{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5089621144652929,
  "eval_steps": 500,
  "global_step": 6215,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.1892536518953e-05,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0,
      "loss": 1.2692375183105469,
      "step": 1
    },
    {
      "epoch": 0.000163785073037906,
      "grad_norm": 0.138671875,
      "learning_rate": 1.2244897959183673e-07,
      "loss": 0.7788365483283997,
      "step": 2
    },
    {
      "epoch": 0.000245677609556859,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.4489795918367347e-07,
      "loss": 1.3062002658843994,
      "step": 3
    },
    {
      "epoch": 0.000327570146075812,
      "grad_norm": 0.11669921875,
      "learning_rate": 3.673469387755102e-07,
      "loss": 0.7250372767448425,
      "step": 4
    },
    {
      "epoch": 0.00040946268259476503,
      "grad_norm": 0.1337890625,
      "learning_rate": 4.897959183673469e-07,
      "loss": 0.961167573928833,
      "step": 5
    },
    {
      "epoch": 0.000491355219113718,
      "grad_norm": 0.16015625,
      "learning_rate": 6.122448979591837e-07,
      "loss": 1.0844104290008545,
      "step": 6
    },
    {
      "epoch": 0.000573247755632671,
      "grad_norm": 0.1455078125,
      "learning_rate": 7.346938775510204e-07,
      "loss": 0.9477209448814392,
      "step": 7
    },
    {
      "epoch": 0.000655140292151624,
      "grad_norm": 0.1484375,
      "learning_rate": 8.571428571428571e-07,
      "loss": 0.8315292000770569,
      "step": 8
    },
    {
      "epoch": 0.000737032828670577,
      "grad_norm": 0.142578125,
      "learning_rate": 9.795918367346939e-07,
      "loss": 0.8303043246269226,
      "step": 9
    },
    {
      "epoch": 0.0008189253651895301,
      "grad_norm": 0.1552734375,
      "learning_rate": 1.1020408163265306e-06,
      "loss": 0.748382568359375,
      "step": 10
    },
    {
      "epoch": 0.000900817901708483,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.2244897959183673e-06,
      "loss": 0.7946436405181885,
      "step": 11
    },
    {
      "epoch": 0.000982710438227436,
      "grad_norm": 0.1484375,
      "learning_rate": 1.346938775510204e-06,
      "loss": 0.8571218848228455,
      "step": 12
    },
    {
      "epoch": 0.001064602974746389,
      "grad_norm": 0.1552734375,
      "learning_rate": 1.4693877551020408e-06,
      "loss": 0.8640573024749756,
      "step": 13
    },
    {
      "epoch": 0.001146495511265342,
      "grad_norm": 0.14453125,
      "learning_rate": 1.5918367346938777e-06,
      "loss": 0.8880600333213806,
      "step": 14
    },
    {
      "epoch": 0.0012283880477842951,
      "grad_norm": 0.158203125,
      "learning_rate": 1.7142857142857143e-06,
      "loss": 0.8768412470817566,
      "step": 15
    },
    {
      "epoch": 0.001310280584303248,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.836734693877551e-06,
      "loss": 0.8432092070579529,
      "step": 16
    },
    {
      "epoch": 0.001392173120822201,
      "grad_norm": 0.12451171875,
      "learning_rate": 1.9591836734693877e-06,
      "loss": 0.7979586124420166,
      "step": 17
    },
    {
      "epoch": 0.001474065657341154,
      "grad_norm": 0.201171875,
      "learning_rate": 2.0816326530612247e-06,
      "loss": 1.0535919666290283,
      "step": 18
    },
    {
      "epoch": 0.001555958193860107,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.204081632653061e-06,
      "loss": 0.7070658206939697,
      "step": 19
    },
    {
      "epoch": 0.0016378507303790601,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.326530612244898e-06,
      "loss": 1.01751708984375,
      "step": 20
    },
    {
      "epoch": 0.0017197432668980132,
      "grad_norm": 0.154296875,
      "learning_rate": 2.4489795918367347e-06,
      "loss": 1.043867588043213,
      "step": 21
    },
    {
      "epoch": 0.001801635803416966,
      "grad_norm": 0.158203125,
      "learning_rate": 2.571428571428571e-06,
      "loss": 0.9232407808303833,
      "step": 22
    },
    {
      "epoch": 0.001883528339935919,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.693877551020408e-06,
      "loss": 0.9462562203407288,
      "step": 23
    },
    {
      "epoch": 0.001965420876454872,
      "grad_norm": 0.1298828125,
      "learning_rate": 2.816326530612245e-06,
      "loss": 0.9328899383544922,
      "step": 24
    },
    {
      "epoch": 0.002047313412973825,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.9387755102040816e-06,
      "loss": 0.8777925372123718,
      "step": 25
    },
    {
      "epoch": 0.002129205949492778,
      "grad_norm": 0.140625,
      "learning_rate": 3.0612244897959185e-06,
      "loss": 0.6303024888038635,
      "step": 26
    },
    {
      "epoch": 0.002211098486011731,
      "grad_norm": 0.1396484375,
      "learning_rate": 3.1836734693877555e-06,
      "loss": 0.9259619116783142,
      "step": 27
    },
    {
      "epoch": 0.002292991022530684,
      "grad_norm": 0.138671875,
      "learning_rate": 3.3061224489795916e-06,
      "loss": 0.8729074597358704,
      "step": 28
    },
    {
      "epoch": 0.0023748835590496373,
      "grad_norm": 0.1298828125,
      "learning_rate": 3.4285714285714285e-06,
      "loss": 0.5940540432929993,
      "step": 29
    },
    {
      "epoch": 0.0024567760955685903,
      "grad_norm": 0.142578125,
      "learning_rate": 3.5510204081632655e-06,
      "loss": 0.7300029397010803,
      "step": 30
    },
    {
      "epoch": 0.0025386686320875433,
      "grad_norm": 0.14453125,
      "learning_rate": 3.673469387755102e-06,
      "loss": 0.6736958026885986,
      "step": 31
    },
    {
      "epoch": 0.002620561168606496,
      "grad_norm": 0.1572265625,
      "learning_rate": 3.795918367346939e-06,
      "loss": 0.6989753842353821,
      "step": 32
    },
    {
      "epoch": 0.002702453705125449,
      "grad_norm": 0.146484375,
      "learning_rate": 3.9183673469387755e-06,
      "loss": 1.0111935138702393,
      "step": 33
    },
    {
      "epoch": 0.002784346241644402,
      "grad_norm": 0.13671875,
      "learning_rate": 4.040816326530612e-06,
      "loss": 0.7319263219833374,
      "step": 34
    },
    {
      "epoch": 0.002866238778163355,
      "grad_norm": 0.1337890625,
      "learning_rate": 4.163265306122449e-06,
      "loss": 0.8890162706375122,
      "step": 35
    },
    {
      "epoch": 0.002948131314682308,
      "grad_norm": 0.150390625,
      "learning_rate": 4.285714285714286e-06,
      "loss": 1.323777198791504,
      "step": 36
    },
    {
      "epoch": 0.003030023851201261,
      "grad_norm": 0.162109375,
      "learning_rate": 4.408163265306122e-06,
      "loss": 1.0636528730392456,
      "step": 37
    },
    {
      "epoch": 0.003111916387720214,
      "grad_norm": 0.1416015625,
      "learning_rate": 4.530612244897959e-06,
      "loss": 1.1280652284622192,
      "step": 38
    },
    {
      "epoch": 0.003193808924239167,
      "grad_norm": 0.1357421875,
      "learning_rate": 4.653061224489796e-06,
      "loss": 0.8183513879776001,
      "step": 39
    },
    {
      "epoch": 0.0032757014607581202,
      "grad_norm": 0.1328125,
      "learning_rate": 4.775510204081632e-06,
      "loss": 0.7616685032844543,
      "step": 40
    },
    {
      "epoch": 0.0033575939972770733,
      "grad_norm": 0.17578125,
      "learning_rate": 4.897959183673469e-06,
      "loss": 0.9517663717269897,
      "step": 41
    },
    {
      "epoch": 0.0034394865337960263,
      "grad_norm": 0.1767578125,
      "learning_rate": 5.020408163265306e-06,
      "loss": 0.9238744974136353,
      "step": 42
    },
    {
      "epoch": 0.0035213790703149794,
      "grad_norm": 0.1181640625,
      "learning_rate": 5.142857142857142e-06,
      "loss": 0.6035186648368835,
      "step": 43
    },
    {
      "epoch": 0.003603271606833932,
      "grad_norm": 0.146484375,
      "learning_rate": 5.265306122448979e-06,
      "loss": 0.504254937171936,
      "step": 44
    },
    {
      "epoch": 0.003685164143352885,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.387755102040816e-06,
      "loss": 0.8422408699989319,
      "step": 45
    },
    {
      "epoch": 0.003767056679871838,
      "grad_norm": 0.1484375,
      "learning_rate": 5.510204081632653e-06,
      "loss": 0.7991898059844971,
      "step": 46
    },
    {
      "epoch": 0.003848949216390791,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.63265306122449e-06,
      "loss": 0.6923357248306274,
      "step": 47
    },
    {
      "epoch": 0.003930841752909744,
      "grad_norm": 0.166015625,
      "learning_rate": 5.755102040816326e-06,
      "loss": 0.8725593090057373,
      "step": 48
    },
    {
      "epoch": 0.004012734289428698,
      "grad_norm": 0.1376953125,
      "learning_rate": 5.877551020408163e-06,
      "loss": 0.8000404238700867,
      "step": 49
    },
    {
      "epoch": 0.00409462682594765,
      "grad_norm": 0.1337890625,
      "learning_rate": 6e-06,
      "loss": 0.8268479108810425,
      "step": 50
    },
    {
      "epoch": 0.004176519362466603,
      "grad_norm": 0.1552734375,
      "learning_rate": 6.122448979591837e-06,
      "loss": 0.9731822609901428,
      "step": 51
    },
    {
      "epoch": 0.004258411898985556,
      "grad_norm": 0.126953125,
      "learning_rate": 6.244897959183674e-06,
      "loss": 0.6076136231422424,
      "step": 52
    },
    {
      "epoch": 0.004340304435504509,
      "grad_norm": 0.1533203125,
      "learning_rate": 6.367346938775511e-06,
      "loss": 0.8989383578300476,
      "step": 53
    },
    {
      "epoch": 0.004422196972023462,
      "grad_norm": 0.1318359375,
      "learning_rate": 6.489795918367348e-06,
      "loss": 1.2253038883209229,
      "step": 54
    },
    {
      "epoch": 0.004504089508542415,
      "grad_norm": 0.1416015625,
      "learning_rate": 6.612244897959183e-06,
      "loss": 1.0751113891601562,
      "step": 55
    },
    {
      "epoch": 0.004585982045061368,
      "grad_norm": 0.15625,
      "learning_rate": 6.73469387755102e-06,
      "loss": 0.976893961429596,
      "step": 56
    },
    {
      "epoch": 0.004667874581580321,
      "grad_norm": 0.1630859375,
      "learning_rate": 6.857142857142857e-06,
      "loss": 1.0483677387237549,
      "step": 57
    },
    {
      "epoch": 0.0047497671180992745,
      "grad_norm": 0.14453125,
      "learning_rate": 6.979591836734694e-06,
      "loss": 1.0369893312454224,
      "step": 58
    },
    {
      "epoch": 0.004831659654618227,
      "grad_norm": 0.1533203125,
      "learning_rate": 7.102040816326531e-06,
      "loss": 0.9358440637588501,
      "step": 59
    },
    {
      "epoch": 0.004913552191137181,
      "grad_norm": 0.1484375,
      "learning_rate": 7.224489795918368e-06,
      "loss": 0.8366637229919434,
      "step": 60
    },
    {
      "epoch": 0.004995444727656133,
      "grad_norm": 0.158203125,
      "learning_rate": 7.346938775510204e-06,
      "loss": 0.9385107755661011,
      "step": 61
    },
    {
      "epoch": 0.005077337264175087,
      "grad_norm": 0.1474609375,
      "learning_rate": 7.469387755102041e-06,
      "loss": 0.8058649301528931,
      "step": 62
    },
    {
      "epoch": 0.005159229800694039,
      "grad_norm": 0.1572265625,
      "learning_rate": 7.591836734693878e-06,
      "loss": 1.2202202081680298,
      "step": 63
    },
    {
      "epoch": 0.005241122337212992,
      "grad_norm": 0.1259765625,
      "learning_rate": 7.714285714285716e-06,
      "loss": 0.6693165898323059,
      "step": 64
    },
    {
      "epoch": 0.005323014873731945,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.836734693877551e-06,
      "loss": 0.6848925948143005,
      "step": 65
    },
    {
      "epoch": 0.005404907410250898,
      "grad_norm": 0.126953125,
      "learning_rate": 7.959183673469388e-06,
      "loss": 0.5377231240272522,
      "step": 66
    },
    {
      "epoch": 0.005486799946769851,
      "grad_norm": 0.12890625,
      "learning_rate": 8.081632653061225e-06,
      "loss": 0.7365704774856567,
      "step": 67
    },
    {
      "epoch": 0.005568692483288804,
      "grad_norm": 0.142578125,
      "learning_rate": 8.204081632653062e-06,
      "loss": 0.8671019673347473,
      "step": 68
    },
    {
      "epoch": 0.0056505850198077575,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.326530612244899e-06,
      "loss": 0.8263939023017883,
      "step": 69
    },
    {
      "epoch": 0.00573247755632671,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.448979591836736e-06,
      "loss": 0.8309429883956909,
      "step": 70
    },
    {
      "epoch": 0.005814370092845664,
      "grad_norm": 0.142578125,
      "learning_rate": 8.571428571428573e-06,
      "loss": 0.868381917476654,
      "step": 71
    },
    {
      "epoch": 0.005896262629364616,
      "grad_norm": 0.1455078125,
      "learning_rate": 8.693877551020408e-06,
      "loss": 0.8100480437278748,
      "step": 72
    },
    {
      "epoch": 0.00597815516588357,
      "grad_norm": 0.154296875,
      "learning_rate": 8.816326530612245e-06,
      "loss": 0.8355422616004944,
      "step": 73
    },
    {
      "epoch": 0.006060047702402522,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.938775510204082e-06,
      "loss": 0.6954861879348755,
      "step": 74
    },
    {
      "epoch": 0.006141940238921475,
      "grad_norm": 0.1826171875,
      "learning_rate": 9.061224489795919e-06,
      "loss": 1.3356058597564697,
      "step": 75
    },
    {
      "epoch": 0.006223832775440428,
      "grad_norm": 0.1533203125,
      "learning_rate": 9.183673469387756e-06,
      "loss": 0.8860036134719849,
      "step": 76
    },
    {
      "epoch": 0.006305725311959381,
      "grad_norm": 0.1474609375,
      "learning_rate": 9.306122448979593e-06,
      "loss": 0.8816828727722168,
      "step": 77
    },
    {
      "epoch": 0.006387617848478334,
      "grad_norm": 0.146484375,
      "learning_rate": 9.428571428571428e-06,
      "loss": 0.8622653484344482,
      "step": 78
    },
    {
      "epoch": 0.006469510384997287,
      "grad_norm": 0.150390625,
      "learning_rate": 9.551020408163265e-06,
      "loss": 1.005674123764038,
      "step": 79
    },
    {
      "epoch": 0.0065514029215162405,
      "grad_norm": 0.208984375,
      "learning_rate": 9.673469387755102e-06,
      "loss": 0.9369722604751587,
      "step": 80
    },
    {
      "epoch": 0.006633295458035193,
      "grad_norm": 0.130859375,
      "learning_rate": 9.795918367346939e-06,
      "loss": 0.881393551826477,
      "step": 81
    },
    {
      "epoch": 0.006715187994554147,
      "grad_norm": 0.197265625,
      "learning_rate": 9.918367346938776e-06,
      "loss": 0.801750898361206,
      "step": 82
    },
    {
      "epoch": 0.006797080531073099,
      "grad_norm": 0.1640625,
      "learning_rate": 1.0040816326530613e-05,
      "loss": 0.88079434633255,
      "step": 83
    },
    {
      "epoch": 0.006878973067592053,
      "grad_norm": 0.162109375,
      "learning_rate": 1.016326530612245e-05,
      "loss": 1.0499800443649292,
      "step": 84
    },
    {
      "epoch": 0.006960865604111005,
      "grad_norm": 0.162109375,
      "learning_rate": 1.0285714285714285e-05,
      "loss": 0.7700325846672058,
      "step": 85
    },
    {
      "epoch": 0.007042758140629959,
      "grad_norm": 0.150390625,
      "learning_rate": 1.0408163265306122e-05,
      "loss": 0.8462141156196594,
      "step": 86
    },
    {
      "epoch": 0.007124650677148911,
      "grad_norm": 0.1533203125,
      "learning_rate": 1.0530612244897959e-05,
      "loss": 0.9165534377098083,
      "step": 87
    },
    {
      "epoch": 0.007206543213667864,
      "grad_norm": 0.322265625,
      "learning_rate": 1.0653061224489796e-05,
      "loss": 0.8755977153778076,
      "step": 88
    },
    {
      "epoch": 0.007288435750186817,
      "grad_norm": 0.1513671875,
      "learning_rate": 1.0775510204081633e-05,
      "loss": 0.9935065507888794,
      "step": 89
    },
    {
      "epoch": 0.00737032828670577,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.089795918367347e-05,
      "loss": 0.7393916845321655,
      "step": 90
    },
    {
      "epoch": 0.0074522208232247235,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.1020408163265306e-05,
      "loss": 1.090970754623413,
      "step": 91
    },
    {
      "epoch": 0.007534113359743676,
      "grad_norm": 0.1591796875,
      "learning_rate": 1.1142857142857143e-05,
      "loss": 0.5876522064208984,
      "step": 92
    },
    {
      "epoch": 0.0076160058962626296,
      "grad_norm": 0.125,
      "learning_rate": 1.126530612244898e-05,
      "loss": 0.6235696077346802,
      "step": 93
    },
    {
      "epoch": 0.007697898432781582,
      "grad_norm": 0.419921875,
      "learning_rate": 1.1387755102040817e-05,
      "loss": 0.8901874423027039,
      "step": 94
    },
    {
      "epoch": 0.007779790969300536,
      "grad_norm": 0.169921875,
      "learning_rate": 1.1510204081632653e-05,
      "loss": 0.7616423964500427,
      "step": 95
    },
    {
      "epoch": 0.007861683505819488,
      "grad_norm": 0.193359375,
      "learning_rate": 1.163265306122449e-05,
      "loss": 1.2576884031295776,
      "step": 96
    },
    {
      "epoch": 0.00794357604233844,
      "grad_norm": 0.177734375,
      "learning_rate": 1.1755102040816326e-05,
      "loss": 0.6757850050926208,
      "step": 97
    },
    {
      "epoch": 0.008025468578857395,
      "grad_norm": 0.140625,
      "learning_rate": 1.1877551020408163e-05,
      "loss": 0.775367259979248,
      "step": 98
    },
    {
      "epoch": 0.008107361115376348,
      "grad_norm": 0.1484375,
      "learning_rate": 1.2e-05,
      "loss": 0.9876186847686768,
      "step": 99
    },
    {
      "epoch": 0.0081892536518953,
      "grad_norm": 0.1591796875,
      "learning_rate": 1.2122448979591836e-05,
      "loss": 0.8306399583816528,
      "step": 100
    },
    {
      "epoch": 0.008271146188414253,
      "grad_norm": 0.16796875,
      "learning_rate": 1.2244897959183674e-05,
      "loss": 0.8706358075141907,
      "step": 101
    },
    {
      "epoch": 0.008353038724933206,
      "grad_norm": 0.1611328125,
      "learning_rate": 1.236734693877551e-05,
      "loss": 0.7968311309814453,
      "step": 102
    },
    {
      "epoch": 0.00843493126145216,
      "grad_norm": 0.1318359375,
      "learning_rate": 1.2489795918367348e-05,
      "loss": 0.7418109178543091,
      "step": 103
    },
    {
      "epoch": 0.008516823797971113,
      "grad_norm": 0.1474609375,
      "learning_rate": 1.2612244897959183e-05,
      "loss": 0.8188655972480774,
      "step": 104
    },
    {
      "epoch": 0.008598716334490065,
      "grad_norm": 0.1533203125,
      "learning_rate": 1.2734693877551022e-05,
      "loss": 0.6840787529945374,
      "step": 105
    },
    {
      "epoch": 0.008680608871009018,
      "grad_norm": 0.16015625,
      "learning_rate": 1.2857142857142857e-05,
      "loss": 0.8916717767715454,
      "step": 106
    },
    {
      "epoch": 0.008762501407527972,
      "grad_norm": 0.1318359375,
      "learning_rate": 1.2979591836734696e-05,
      "loss": 0.7145904302597046,
      "step": 107
    },
    {
      "epoch": 0.008844393944046925,
      "grad_norm": 0.1572265625,
      "learning_rate": 1.3102040816326531e-05,
      "loss": 0.8699553608894348,
      "step": 108
    },
    {
      "epoch": 0.008926286480565877,
      "grad_norm": 0.177734375,
      "learning_rate": 1.3224489795918366e-05,
      "loss": 1.0870182514190674,
      "step": 109
    },
    {
      "epoch": 0.00900817901708483,
      "grad_norm": 0.1533203125,
      "learning_rate": 1.3346938775510205e-05,
      "loss": 0.830143392086029,
      "step": 110
    },
    {
      "epoch": 0.009090071553603784,
      "grad_norm": 0.3125,
      "learning_rate": 1.346938775510204e-05,
      "loss": 0.9132651090621948,
      "step": 111
    },
    {
      "epoch": 0.009171964090122737,
      "grad_norm": 0.1806640625,
      "learning_rate": 1.3591836734693879e-05,
      "loss": 0.7668812274932861,
      "step": 112
    },
    {
      "epoch": 0.00925385662664169,
      "grad_norm": 0.169921875,
      "learning_rate": 1.3714285714285714e-05,
      "loss": 0.7226609587669373,
      "step": 113
    },
    {
      "epoch": 0.009335749163160642,
      "grad_norm": 0.1748046875,
      "learning_rate": 1.3836734693877553e-05,
      "loss": 0.9787721633911133,
      "step": 114
    },
    {
      "epoch": 0.009417641699679595,
      "grad_norm": 0.1591796875,
      "learning_rate": 1.3959183673469388e-05,
      "loss": 0.8557654619216919,
      "step": 115
    },
    {
      "epoch": 0.009499534236198549,
      "grad_norm": 0.1650390625,
      "learning_rate": 1.4081632653061225e-05,
      "loss": 0.9546581506729126,
      "step": 116
    },
    {
      "epoch": 0.009581426772717502,
      "grad_norm": 0.146484375,
      "learning_rate": 1.4204081632653062e-05,
      "loss": 0.629300594329834,
      "step": 117
    },
    {
      "epoch": 0.009663319309236454,
      "grad_norm": 0.158203125,
      "learning_rate": 1.4326530612244899e-05,
      "loss": 1.1067485809326172,
      "step": 118
    },
    {
      "epoch": 0.009745211845755407,
      "grad_norm": 0.1806640625,
      "learning_rate": 1.4448979591836736e-05,
      "loss": 0.8721896409988403,
      "step": 119
    },
    {
      "epoch": 0.009827104382274361,
      "grad_norm": 0.1484375,
      "learning_rate": 1.4571428571428571e-05,
      "loss": 0.671083927154541,
      "step": 120
    },
    {
      "epoch": 0.009908996918793314,
      "grad_norm": 0.138671875,
      "learning_rate": 1.4693877551020408e-05,
      "loss": 1.0528693199157715,
      "step": 121
    },
    {
      "epoch": 0.009990889455312266,
      "grad_norm": 0.1962890625,
      "learning_rate": 1.4816326530612245e-05,
      "loss": 0.8432303667068481,
      "step": 122
    },
    {
      "epoch": 0.010072781991831219,
      "grad_norm": 0.1396484375,
      "learning_rate": 1.4938775510204082e-05,
      "loss": 0.7435109615325928,
      "step": 123
    },
    {
      "epoch": 0.010154674528350173,
      "grad_norm": 0.169921875,
      "learning_rate": 1.5061224489795919e-05,
      "loss": 0.7423120141029358,
      "step": 124
    },
    {
      "epoch": 0.010236567064869126,
      "grad_norm": 0.14453125,
      "learning_rate": 1.5183673469387756e-05,
      "loss": 0.9071717858314514,
      "step": 125
    },
    {
      "epoch": 0.010318459601388079,
      "grad_norm": 0.1748046875,
      "learning_rate": 1.530612244897959e-05,
      "loss": 0.7631561160087585,
      "step": 126
    },
    {
      "epoch": 0.010400352137907031,
      "grad_norm": 0.1796875,
      "learning_rate": 1.542857142857143e-05,
      "loss": 0.8060978651046753,
      "step": 127
    },
    {
      "epoch": 0.010482244674425984,
      "grad_norm": 0.1611328125,
      "learning_rate": 1.5551020408163265e-05,
      "loss": 0.7506223320960999,
      "step": 128
    },
    {
      "epoch": 0.010564137210944938,
      "grad_norm": 0.169921875,
      "learning_rate": 1.5673469387755102e-05,
      "loss": 0.9428765773773193,
      "step": 129
    },
    {
      "epoch": 0.01064602974746389,
      "grad_norm": 0.1689453125,
      "learning_rate": 1.579591836734694e-05,
      "loss": 0.5963538289070129,
      "step": 130
    },
    {
      "epoch": 0.010727922283982843,
      "grad_norm": 0.158203125,
      "learning_rate": 1.5918367346938776e-05,
      "loss": 0.7066556215286255,
      "step": 131
    },
    {
      "epoch": 0.010809814820501796,
      "grad_norm": 0.1484375,
      "learning_rate": 1.6040816326530613e-05,
      "loss": 0.882148265838623,
      "step": 132
    },
    {
      "epoch": 0.01089170735702075,
      "grad_norm": 0.205078125,
      "learning_rate": 1.616326530612245e-05,
      "loss": 0.797228991985321,
      "step": 133
    },
    {
      "epoch": 0.010973599893539703,
      "grad_norm": 0.173828125,
      "learning_rate": 1.6285714285714287e-05,
      "loss": 0.7989157438278198,
      "step": 134
    },
    {
      "epoch": 0.011055492430058655,
      "grad_norm": 0.1806640625,
      "learning_rate": 1.6408163265306124e-05,
      "loss": 1.149370789527893,
      "step": 135
    },
    {
      "epoch": 0.011137384966577608,
      "grad_norm": 0.1875,
      "learning_rate": 1.653061224489796e-05,
      "loss": 0.9384973049163818,
      "step": 136
    },
    {
      "epoch": 0.011219277503096562,
      "grad_norm": 0.1748046875,
      "learning_rate": 1.6653061224489797e-05,
      "loss": 0.6930122375488281,
      "step": 137
    },
    {
      "epoch": 0.011301170039615515,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.6775510204081634e-05,
      "loss": 1.0194259881973267,
      "step": 138
    },
    {
      "epoch": 0.011383062576134468,
      "grad_norm": 0.1748046875,
      "learning_rate": 1.689795918367347e-05,
      "loss": 0.8852027058601379,
      "step": 139
    },
    {
      "epoch": 0.01146495511265342,
      "grad_norm": 0.1708984375,
      "learning_rate": 1.7020408163265305e-05,
      "loss": 0.8619977831840515,
      "step": 140
    },
    {
      "epoch": 0.011546847649172373,
      "grad_norm": 0.158203125,
      "learning_rate": 1.7142857142857145e-05,
      "loss": 0.7176647186279297,
      "step": 141
    },
    {
      "epoch": 0.011628740185691327,
      "grad_norm": 0.1953125,
      "learning_rate": 1.726530612244898e-05,
      "loss": 1.1816068887710571,
      "step": 142
    },
    {
      "epoch": 0.01171063272221028,
      "grad_norm": 0.1767578125,
      "learning_rate": 1.7387755102040816e-05,
      "loss": 0.9537625908851624,
      "step": 143
    },
    {
      "epoch": 0.011792525258729232,
      "grad_norm": 0.275390625,
      "learning_rate": 1.7510204081632653e-05,
      "loss": 0.7848352193832397,
      "step": 144
    },
    {
      "epoch": 0.011874417795248185,
      "grad_norm": 0.212890625,
      "learning_rate": 1.763265306122449e-05,
      "loss": 0.8986237049102783,
      "step": 145
    },
    {
      "epoch": 0.01195631033176714,
      "grad_norm": 0.2080078125,
      "learning_rate": 1.7755102040816327e-05,
      "loss": 0.7229043245315552,
      "step": 146
    },
    {
      "epoch": 0.012038202868286092,
      "grad_norm": 0.1875,
      "learning_rate": 1.7877551020408164e-05,
      "loss": 1.0595723390579224,
      "step": 147
    },
    {
      "epoch": 0.012120095404805045,
      "grad_norm": 0.19140625,
      "learning_rate": 1.8e-05,
      "loss": 0.9196676015853882,
      "step": 148
    },
    {
      "epoch": 0.012201987941323997,
      "grad_norm": 0.193359375,
      "learning_rate": 1.8122448979591837e-05,
      "loss": 0.7915199995040894,
      "step": 149
    },
    {
      "epoch": 0.01228388047784295,
      "grad_norm": 0.181640625,
      "learning_rate": 1.8244897959183674e-05,
      "loss": 0.683129072189331,
      "step": 150
    },
    {
      "epoch": 0.012365773014361904,
      "grad_norm": 0.162109375,
      "learning_rate": 1.836734693877551e-05,
      "loss": 0.8023064732551575,
      "step": 151
    },
    {
      "epoch": 0.012447665550880857,
      "grad_norm": 0.197265625,
      "learning_rate": 1.8489795918367348e-05,
      "loss": 0.6344118714332581,
      "step": 152
    },
    {
      "epoch": 0.01252955808739981,
      "grad_norm": 0.17578125,
      "learning_rate": 1.8612244897959185e-05,
      "loss": 0.7012781500816345,
      "step": 153
    },
    {
      "epoch": 0.012611450623918762,
      "grad_norm": 0.16796875,
      "learning_rate": 1.8734693877551022e-05,
      "loss": 0.9908215999603271,
      "step": 154
    },
    {
      "epoch": 0.012693343160437716,
      "grad_norm": 0.1796875,
      "learning_rate": 1.8857142857142856e-05,
      "loss": 0.8533449172973633,
      "step": 155
    },
    {
      "epoch": 0.012775235696956669,
      "grad_norm": 0.15625,
      "learning_rate": 1.8979591836734696e-05,
      "loss": 0.7644760012626648,
      "step": 156
    },
    {
      "epoch": 0.012857128233475621,
      "grad_norm": 0.162109375,
      "learning_rate": 1.910204081632653e-05,
      "loss": 1.0475883483886719,
      "step": 157
    },
    {
      "epoch": 0.012939020769994574,
      "grad_norm": 0.1826171875,
      "learning_rate": 1.9224489795918367e-05,
      "loss": 0.7506498098373413,
      "step": 158
    },
    {
      "epoch": 0.013020913306513528,
      "grad_norm": 0.177734375,
      "learning_rate": 1.9346938775510203e-05,
      "loss": 0.6886553764343262,
      "step": 159
    },
    {
      "epoch": 0.013102805843032481,
      "grad_norm": 0.189453125,
      "learning_rate": 1.946938775510204e-05,
      "loss": 0.8300869464874268,
      "step": 160
    },
    {
      "epoch": 0.013184698379551434,
      "grad_norm": 0.1767578125,
      "learning_rate": 1.9591836734693877e-05,
      "loss": 1.0316212177276611,
      "step": 161
    },
    {
      "epoch": 0.013266590916070386,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.9714285714285714e-05,
      "loss": 0.7299474477767944,
      "step": 162
    },
    {
      "epoch": 0.013348483452589339,
      "grad_norm": 0.216796875,
      "learning_rate": 1.983673469387755e-05,
      "loss": 0.8153265714645386,
      "step": 163
    },
    {
      "epoch": 0.013430375989108293,
      "grad_norm": 0.1630859375,
      "learning_rate": 1.9959183673469388e-05,
      "loss": 0.9564380049705505,
      "step": 164
    },
    {
      "epoch": 0.013512268525627246,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.0081632653061225e-05,
      "loss": 1.0247093439102173,
      "step": 165
    },
    {
      "epoch": 0.013594161062146198,
      "grad_norm": 0.169921875,
      "learning_rate": 2.0204081632653062e-05,
      "loss": 0.9854190349578857,
      "step": 166
    },
    {
      "epoch": 0.013676053598665151,
      "grad_norm": 0.232421875,
      "learning_rate": 2.03265306122449e-05,
      "loss": 0.8826335072517395,
      "step": 167
    },
    {
      "epoch": 0.013757946135184105,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.0448979591836736e-05,
      "loss": 0.9545683264732361,
      "step": 168
    },
    {
      "epoch": 0.013839838671703058,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.057142857142857e-05,
      "loss": 0.7048317790031433,
      "step": 169
    },
    {
      "epoch": 0.01392173120822201,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.069387755102041e-05,
      "loss": 1.383650541305542,
      "step": 170
    },
    {
      "epoch": 0.014003623744740963,
      "grad_norm": 0.2490234375,
      "learning_rate": 2.0816326530612243e-05,
      "loss": 1.3275967836380005,
      "step": 171
    },
    {
      "epoch": 0.014085516281259917,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.0938775510204084e-05,
      "loss": 0.9396587610244751,
      "step": 172
    },
    {
      "epoch": 0.01416740881777887,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.1061224489795917e-05,
      "loss": 0.7036116123199463,
      "step": 173
    },
    {
      "epoch": 0.014249301354297823,
      "grad_norm": 0.20703125,
      "learning_rate": 2.1183673469387758e-05,
      "loss": 0.9278486967086792,
      "step": 174
    },
    {
      "epoch": 0.014331193890816775,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.130612244897959e-05,
      "loss": 0.7104832530021667,
      "step": 175
    },
    {
      "epoch": 0.014413086427335728,
      "grad_norm": 0.177734375,
      "learning_rate": 2.142857142857143e-05,
      "loss": 0.9519690275192261,
      "step": 176
    },
    {
      "epoch": 0.014494978963854682,
      "grad_norm": 0.177734375,
      "learning_rate": 2.1551020408163265e-05,
      "loss": 0.84523606300354,
      "step": 177
    },
    {
      "epoch": 0.014576871500373635,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.1673469387755102e-05,
      "loss": 0.9742757081985474,
      "step": 178
    },
    {
      "epoch": 0.014658764036892587,
      "grad_norm": 0.2265625,
      "learning_rate": 2.179591836734694e-05,
      "loss": 0.9848318099975586,
      "step": 179
    },
    {
      "epoch": 0.01474065657341154,
      "grad_norm": 0.18359375,
      "learning_rate": 2.1918367346938776e-05,
      "loss": 0.790285587310791,
      "step": 180
    },
    {
      "epoch": 0.014822549109930494,
      "grad_norm": 0.24609375,
      "learning_rate": 2.2040816326530613e-05,
      "loss": 1.0562388896942139,
      "step": 181
    },
    {
      "epoch": 0.014904441646449447,
      "grad_norm": 0.203125,
      "learning_rate": 2.216326530612245e-05,
      "loss": 0.6656460762023926,
      "step": 182
    },
    {
      "epoch": 0.0149863341829684,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2285714285714287e-05,
      "loss": 0.7126745581626892,
      "step": 183
    },
    {
      "epoch": 0.015068226719487352,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.2408163265306124e-05,
      "loss": 0.5595638751983643,
      "step": 184
    },
    {
      "epoch": 0.015150119256006307,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.253061224489796e-05,
      "loss": 0.837664008140564,
      "step": 185
    },
    {
      "epoch": 0.015232011792525259,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.2653061224489798e-05,
      "loss": 1.0549743175506592,
      "step": 186
    },
    {
      "epoch": 0.015313904329044212,
      "grad_norm": 0.158203125,
      "learning_rate": 2.2775510204081635e-05,
      "loss": 0.6522034406661987,
      "step": 187
    },
    {
      "epoch": 0.015395796865563164,
      "grad_norm": 0.181640625,
      "learning_rate": 2.289795918367347e-05,
      "loss": 0.88138747215271,
      "step": 188
    },
    {
      "epoch": 0.015477689402082117,
      "grad_norm": 0.23046875,
      "learning_rate": 2.3020408163265305e-05,
      "loss": 0.7675457000732422,
      "step": 189
    },
    {
      "epoch": 0.015559581938601071,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3142857142857145e-05,
      "loss": 0.7595466375350952,
      "step": 190
    },
    {
      "epoch": 0.015641474475120022,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.326530612244898e-05,
      "loss": 0.6058135628700256,
      "step": 191
    },
    {
      "epoch": 0.015723367011638976,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3387755102040816e-05,
      "loss": 0.8671406507492065,
      "step": 192
    },
    {
      "epoch": 0.01580525954815793,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3510204081632653e-05,
      "loss": 0.9157660007476807,
      "step": 193
    },
    {
      "epoch": 0.01588715208467688,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.363265306122449e-05,
      "loss": 0.879551112651825,
      "step": 194
    },
    {
      "epoch": 0.015969044621195836,
      "grad_norm": 0.28515625,
      "learning_rate": 2.3755102040816327e-05,
      "loss": 0.641154944896698,
      "step": 195
    },
    {
      "epoch": 0.01605093715771479,
      "grad_norm": 0.171875,
      "learning_rate": 2.3877551020408164e-05,
      "loss": 0.6814965605735779,
      "step": 196
    },
    {
      "epoch": 0.01613282969423374,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.4e-05,
      "loss": 0.8545130491256714,
      "step": 197
    },
    {
      "epoch": 0.016214722230752696,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.399999973921529e-05,
      "loss": 0.9945880174636841,
      "step": 198
    },
    {
      "epoch": 0.016296614767271646,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3999998956861162e-05,
      "loss": 0.7373678088188171,
      "step": 199
    },
    {
      "epoch": 0.0163785073037906,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3999997652937657e-05,
      "loss": 0.8678381443023682,
      "step": 200
    },
    {
      "epoch": 0.016460399840309555,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3999995827444833e-05,
      "loss": 0.6554551720619202,
      "step": 201
    },
    {
      "epoch": 0.016542292376828506,
      "grad_norm": 0.134765625,
      "learning_rate": 2.3999993480382764e-05,
      "loss": 0.6440178751945496,
      "step": 202
    },
    {
      "epoch": 0.01662418491334746,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3999990611751552e-05,
      "loss": 0.9317048788070679,
      "step": 203
    },
    {
      "epoch": 0.01670607744986641,
      "grad_norm": 0.21875,
      "learning_rate": 2.3999987221551325e-05,
      "loss": 0.949978232383728,
      "step": 204
    },
    {
      "epoch": 0.016787969986385366,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3999983309782227e-05,
      "loss": 0.7016098499298096,
      "step": 205
    },
    {
      "epoch": 0.01686986252290432,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3999978876444435e-05,
      "loss": 0.8135285973548889,
      "step": 206
    },
    {
      "epoch": 0.01695175505942327,
      "grad_norm": 0.240234375,
      "learning_rate": 2.3999973921538135e-05,
      "loss": 0.6167044043540955,
      "step": 207
    },
    {
      "epoch": 0.017033647595942225,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3999968445063538e-05,
      "loss": 0.8787804841995239,
      "step": 208
    },
    {
      "epoch": 0.01711554013246118,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3999962447020896e-05,
      "loss": 1.1545552015304565,
      "step": 209
    },
    {
      "epoch": 0.01719743266898013,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.399995592741046e-05,
      "loss": 0.711759090423584,
      "step": 210
    },
    {
      "epoch": 0.017279325205499085,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3999948886232516e-05,
      "loss": 0.8914077877998352,
      "step": 211
    },
    {
      "epoch": 0.017361217742018036,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3999941323487364e-05,
      "loss": 0.8630354404449463,
      "step": 212
    },
    {
      "epoch": 0.01744311027853699,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3999933239175342e-05,
      "loss": 1.1325874328613281,
      "step": 213
    },
    {
      "epoch": 0.017525002815055944,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3999924633296796e-05,
      "loss": 0.9854510426521301,
      "step": 214
    },
    {
      "epoch": 0.017606895351574895,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3999915505852103e-05,
      "loss": 0.8229562044143677,
      "step": 215
    },
    {
      "epoch": 0.01768878788809385,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3999905856841657e-05,
      "loss": 1.047928810119629,
      "step": 216
    },
    {
      "epoch": 0.0177706804246128,
      "grad_norm": 0.16015625,
      "learning_rate": 2.399989568626588e-05,
      "loss": 0.8584483861923218,
      "step": 217
    },
    {
      "epoch": 0.017852572961131755,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.399988499412521e-05,
      "loss": 0.7633299827575684,
      "step": 218
    },
    {
      "epoch": 0.01793446549765071,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3999873780420115e-05,
      "loss": 0.6681854128837585,
      "step": 219
    },
    {
      "epoch": 0.01801635803416966,
      "grad_norm": 0.2294921875,
      "learning_rate": 2.3999862045151084e-05,
      "loss": 0.8275490999221802,
      "step": 220
    },
    {
      "epoch": 0.018098250570688614,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.399984978831862e-05,
      "loss": 1.0392675399780273,
      "step": 221
    },
    {
      "epoch": 0.01818014310720757,
      "grad_norm": 0.1875,
      "learning_rate": 2.3999837009923263e-05,
      "loss": 0.5126798748970032,
      "step": 222
    },
    {
      "epoch": 0.01826203564372652,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3999823709965567e-05,
      "loss": 1.0829551219940186,
      "step": 223
    },
    {
      "epoch": 0.018343928180245474,
      "grad_norm": 0.169921875,
      "learning_rate": 2.399980988844611e-05,
      "loss": 0.9590635299682617,
      "step": 224
    },
    {
      "epoch": 0.018425820716764425,
      "grad_norm": 0.1875,
      "learning_rate": 2.3999795545365487e-05,
      "loss": 1.1667808294296265,
      "step": 225
    },
    {
      "epoch": 0.01850771325328338,
      "grad_norm": 0.203125,
      "learning_rate": 2.399978068072433e-05,
      "loss": 0.8245648145675659,
      "step": 226
    },
    {
      "epoch": 0.018589605789802333,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.399976529452328e-05,
      "loss": 1.0823947191238403,
      "step": 227
    },
    {
      "epoch": 0.018671498326321284,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3999749386763007e-05,
      "loss": 0.8102518916130066,
      "step": 228
    },
    {
      "epoch": 0.01875339086284024,
      "grad_norm": 0.1875,
      "learning_rate": 2.39997329574442e-05,
      "loss": 0.6749840974807739,
      "step": 229
    },
    {
      "epoch": 0.01883528339935919,
      "grad_norm": 0.15625,
      "learning_rate": 2.399971600656758e-05,
      "loss": 0.8160229921340942,
      "step": 230
    },
    {
      "epoch": 0.018917175935878144,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3999698534133875e-05,
      "loss": 0.9988230466842651,
      "step": 231
    },
    {
      "epoch": 0.018999068472397098,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.399968054014385e-05,
      "loss": 1.023828148841858,
      "step": 232
    },
    {
      "epoch": 0.01908096100891605,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3999662024598287e-05,
      "loss": 0.9382991790771484,
      "step": 233
    },
    {
      "epoch": 0.019162853545435003,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3999642987497988e-05,
      "loss": 0.8369897603988647,
      "step": 234
    },
    {
      "epoch": 0.019244746081953958,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.399962342884378e-05,
      "loss": 0.9899635314941406,
      "step": 235
    },
    {
      "epoch": 0.01932663861847291,
      "grad_norm": 0.240234375,
      "learning_rate": 2.399960334863652e-05,
      "loss": 1.175307035446167,
      "step": 236
    },
    {
      "epoch": 0.019408531154991863,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3999582746877072e-05,
      "loss": 0.9381195902824402,
      "step": 237
    },
    {
      "epoch": 0.019490423691510814,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3999561623566336e-05,
      "loss": 1.063532829284668,
      "step": 238
    },
    {
      "epoch": 0.019572316228029768,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.399953997870523e-05,
      "loss": 0.6051982641220093,
      "step": 239
    },
    {
      "epoch": 0.019654208764548722,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3999517812294692e-05,
      "loss": 0.8455670475959778,
      "step": 240
    },
    {
      "epoch": 0.019736101301067673,
      "grad_norm": 0.236328125,
      "learning_rate": 2.399949512433569e-05,
      "loss": 0.715160608291626,
      "step": 241
    },
    {
      "epoch": 0.019817993837586628,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3999471914829207e-05,
      "loss": 0.8958784341812134,
      "step": 242
    },
    {
      "epoch": 0.01989988637410558,
      "grad_norm": 0.17578125,
      "learning_rate": 2.399944818377625e-05,
      "loss": 0.6496281623840332,
      "step": 243
    },
    {
      "epoch": 0.019981778910624533,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.3999423931177856e-05,
      "loss": 1.1493349075317383,
      "step": 244
    },
    {
      "epoch": 0.020063671447143487,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3999399157035073e-05,
      "loss": 0.5976084470748901,
      "step": 245
    },
    {
      "epoch": 0.020145563983662438,
      "grad_norm": 0.2109375,
      "learning_rate": 2.3999373861348982e-05,
      "loss": 0.9261276721954346,
      "step": 246
    },
    {
      "epoch": 0.020227456520181392,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3999348044120683e-05,
      "loss": 1.0355850458145142,
      "step": 247
    },
    {
      "epoch": 0.020309349056700347,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.3999321705351298e-05,
      "loss": 0.9628753066062927,
      "step": 248
    },
    {
      "epoch": 0.020391241593219298,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3999294845041965e-05,
      "loss": 0.9781221151351929,
      "step": 249
    },
    {
      "epoch": 0.020473134129738252,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3999267463193857e-05,
      "loss": 0.9014031291007996,
      "step": 250
    },
    {
      "epoch": 0.020555026666257203,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3999239559808167e-05,
      "loss": 0.7805596590042114,
      "step": 251
    },
    {
      "epoch": 0.020636919202776157,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3999211134886103e-05,
      "loss": 0.9792352914810181,
      "step": 252
    },
    {
      "epoch": 0.02071881173929511,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.39991821884289e-05,
      "loss": 0.6729113459587097,
      "step": 253
    },
    {
      "epoch": 0.020800704275814062,
      "grad_norm": 0.203125,
      "learning_rate": 2.399915272043782e-05,
      "loss": 0.7053857445716858,
      "step": 254
    },
    {
      "epoch": 0.020882596812333017,
      "grad_norm": 0.1640625,
      "learning_rate": 2.399912273091414e-05,
      "loss": 0.730910062789917,
      "step": 255
    },
    {
      "epoch": 0.020964489348851967,
      "grad_norm": 0.189453125,
      "learning_rate": 2.399909221985917e-05,
      "loss": 1.0160516500473022,
      "step": 256
    },
    {
      "epoch": 0.021046381885370922,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.399906118727423e-05,
      "loss": 1.0672346353530884,
      "step": 257
    },
    {
      "epoch": 0.021128274421889876,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3999029633160666e-05,
      "loss": 0.8804312944412231,
      "step": 258
    },
    {
      "epoch": 0.021210166958408827,
      "grad_norm": 0.23828125,
      "learning_rate": 2.3998997557519854e-05,
      "loss": 0.6796444058418274,
      "step": 259
    },
    {
      "epoch": 0.02129205949492778,
      "grad_norm": 0.1875,
      "learning_rate": 2.3998964960353192e-05,
      "loss": 0.9091837406158447,
      "step": 260
    },
    {
      "epoch": 0.021373952031446736,
      "grad_norm": 0.248046875,
      "learning_rate": 2.3998931841662094e-05,
      "loss": 0.8179640173912048,
      "step": 261
    },
    {
      "epoch": 0.021455844567965687,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3998898201447995e-05,
      "loss": 0.8405275940895081,
      "step": 262
    },
    {
      "epoch": 0.02153773710448464,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3998864039712357e-05,
      "loss": 0.7459143400192261,
      "step": 263
    },
    {
      "epoch": 0.021619629641003592,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3998829356456676e-05,
      "loss": 0.6243706345558167,
      "step": 264
    },
    {
      "epoch": 0.021701522177522546,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3998794151682444e-05,
      "loss": 0.5267554521560669,
      "step": 265
    },
    {
      "epoch": 0.0217834147140415,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.39987584253912e-05,
      "loss": 1.0524226427078247,
      "step": 266
    },
    {
      "epoch": 0.02186530725056045,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.39987221775845e-05,
      "loss": 1.0620660781860352,
      "step": 267
    },
    {
      "epoch": 0.021947199787079406,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3998685408263914e-05,
      "loss": 0.8309974074363708,
      "step": 268
    },
    {
      "epoch": 0.022029092323598357,
      "grad_norm": 0.19140625,
      "learning_rate": 2.399864811743104e-05,
      "loss": 0.8603475689888,
      "step": 269
    },
    {
      "epoch": 0.02211098486011731,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.39986103050875e-05,
      "loss": 0.7075223922729492,
      "step": 270
    },
    {
      "epoch": 0.022192877396636265,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3998571971234938e-05,
      "loss": 0.9101685285568237,
      "step": 271
    },
    {
      "epoch": 0.022274769933155216,
      "grad_norm": 0.18359375,
      "learning_rate": 2.399853311587502e-05,
      "loss": 0.6626906394958496,
      "step": 272
    },
    {
      "epoch": 0.02235666246967417,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3998493739009434e-05,
      "loss": 1.0679376125335693,
      "step": 273
    },
    {
      "epoch": 0.022438555006193125,
      "grad_norm": 0.169921875,
      "learning_rate": 2.399845384063989e-05,
      "loss": 0.9680445790290833,
      "step": 274
    },
    {
      "epoch": 0.022520447542712076,
      "grad_norm": 0.140625,
      "learning_rate": 2.3998413420768126e-05,
      "loss": 0.7406020164489746,
      "step": 275
    },
    {
      "epoch": 0.02260234007923103,
      "grad_norm": 0.216796875,
      "learning_rate": 2.3998372479395895e-05,
      "loss": 1.1959878206253052,
      "step": 276
    },
    {
      "epoch": 0.02268423261574998,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3998331016524984e-05,
      "loss": 0.8100699186325073,
      "step": 277
    },
    {
      "epoch": 0.022766125152268935,
      "grad_norm": 0.146484375,
      "learning_rate": 2.3998289032157183e-05,
      "loss": 0.9156461954116821,
      "step": 278
    },
    {
      "epoch": 0.02284801768878789,
      "grad_norm": 0.236328125,
      "learning_rate": 2.3998246526294326e-05,
      "loss": 0.8663290739059448,
      "step": 279
    },
    {
      "epoch": 0.02292991022530684,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3998203498938257e-05,
      "loss": 0.6476192474365234,
      "step": 280
    },
    {
      "epoch": 0.023011802761825795,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.399815995009085e-05,
      "loss": 0.7882580161094666,
      "step": 281
    },
    {
      "epoch": 0.023093695298344746,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.399811587975399e-05,
      "loss": 0.760831892490387,
      "step": 282
    },
    {
      "epoch": 0.0231755878348637,
      "grad_norm": 0.18359375,
      "learning_rate": 2.39980712879296e-05,
      "loss": 0.7089680433273315,
      "step": 283
    },
    {
      "epoch": 0.023257480371382654,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3998026174619617e-05,
      "loss": 0.902609646320343,
      "step": 284
    },
    {
      "epoch": 0.023339372907901605,
      "grad_norm": 0.142578125,
      "learning_rate": 2.3997980539825998e-05,
      "loss": 0.6368549466133118,
      "step": 285
    },
    {
      "epoch": 0.02342126544442056,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3997934383550727e-05,
      "loss": 1.0773712396621704,
      "step": 286
    },
    {
      "epoch": 0.02350315798093951,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3997887705795815e-05,
      "loss": 0.6845366358757019,
      "step": 287
    },
    {
      "epoch": 0.023585050517458465,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3997840506563286e-05,
      "loss": 0.8296812772750854,
      "step": 288
    },
    {
      "epoch": 0.02366694305397742,
      "grad_norm": 0.244140625,
      "learning_rate": 2.3997792785855194e-05,
      "loss": 0.964162290096283,
      "step": 289
    },
    {
      "epoch": 0.02374883559049637,
      "grad_norm": 0.220703125,
      "learning_rate": 2.3997744543673612e-05,
      "loss": 1.0193179845809937,
      "step": 290
    },
    {
      "epoch": 0.023830728127015324,
      "grad_norm": 0.15625,
      "learning_rate": 2.3997695780020636e-05,
      "loss": 0.8332913517951965,
      "step": 291
    },
    {
      "epoch": 0.02391262066353428,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3997646494898384e-05,
      "loss": 0.930297315120697,
      "step": 292
    },
    {
      "epoch": 0.02399451320005323,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3997596688309006e-05,
      "loss": 0.867175817489624,
      "step": 293
    },
    {
      "epoch": 0.024076405736572184,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3997546360254664e-05,
      "loss": 0.875670313835144,
      "step": 294
    },
    {
      "epoch": 0.024158298273091135,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3997495510737535e-05,
      "loss": 0.8585609197616577,
      "step": 295
    },
    {
      "epoch": 0.02424019080961009,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3997444139759842e-05,
      "loss": 0.7514534592628479,
      "step": 296
    },
    {
      "epoch": 0.024322083346129043,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.399739224732381e-05,
      "loss": 0.8820545077323914,
      "step": 297
    },
    {
      "epoch": 0.024403975882647994,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.39973398334317e-05,
      "loss": 0.8920679688453674,
      "step": 298
    },
    {
      "epoch": 0.02448586841916695,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3997286898085784e-05,
      "loss": 1.2037074565887451,
      "step": 299
    },
    {
      "epoch": 0.0245677609556859,
      "grad_norm": 0.1875,
      "learning_rate": 2.3997233441288368e-05,
      "loss": 0.6516916155815125,
      "step": 300
    },
    {
      "epoch": 0.024649653492204854,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3997179463041772e-05,
      "loss": 0.6991357803344727,
      "step": 301
    },
    {
      "epoch": 0.024731546028723808,
      "grad_norm": 0.18359375,
      "learning_rate": 2.399712496334835e-05,
      "loss": 0.7480301260948181,
      "step": 302
    },
    {
      "epoch": 0.02481343856524276,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3997069942210458e-05,
      "loss": 0.8925761580467224,
      "step": 303
    },
    {
      "epoch": 0.024895331101761713,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3997014399630496e-05,
      "loss": 0.838609516620636,
      "step": 304
    },
    {
      "epoch": 0.024977223638280668,
      "grad_norm": 0.138671875,
      "learning_rate": 2.3996958335610875e-05,
      "loss": 0.7044812440872192,
      "step": 305
    },
    {
      "epoch": 0.02505911617479962,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3996901750154035e-05,
      "loss": 0.78529292345047,
      "step": 306
    },
    {
      "epoch": 0.025141008711318573,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3996844643262432e-05,
      "loss": 1.0998769998550415,
      "step": 307
    },
    {
      "epoch": 0.025222901247837524,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3996787014938552e-05,
      "loss": 0.8928024768829346,
      "step": 308
    },
    {
      "epoch": 0.025304793784356478,
      "grad_norm": 0.20703125,
      "learning_rate": 2.3996728865184897e-05,
      "loss": 0.8206113576889038,
      "step": 309
    },
    {
      "epoch": 0.025386686320875432,
      "grad_norm": 0.169921875,
      "learning_rate": 2.399667019400399e-05,
      "loss": 0.6971143484115601,
      "step": 310
    },
    {
      "epoch": 0.025468578857394383,
      "grad_norm": 0.1953125,
      "learning_rate": 2.399661100139839e-05,
      "loss": 0.9175295829772949,
      "step": 311
    },
    {
      "epoch": 0.025550471393913338,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.399655128737066e-05,
      "loss": 0.6745611429214478,
      "step": 312
    },
    {
      "epoch": 0.02563236393043229,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3996491051923406e-05,
      "loss": 1.0425691604614258,
      "step": 313
    },
    {
      "epoch": 0.025714256466951243,
      "grad_norm": 0.232421875,
      "learning_rate": 2.399643029505924e-05,
      "loss": 0.807881236076355,
      "step": 314
    },
    {
      "epoch": 0.025796149003470197,
      "grad_norm": 0.169921875,
      "learning_rate": 2.39963690167808e-05,
      "loss": 0.9455503225326538,
      "step": 315
    },
    {
      "epoch": 0.025878041539989148,
      "grad_norm": 0.181640625,
      "learning_rate": 2.399630721709076e-05,
      "loss": 0.6920031309127808,
      "step": 316
    },
    {
      "epoch": 0.025959934076508102,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3996244895991797e-05,
      "loss": 1.0862457752227783,
      "step": 317
    },
    {
      "epoch": 0.026041826613027057,
      "grad_norm": 0.1298828125,
      "learning_rate": 2.399618205348662e-05,
      "loss": 0.6960668563842773,
      "step": 318
    },
    {
      "epoch": 0.026123719149546008,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3996118689577964e-05,
      "loss": 0.9928886294364929,
      "step": 319
    },
    {
      "epoch": 0.026205611686064962,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.399605480426858e-05,
      "loss": 0.8675575256347656,
      "step": 320
    },
    {
      "epoch": 0.026287504222583913,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3995990397561248e-05,
      "loss": 0.8609463572502136,
      "step": 321
    },
    {
      "epoch": 0.026369396759102867,
      "grad_norm": 0.52734375,
      "learning_rate": 2.3995925469458765e-05,
      "loss": 0.8016886115074158,
      "step": 322
    },
    {
      "epoch": 0.02645128929562182,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.399586001996395e-05,
      "loss": 0.7719072103500366,
      "step": 323
    },
    {
      "epoch": 0.026533181832140772,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3995794049079652e-05,
      "loss": 0.9414935111999512,
      "step": 324
    },
    {
      "epoch": 0.026615074368659727,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.399572755680874e-05,
      "loss": 0.6840108633041382,
      "step": 325
    },
    {
      "epoch": 0.026696966905178678,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.39956605431541e-05,
      "loss": 0.9565162658691406,
      "step": 326
    },
    {
      "epoch": 0.026778859441697632,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3995593008118647e-05,
      "loss": 0.5880173444747925,
      "step": 327
    },
    {
      "epoch": 0.026860751978216586,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3995524951705315e-05,
      "loss": 1.1901681423187256,
      "step": 328
    },
    {
      "epoch": 0.026942644514735537,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3995456373917064e-05,
      "loss": 0.7729980945587158,
      "step": 329
    },
    {
      "epoch": 0.02702453705125449,
      "grad_norm": 0.228515625,
      "learning_rate": 2.3995387274756873e-05,
      "loss": 0.8738072514533997,
      "step": 330
    },
    {
      "epoch": 0.027106429587773446,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3995317654227747e-05,
      "loss": 0.73838210105896,
      "step": 331
    },
    {
      "epoch": 0.027188322124292397,
      "grad_norm": 0.14453125,
      "learning_rate": 2.399524751233271e-05,
      "loss": 0.6781070828437805,
      "step": 332
    },
    {
      "epoch": 0.02727021466081135,
      "grad_norm": 0.201171875,
      "learning_rate": 2.399517684907481e-05,
      "loss": 1.1951383352279663,
      "step": 333
    },
    {
      "epoch": 0.027352107197330302,
      "grad_norm": 0.154296875,
      "learning_rate": 2.399510566445712e-05,
      "loss": 0.5185015201568604,
      "step": 334
    },
    {
      "epoch": 0.027433999733849256,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3995033958482734e-05,
      "loss": 0.7132613062858582,
      "step": 335
    },
    {
      "epoch": 0.02751589227036821,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3994961731154767e-05,
      "loss": 0.6685208678245544,
      "step": 336
    },
    {
      "epoch": 0.02759778480688716,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.399488898247636e-05,
      "loss": 0.5686011910438538,
      "step": 337
    },
    {
      "epoch": 0.027679677343406116,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3994815712450675e-05,
      "loss": 0.8996471762657166,
      "step": 338
    },
    {
      "epoch": 0.027761569879925067,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3994741921080894e-05,
      "loss": 0.5920256972312927,
      "step": 339
    },
    {
      "epoch": 0.02784346241644402,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.399466760837023e-05,
      "loss": 0.7819945216178894,
      "step": 340
    },
    {
      "epoch": 0.027925354952962975,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3994592774321908e-05,
      "loss": 0.8822646141052246,
      "step": 341
    },
    {
      "epoch": 0.028007247489481926,
      "grad_norm": 0.3125,
      "learning_rate": 2.399451741893918e-05,
      "loss": 1.0437325239181519,
      "step": 342
    },
    {
      "epoch": 0.02808914002600088,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3994441542225326e-05,
      "loss": 0.7110456824302673,
      "step": 343
    },
    {
      "epoch": 0.028171032562519835,
      "grad_norm": 0.1875,
      "learning_rate": 2.399436514418364e-05,
      "loss": 1.2693532705307007,
      "step": 344
    },
    {
      "epoch": 0.028252925099038786,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3994288224817447e-05,
      "loss": 0.8693782091140747,
      "step": 345
    },
    {
      "epoch": 0.02833481763555774,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.399421078413008e-05,
      "loss": 0.9717293381690979,
      "step": 346
    },
    {
      "epoch": 0.02841671017207669,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3994132822124916e-05,
      "loss": 0.9909393191337585,
      "step": 347
    },
    {
      "epoch": 0.028498602708595645,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.399405433880534e-05,
      "loss": 0.7639890909194946,
      "step": 348
    },
    {
      "epoch": 0.0285804952451146,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.399397533417476e-05,
      "loss": 0.8804298639297485,
      "step": 349
    },
    {
      "epoch": 0.02866238778163355,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3993895808236613e-05,
      "loss": 0.6749045848846436,
      "step": 350
    },
    {
      "epoch": 0.028744280318152505,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3993815760994358e-05,
      "loss": 0.9385254383087158,
      "step": 351
    },
    {
      "epoch": 0.028826172854671456,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3993735192451465e-05,
      "loss": 0.9184956550598145,
      "step": 352
    },
    {
      "epoch": 0.02890806539119041,
      "grad_norm": 0.19140625,
      "learning_rate": 2.399365410261145e-05,
      "loss": 0.8687407970428467,
      "step": 353
    },
    {
      "epoch": 0.028989957927709364,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.399357249147782e-05,
      "loss": 0.7708573937416077,
      "step": 354
    },
    {
      "epoch": 0.029071850464228315,
      "grad_norm": 0.1796875,
      "learning_rate": 2.399349035905414e-05,
      "loss": 0.9318118691444397,
      "step": 355
    },
    {
      "epoch": 0.02915374300074727,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3993407705343968e-05,
      "loss": 0.8875454664230347,
      "step": 356
    },
    {
      "epoch": 0.029235635537266224,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3993324530350898e-05,
      "loss": 0.8044502139091492,
      "step": 357
    },
    {
      "epoch": 0.029317528073785175,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3993240834078553e-05,
      "loss": 0.6632757782936096,
      "step": 358
    },
    {
      "epoch": 0.02939942061030413,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3993156616530558e-05,
      "loss": 0.9331107139587402,
      "step": 359
    },
    {
      "epoch": 0.02948131314682308,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3993071877710586e-05,
      "loss": 1.1078283786773682,
      "step": 360
    },
    {
      "epoch": 0.029563205683342034,
      "grad_norm": 0.177734375,
      "learning_rate": 2.399298661762231e-05,
      "loss": 1.0519784688949585,
      "step": 361
    },
    {
      "epoch": 0.02964509821986099,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3992900836269444e-05,
      "loss": 0.8882368803024292,
      "step": 362
    },
    {
      "epoch": 0.02972699075637994,
      "grad_norm": 0.203125,
      "learning_rate": 2.399281453365571e-05,
      "loss": 1.2736127376556396,
      "step": 363
    },
    {
      "epoch": 0.029808883292898894,
      "grad_norm": 0.177734375,
      "learning_rate": 2.399272770978486e-05,
      "loss": 1.0872962474822998,
      "step": 364
    },
    {
      "epoch": 0.029890775829417845,
      "grad_norm": 0.263671875,
      "learning_rate": 2.399264036466067e-05,
      "loss": 1.2390316724777222,
      "step": 365
    },
    {
      "epoch": 0.0299726683659368,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3992552498286938e-05,
      "loss": 0.9149156212806702,
      "step": 366
    },
    {
      "epoch": 0.030054560902455753,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3992464110667476e-05,
      "loss": 0.7826298475265503,
      "step": 367
    },
    {
      "epoch": 0.030136453438974704,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3992375201806138e-05,
      "loss": 0.9323840737342834,
      "step": 368
    },
    {
      "epoch": 0.03021834597549366,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3992285771706774e-05,
      "loss": 0.7545689344406128,
      "step": 369
    },
    {
      "epoch": 0.030300238512012613,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3992195820373283e-05,
      "loss": 0.6738094687461853,
      "step": 370
    },
    {
      "epoch": 0.030382131048531564,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3992105347809564e-05,
      "loss": 0.7620134949684143,
      "step": 371
    },
    {
      "epoch": 0.030464023585050518,
      "grad_norm": 0.146484375,
      "learning_rate": 2.399201435401956e-05,
      "loss": 0.6103556156158447,
      "step": 372
    },
    {
      "epoch": 0.03054591612156947,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.399192283900722e-05,
      "loss": 0.6378960609436035,
      "step": 373
    },
    {
      "epoch": 0.030627808658088423,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3991830802776518e-05,
      "loss": 0.8057185411453247,
      "step": 374
    },
    {
      "epoch": 0.030709701194607378,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3991738245331465e-05,
      "loss": 0.9602435231208801,
      "step": 375
    },
    {
      "epoch": 0.03079159373112633,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3991645166676073e-05,
      "loss": 0.9772658348083496,
      "step": 376
    },
    {
      "epoch": 0.030873486267645283,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3991551566814392e-05,
      "loss": 1.1092474460601807,
      "step": 377
    },
    {
      "epoch": 0.030955378804164234,
      "grad_norm": 0.177734375,
      "learning_rate": 2.399145744575049e-05,
      "loss": 0.8104967474937439,
      "step": 378
    },
    {
      "epoch": 0.031037271340683188,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.399136280348846e-05,
      "loss": 1.0451630353927612,
      "step": 379
    },
    {
      "epoch": 0.031119163877202143,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3991267640032415e-05,
      "loss": 0.6408492922782898,
      "step": 380
    },
    {
      "epoch": 0.031201056413721093,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3991171955386485e-05,
      "loss": 1.005161166191101,
      "step": 381
    },
    {
      "epoch": 0.031282948950240044,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3991075749554837e-05,
      "loss": 0.7929934859275818,
      "step": 382
    },
    {
      "epoch": 0.031364841486759,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.3990979022541648e-05,
      "loss": 0.9439051151275635,
      "step": 383
    },
    {
      "epoch": 0.03144673402327795,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3990881774351126e-05,
      "loss": 1.241767168045044,
      "step": 384
    },
    {
      "epoch": 0.03152862655979691,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.3990784004987496e-05,
      "loss": 0.7989773750305176,
      "step": 385
    },
    {
      "epoch": 0.03161051909631586,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3990685714455004e-05,
      "loss": 0.6661219596862793,
      "step": 386
    },
    {
      "epoch": 0.03169241163283481,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3990586902757926e-05,
      "loss": 0.8293935060501099,
      "step": 387
    },
    {
      "epoch": 0.03177430416935376,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3990487569900553e-05,
      "loss": 0.7385613918304443,
      "step": 388
    },
    {
      "epoch": 0.03185619670587272,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3990387715887206e-05,
      "loss": 1.1216137409210205,
      "step": 389
    },
    {
      "epoch": 0.03193808924239167,
      "grad_norm": 0.16796875,
      "learning_rate": 2.399028734072223e-05,
      "loss": 0.8001031875610352,
      "step": 390
    },
    {
      "epoch": 0.032019981778910626,
      "grad_norm": 0.23046875,
      "learning_rate": 2.3990186444409972e-05,
      "loss": 0.6830602288246155,
      "step": 391
    },
    {
      "epoch": 0.03210187431542958,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3990085026954836e-05,
      "loss": 0.9557156562805176,
      "step": 392
    },
    {
      "epoch": 0.03218376685194853,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3989983088361216e-05,
      "loss": 0.6188973188400269,
      "step": 393
    },
    {
      "epoch": 0.03226565938846748,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.398988062863355e-05,
      "loss": 0.7792149782180786,
      "step": 394
    },
    {
      "epoch": 0.03234755192498644,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.398977764777629e-05,
      "loss": 1.051107406616211,
      "step": 395
    },
    {
      "epoch": 0.03242944446150539,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3989674145793907e-05,
      "loss": 1.1782621145248413,
      "step": 396
    },
    {
      "epoch": 0.032511336998024346,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3989570122690906e-05,
      "loss": 0.9232417941093445,
      "step": 397
    },
    {
      "epoch": 0.03259322953454329,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3989465578471806e-05,
      "loss": 0.7142661213874817,
      "step": 398
    },
    {
      "epoch": 0.03267512207106225,
      "grad_norm": 0.18359375,
      "learning_rate": 2.398936051314115e-05,
      "loss": 0.8970451354980469,
      "step": 399
    },
    {
      "epoch": 0.0327570146075812,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.398925492670351e-05,
      "loss": 0.8157063722610474,
      "step": 400
    },
    {
      "epoch": 0.032838907144100156,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3989148819163468e-05,
      "loss": 0.6992770433425903,
      "step": 401
    },
    {
      "epoch": 0.03292079968061911,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3989042190525637e-05,
      "loss": 0.5834121108055115,
      "step": 402
    },
    {
      "epoch": 0.03300269221713806,
      "grad_norm": 0.26953125,
      "learning_rate": 2.3988935040794654e-05,
      "loss": 0.6128764748573303,
      "step": 403
    },
    {
      "epoch": 0.03308458475365701,
      "grad_norm": 0.18359375,
      "learning_rate": 2.398882736997518e-05,
      "loss": 1.056875228881836,
      "step": 404
    },
    {
      "epoch": 0.033166477290175966,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3988719178071884e-05,
      "loss": 0.7617940902709961,
      "step": 405
    },
    {
      "epoch": 0.03324836982669492,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3988610465089474e-05,
      "loss": 0.6815392374992371,
      "step": 406
    },
    {
      "epoch": 0.033330262363213875,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.398850123103268e-05,
      "loss": 0.9706563353538513,
      "step": 407
    },
    {
      "epoch": 0.03341215489973282,
      "grad_norm": 0.189453125,
      "learning_rate": 2.398839147590624e-05,
      "loss": 1.1680763959884644,
      "step": 408
    },
    {
      "epoch": 0.03349404743625178,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.398828119971494e-05,
      "loss": 0.8144336938858032,
      "step": 409
    },
    {
      "epoch": 0.03357593997277073,
      "grad_norm": 0.208984375,
      "learning_rate": 2.3988170402463554e-05,
      "loss": 0.9499691128730774,
      "step": 410
    },
    {
      "epoch": 0.033657832509289685,
      "grad_norm": 0.125,
      "learning_rate": 2.398805908415691e-05,
      "loss": 0.4251038730144501,
      "step": 411
    },
    {
      "epoch": 0.03373972504580864,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3987947244799843e-05,
      "loss": 0.8208865523338318,
      "step": 412
    },
    {
      "epoch": 0.03382161758232759,
      "grad_norm": 0.265625,
      "learning_rate": 2.398783488439721e-05,
      "loss": 0.991141140460968,
      "step": 413
    },
    {
      "epoch": 0.03390351011884654,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.3987722002953903e-05,
      "loss": 1.2578606605529785,
      "step": 414
    },
    {
      "epoch": 0.033985402655365496,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3987608600474825e-05,
      "loss": 1.044303059577942,
      "step": 415
    },
    {
      "epoch": 0.03406729519188445,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.39874946769649e-05,
      "loss": 0.9341923594474792,
      "step": 416
    },
    {
      "epoch": 0.034149187728403405,
      "grad_norm": 0.271484375,
      "learning_rate": 2.398738023242909e-05,
      "loss": 0.9305179119110107,
      "step": 417
    },
    {
      "epoch": 0.03423108026492236,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3987265266872357e-05,
      "loss": 1.0452042818069458,
      "step": 418
    },
    {
      "epoch": 0.034312972801441306,
      "grad_norm": 0.171875,
      "learning_rate": 2.3987149780299704e-05,
      "loss": 0.7530025839805603,
      "step": 419
    },
    {
      "epoch": 0.03439486533796026,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.398703377271615e-05,
      "loss": 0.7470859885215759,
      "step": 420
    },
    {
      "epoch": 0.034476757874479215,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3986917244126736e-05,
      "loss": 0.7884257435798645,
      "step": 421
    },
    {
      "epoch": 0.03455865041099817,
      "grad_norm": 0.345703125,
      "learning_rate": 2.398680019453653e-05,
      "loss": 1.1965625286102295,
      "step": 422
    },
    {
      "epoch": 0.034640542947517124,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3986682623950618e-05,
      "loss": 1.0573604106903076,
      "step": 423
    },
    {
      "epoch": 0.03472243548403607,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3986564532374112e-05,
      "loss": 0.8759225606918335,
      "step": 424
    },
    {
      "epoch": 0.034804328020555025,
      "grad_norm": 0.171875,
      "learning_rate": 2.3986445919812137e-05,
      "loss": 0.8650780916213989,
      "step": 425
    },
    {
      "epoch": 0.03488622055707398,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.398632678626986e-05,
      "loss": 0.5486708879470825,
      "step": 426
    },
    {
      "epoch": 0.034968113093592934,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.3986207131752448e-05,
      "loss": 1.3964512348175049,
      "step": 427
    },
    {
      "epoch": 0.03505000563011189,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3986086956265106e-05,
      "loss": 1.2789416313171387,
      "step": 428
    },
    {
      "epoch": 0.035131898166630836,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3985966259813063e-05,
      "loss": 1.0349560976028442,
      "step": 429
    },
    {
      "epoch": 0.03521379070314979,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3985845042401554e-05,
      "loss": 0.845013439655304,
      "step": 430
    },
    {
      "epoch": 0.035295683239668744,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3985723304035855e-05,
      "loss": 0.8173735737800598,
      "step": 431
    },
    {
      "epoch": 0.0353775757761877,
      "grad_norm": 0.1376953125,
      "learning_rate": 2.3985601044721255e-05,
      "loss": 0.6505873799324036,
      "step": 432
    },
    {
      "epoch": 0.03545946831270665,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3985478264463072e-05,
      "loss": 0.8953864574432373,
      "step": 433
    },
    {
      "epoch": 0.0355413608492256,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3985354963266637e-05,
      "loss": 0.7566075921058655,
      "step": 434
    },
    {
      "epoch": 0.035623253385744555,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3985231141137312e-05,
      "loss": 1.0244107246398926,
      "step": 435
    },
    {
      "epoch": 0.03570514592226351,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3985106798080475e-05,
      "loss": 0.9059413075447083,
      "step": 436
    },
    {
      "epoch": 0.035787038458782464,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3984981934101537e-05,
      "loss": 0.833506166934967,
      "step": 437
    },
    {
      "epoch": 0.03586893099530142,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.398485654920592e-05,
      "loss": 0.8970056772232056,
      "step": 438
    },
    {
      "epoch": 0.035950823531820365,
      "grad_norm": 0.203125,
      "learning_rate": 2.3984730643399077e-05,
      "loss": 0.8806341886520386,
      "step": 439
    },
    {
      "epoch": 0.03603271606833932,
      "grad_norm": 0.177734375,
      "learning_rate": 2.398460421668648e-05,
      "loss": 0.8865904808044434,
      "step": 440
    },
    {
      "epoch": 0.036114608604858274,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3984477269073617e-05,
      "loss": 0.7775003910064697,
      "step": 441
    },
    {
      "epoch": 0.03619650114137723,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3984349800566015e-05,
      "loss": 0.9406428337097168,
      "step": 442
    },
    {
      "epoch": 0.03627839367789618,
      "grad_norm": 0.166015625,
      "learning_rate": 2.398422181116921e-05,
      "loss": 0.9077646136283875,
      "step": 443
    },
    {
      "epoch": 0.03636028621441514,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3984093300888767e-05,
      "loss": 0.7018104791641235,
      "step": 444
    },
    {
      "epoch": 0.036442178750934084,
      "grad_norm": 0.220703125,
      "learning_rate": 2.3983964269730267e-05,
      "loss": 0.9777675867080688,
      "step": 445
    },
    {
      "epoch": 0.03652407128745304,
      "grad_norm": 0.1875,
      "learning_rate": 2.3983834717699326e-05,
      "loss": 0.7176735401153564,
      "step": 446
    },
    {
      "epoch": 0.03660596382397199,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3983704644801568e-05,
      "loss": 0.6879844069480896,
      "step": 447
    },
    {
      "epoch": 0.03668785636049095,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.398357405104265e-05,
      "loss": 0.7447598576545715,
      "step": 448
    },
    {
      "epoch": 0.0367697488970099,
      "grad_norm": 0.1875,
      "learning_rate": 2.3983442936428245e-05,
      "loss": 0.9589807987213135,
      "step": 449
    },
    {
      "epoch": 0.03685164143352885,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3983311300964055e-05,
      "loss": 1.155102252960205,
      "step": 450
    },
    {
      "epoch": 0.036933533970047804,
      "grad_norm": 0.203125,
      "learning_rate": 2.3983179144655802e-05,
      "loss": 1.1781085729599,
      "step": 451
    },
    {
      "epoch": 0.03701542650656676,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3983046467509227e-05,
      "loss": 1.3224983215332031,
      "step": 452
    },
    {
      "epoch": 0.03709731904308571,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3982913269530098e-05,
      "loss": 1.1346737146377563,
      "step": 453
    },
    {
      "epoch": 0.03717921157960467,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3982779550724203e-05,
      "loss": 0.901663064956665,
      "step": 454
    },
    {
      "epoch": 0.037261104116123614,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3982645311097356e-05,
      "loss": 0.7664777636528015,
      "step": 455
    },
    {
      "epoch": 0.03734299665264257,
      "grad_norm": 0.1640625,
      "learning_rate": 2.398251055065539e-05,
      "loss": 0.9744863510131836,
      "step": 456
    },
    {
      "epoch": 0.03742488918916152,
      "grad_norm": 0.1875,
      "learning_rate": 2.3982375269404166e-05,
      "loss": 1.1253559589385986,
      "step": 457
    },
    {
      "epoch": 0.03750678172568048,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3982239467349557e-05,
      "loss": 0.750821053981781,
      "step": 458
    },
    {
      "epoch": 0.03758867426219943,
      "grad_norm": 0.16015625,
      "learning_rate": 2.398210314449747e-05,
      "loss": 0.8490447998046875,
      "step": 459
    },
    {
      "epoch": 0.03767056679871838,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3981966300853833e-05,
      "loss": 0.9658976793289185,
      "step": 460
    },
    {
      "epoch": 0.03775245933523733,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3981828936424588e-05,
      "loss": 0.8551173806190491,
      "step": 461
    },
    {
      "epoch": 0.03783435187175629,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.398169105121571e-05,
      "loss": 1.0039170980453491,
      "step": 462
    },
    {
      "epoch": 0.03791624440827524,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.398155264523319e-05,
      "loss": 0.6282235383987427,
      "step": 463
    },
    {
      "epoch": 0.037998136944794196,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3981413718483042e-05,
      "loss": 0.8205994963645935,
      "step": 464
    },
    {
      "epoch": 0.03808002948131314,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3981274270971307e-05,
      "loss": 0.8125453591346741,
      "step": 465
    },
    {
      "epoch": 0.0381619220178321,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3981134302704045e-05,
      "loss": 1.2362008094787598,
      "step": 466
    },
    {
      "epoch": 0.03824381455435105,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.3980993813687338e-05,
      "loss": 1.047678828239441,
      "step": 467
    },
    {
      "epoch": 0.038325707090870006,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3980852803927293e-05,
      "loss": 0.8089118003845215,
      "step": 468
    },
    {
      "epoch": 0.03840759962738896,
      "grad_norm": 0.193359375,
      "learning_rate": 2.398071127343004e-05,
      "loss": 1.0405510663986206,
      "step": 469
    },
    {
      "epoch": 0.038489492163907915,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3980569222201734e-05,
      "loss": 1.1824989318847656,
      "step": 470
    },
    {
      "epoch": 0.03857138470042686,
      "grad_norm": 0.55078125,
      "learning_rate": 2.398042665024854e-05,
      "loss": 0.8795514702796936,
      "step": 471
    },
    {
      "epoch": 0.03865327723694582,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3980283557576664e-05,
      "loss": 0.7451978325843811,
      "step": 472
    },
    {
      "epoch": 0.03873516977346477,
      "grad_norm": 0.2431640625,
      "learning_rate": 2.3980139944192323e-05,
      "loss": 1.1166579723358154,
      "step": 473
    },
    {
      "epoch": 0.038817062309983726,
      "grad_norm": 0.173828125,
      "learning_rate": 2.397999581010175e-05,
      "loss": 0.8764607310295105,
      "step": 474
    },
    {
      "epoch": 0.03889895484650268,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3979851155311226e-05,
      "loss": 0.7599079608917236,
      "step": 475
    },
    {
      "epoch": 0.03898084738302163,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3979705979827023e-05,
      "loss": 1.1147818565368652,
      "step": 476
    },
    {
      "epoch": 0.03906273991954058,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3979560283655458e-05,
      "loss": 0.979387104511261,
      "step": 477
    },
    {
      "epoch": 0.039144632456059536,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3979414066802864e-05,
      "loss": 0.7678375244140625,
      "step": 478
    },
    {
      "epoch": 0.03922652499257849,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3979267329275597e-05,
      "loss": 0.6357822418212891,
      "step": 479
    },
    {
      "epoch": 0.039308417529097445,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.397912007108003e-05,
      "loss": 0.7360219955444336,
      "step": 480
    },
    {
      "epoch": 0.03939031006561639,
      "grad_norm": 0.18359375,
      "learning_rate": 2.397897229222257e-05,
      "loss": 0.8464394211769104,
      "step": 481
    },
    {
      "epoch": 0.039472202602135346,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3978823992709632e-05,
      "loss": 0.779076099395752,
      "step": 482
    },
    {
      "epoch": 0.0395540951386543,
      "grad_norm": 0.1875,
      "learning_rate": 2.397867517254767e-05,
      "loss": 0.8278973698616028,
      "step": 483
    },
    {
      "epoch": 0.039635987675173255,
      "grad_norm": 0.138671875,
      "learning_rate": 2.3978525831743145e-05,
      "loss": 0.6052179336547852,
      "step": 484
    },
    {
      "epoch": 0.03971788021169221,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3978375970302556e-05,
      "loss": 0.8245387673377991,
      "step": 485
    },
    {
      "epoch": 0.03979977274821116,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3978225588232407e-05,
      "loss": 0.7775510549545288,
      "step": 486
    },
    {
      "epoch": 0.03988166528473011,
      "grad_norm": 0.1318359375,
      "learning_rate": 2.397807468553924e-05,
      "loss": 0.786597490310669,
      "step": 487
    },
    {
      "epoch": 0.039963557821249066,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3977923262229616e-05,
      "loss": 0.9905457496643066,
      "step": 488
    },
    {
      "epoch": 0.04004545035776802,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3977771318310113e-05,
      "loss": 0.8521064519882202,
      "step": 489
    },
    {
      "epoch": 0.040127342894286974,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3977618853787332e-05,
      "loss": 0.7497962713241577,
      "step": 490
    },
    {
      "epoch": 0.04020923543080592,
      "grad_norm": 0.146484375,
      "learning_rate": 2.3977465868667906e-05,
      "loss": 0.6652951836585999,
      "step": 491
    },
    {
      "epoch": 0.040291127967324876,
      "grad_norm": 0.17578125,
      "learning_rate": 2.397731236295848e-05,
      "loss": 0.806548535823822,
      "step": 492
    },
    {
      "epoch": 0.04037302050384383,
      "grad_norm": 0.216796875,
      "learning_rate": 2.397715833666573e-05,
      "loss": 0.9164248704910278,
      "step": 493
    },
    {
      "epoch": 0.040454913040362785,
      "grad_norm": 0.16796875,
      "learning_rate": 2.397700378979635e-05,
      "loss": 1.0114997625350952,
      "step": 494
    },
    {
      "epoch": 0.04053680557688174,
      "grad_norm": 0.203125,
      "learning_rate": 2.3976848722357055e-05,
      "loss": 0.9673534631729126,
      "step": 495
    },
    {
      "epoch": 0.04061869811340069,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.3976693134354582e-05,
      "loss": 1.0153312683105469,
      "step": 496
    },
    {
      "epoch": 0.04070059064991964,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.39765370257957e-05,
      "loss": 0.7527498006820679,
      "step": 497
    },
    {
      "epoch": 0.040782483186438595,
      "grad_norm": 0.166015625,
      "learning_rate": 2.397638039668719e-05,
      "loss": 0.6898579597473145,
      "step": 498
    },
    {
      "epoch": 0.04086437572295755,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.397622324703586e-05,
      "loss": 0.6689336895942688,
      "step": 499
    },
    {
      "epoch": 0.040946268259476504,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3976065576848543e-05,
      "loss": 0.9839563369750977,
      "step": 500
    },
    {
      "epoch": 0.04102816079599546,
      "grad_norm": 0.1875,
      "learning_rate": 2.397590738613209e-05,
      "loss": 0.7965925335884094,
      "step": 501
    },
    {
      "epoch": 0.041110053332514405,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3975748674893375e-05,
      "loss": 0.7120146155357361,
      "step": 502
    },
    {
      "epoch": 0.04119194586903336,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.39755894431393e-05,
      "loss": 0.9516551494598389,
      "step": 503
    },
    {
      "epoch": 0.041273838405552314,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3975429690876786e-05,
      "loss": 1.1224391460418701,
      "step": 504
    },
    {
      "epoch": 0.04135573094207127,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3975269418112772e-05,
      "loss": 1.016868233680725,
      "step": 505
    },
    {
      "epoch": 0.04143762347859022,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3975108624854225e-05,
      "loss": 1.1816296577453613,
      "step": 506
    },
    {
      "epoch": 0.04151951601510917,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3974947311108136e-05,
      "loss": 0.7898935675621033,
      "step": 507
    },
    {
      "epoch": 0.041601408551628125,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3974785476881516e-05,
      "loss": 1.2441715002059937,
      "step": 508
    },
    {
      "epoch": 0.04168330108814708,
      "grad_norm": 0.166015625,
      "learning_rate": 2.39746231221814e-05,
      "loss": 0.5922395586967468,
      "step": 509
    },
    {
      "epoch": 0.04176519362466603,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.397446024701484e-05,
      "loss": 0.7211483716964722,
      "step": 510
    },
    {
      "epoch": 0.04184708616118499,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3974296851388924e-05,
      "loss": 0.9107987880706787,
      "step": 511
    },
    {
      "epoch": 0.041928978697703935,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3974132935310744e-05,
      "loss": 0.7573146820068359,
      "step": 512
    },
    {
      "epoch": 0.04201087123422289,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3973968498787433e-05,
      "loss": 0.7425338625907898,
      "step": 513
    },
    {
      "epoch": 0.042092763770741844,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3973803541826128e-05,
      "loss": 0.7786843776702881,
      "step": 514
    },
    {
      "epoch": 0.0421746563072608,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3973638064434007e-05,
      "loss": 0.938799262046814,
      "step": 515
    },
    {
      "epoch": 0.04225654884377975,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3973472066618264e-05,
      "loss": 0.8231399059295654,
      "step": 516
    },
    {
      "epoch": 0.0423384413802987,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3973305548386108e-05,
      "loss": 0.9845247864723206,
      "step": 517
    },
    {
      "epoch": 0.042420333916817654,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3973138509744773e-05,
      "loss": 0.8089458346366882,
      "step": 518
    },
    {
      "epoch": 0.04250222645333661,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.397297095070153e-05,
      "loss": 0.7149685621261597,
      "step": 519
    },
    {
      "epoch": 0.04258411898985556,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3972802871263654e-05,
      "loss": 0.9618757963180542,
      "step": 520
    },
    {
      "epoch": 0.04266601152637452,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.397263427143845e-05,
      "loss": 0.8664198517799377,
      "step": 521
    },
    {
      "epoch": 0.04274790406289347,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3972465151233256e-05,
      "loss": 0.9404294490814209,
      "step": 522
    },
    {
      "epoch": 0.04282979659941242,
      "grad_norm": 0.171875,
      "learning_rate": 2.3972295510655412e-05,
      "loss": 0.8636691570281982,
      "step": 523
    },
    {
      "epoch": 0.04291168913593137,
      "grad_norm": 0.203125,
      "learning_rate": 2.3972125349712293e-05,
      "loss": 0.7196763753890991,
      "step": 524
    },
    {
      "epoch": 0.04299358167245033,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.39719546684113e-05,
      "loss": 1.0419987440109253,
      "step": 525
    },
    {
      "epoch": 0.04307547420896928,
      "grad_norm": 0.203125,
      "learning_rate": 2.397178346675985e-05,
      "loss": 0.8712450861930847,
      "step": 526
    },
    {
      "epoch": 0.043157366745488236,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3971611744765377e-05,
      "loss": 1.0688278675079346,
      "step": 527
    },
    {
      "epoch": 0.043239259282007184,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3971439502435356e-05,
      "loss": 0.7815065979957581,
      "step": 528
    },
    {
      "epoch": 0.04332115181852614,
      "grad_norm": 0.1875,
      "learning_rate": 2.397126673977726e-05,
      "loss": 0.8646108508110046,
      "step": 529
    },
    {
      "epoch": 0.04340304435504509,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3971093456798613e-05,
      "loss": 0.6068559288978577,
      "step": 530
    },
    {
      "epoch": 0.04348493689156405,
      "grad_norm": 0.2265625,
      "learning_rate": 2.3970919653506932e-05,
      "loss": 0.8408427238464355,
      "step": 531
    },
    {
      "epoch": 0.043566829428083,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3970745329909782e-05,
      "loss": 0.7949510216712952,
      "step": 532
    },
    {
      "epoch": 0.04364872196460195,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.397057048601474e-05,
      "loss": 1.1144850254058838,
      "step": 533
    },
    {
      "epoch": 0.0437306145011209,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.3970395121829396e-05,
      "loss": 1.0043058395385742,
      "step": 534
    },
    {
      "epoch": 0.04381250703763986,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3970219237361375e-05,
      "loss": 0.9013270735740662,
      "step": 535
    },
    {
      "epoch": 0.04389439957415881,
      "grad_norm": 0.1796875,
      "learning_rate": 2.397004283261833e-05,
      "loss": 1.0792452096939087,
      "step": 536
    },
    {
      "epoch": 0.043976292110677766,
      "grad_norm": 0.185546875,
      "learning_rate": 2.396986590760792e-05,
      "loss": 0.8610565662384033,
      "step": 537
    },
    {
      "epoch": 0.04405818464719671,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.396968846233784e-05,
      "loss": 1.0946379899978638,
      "step": 538
    },
    {
      "epoch": 0.04414007718371567,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3969510496815797e-05,
      "loss": 0.762466311454773,
      "step": 539
    },
    {
      "epoch": 0.04422196972023462,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.396933201104953e-05,
      "loss": 0.9066336750984192,
      "step": 540
    },
    {
      "epoch": 0.044303862256753576,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3969153005046797e-05,
      "loss": 0.6592026948928833,
      "step": 541
    },
    {
      "epoch": 0.04438575479327253,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3968973478815375e-05,
      "loss": 0.9944949150085449,
      "step": 542
    },
    {
      "epoch": 0.04446764732979148,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.396879343236307e-05,
      "loss": 0.9634906649589539,
      "step": 543
    },
    {
      "epoch": 0.04454953986631043,
      "grad_norm": 0.208984375,
      "learning_rate": 2.3968612865697703e-05,
      "loss": 1.0275566577911377,
      "step": 544
    },
    {
      "epoch": 0.04463143240282939,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.3968431778827134e-05,
      "loss": 0.8620836138725281,
      "step": 545
    },
    {
      "epoch": 0.04471332493934834,
      "grad_norm": 0.16796875,
      "learning_rate": 2.396825017175922e-05,
      "loss": 0.8265810608863831,
      "step": 546
    },
    {
      "epoch": 0.044795217475867295,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3968068044501862e-05,
      "loss": 0.8992477655410767,
      "step": 547
    },
    {
      "epoch": 0.04487711001238625,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.396788539706297e-05,
      "loss": 0.7925514578819275,
      "step": 548
    },
    {
      "epoch": 0.0449590025489052,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.396770222945049e-05,
      "loss": 0.9274776577949524,
      "step": 549
    },
    {
      "epoch": 0.04504089508542415,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3967518541672375e-05,
      "loss": 0.8201966285705566,
      "step": 550
    },
    {
      "epoch": 0.045122787621943106,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.3967334333736622e-05,
      "loss": 1.0400993824005127,
      "step": 551
    },
    {
      "epoch": 0.04520468015846206,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3967149605651224e-05,
      "loss": 0.6288777589797974,
      "step": 552
    },
    {
      "epoch": 0.045286572694981014,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3966964357424215e-05,
      "loss": 0.6736346483230591,
      "step": 553
    },
    {
      "epoch": 0.04536846523149996,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3966778589063648e-05,
      "loss": 0.7843412756919861,
      "step": 554
    },
    {
      "epoch": 0.045450357768018916,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3966592300577593e-05,
      "loss": 0.6924692392349243,
      "step": 555
    },
    {
      "epoch": 0.04553225030453787,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3966405491974156e-05,
      "loss": 0.7438194155693054,
      "step": 556
    },
    {
      "epoch": 0.045614142841056825,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3966218163261444e-05,
      "loss": 0.9577575922012329,
      "step": 557
    },
    {
      "epoch": 0.04569603537757578,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3966030314447608e-05,
      "loss": 0.9418614506721497,
      "step": 558
    },
    {
      "epoch": 0.045777927914094727,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3965841945540808e-05,
      "loss": 0.8478980660438538,
      "step": 559
    },
    {
      "epoch": 0.04585982045061368,
      "grad_norm": 0.1259765625,
      "learning_rate": 2.396565305654924e-05,
      "loss": 0.4260137677192688,
      "step": 560
    },
    {
      "epoch": 0.045941712987132635,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.39654636474811e-05,
      "loss": 0.9933167695999146,
      "step": 561
    },
    {
      "epoch": 0.04602360552365159,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.396527371834463e-05,
      "loss": 1.0423835515975952,
      "step": 562
    },
    {
      "epoch": 0.046105498060170544,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3965083269148086e-05,
      "loss": 0.8101381063461304,
      "step": 563
    },
    {
      "epoch": 0.04618739059668949,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.396489229989974e-05,
      "loss": 0.9623128175735474,
      "step": 564
    },
    {
      "epoch": 0.046269283133208446,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3964700810607892e-05,
      "loss": 1.0519846677780151,
      "step": 565
    },
    {
      "epoch": 0.0463511756697274,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.396450880128087e-05,
      "loss": 0.9688785672187805,
      "step": 566
    },
    {
      "epoch": 0.046433068206246354,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3964316271927016e-05,
      "loss": 0.966496467590332,
      "step": 567
    },
    {
      "epoch": 0.04651496074276531,
      "grad_norm": 0.283203125,
      "learning_rate": 2.3964123222554704e-05,
      "loss": 0.8825667500495911,
      "step": 568
    },
    {
      "epoch": 0.046596853279284256,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3963929653172312e-05,
      "loss": 0.9471298456192017,
      "step": 569
    },
    {
      "epoch": 0.04667874581580321,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3963735563788272e-05,
      "loss": 1.1921699047088623,
      "step": 570
    },
    {
      "epoch": 0.046760638352322165,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3963540954411004e-05,
      "loss": 0.9107367396354675,
      "step": 571
    },
    {
      "epoch": 0.04684253088884112,
      "grad_norm": 0.140625,
      "learning_rate": 2.3963345825048967e-05,
      "loss": 1.1056568622589111,
      "step": 572
    },
    {
      "epoch": 0.04692442342536007,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3963150175710654e-05,
      "loss": 0.5379326939582825,
      "step": 573
    },
    {
      "epoch": 0.04700631596187902,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.396295400640456e-05,
      "loss": 0.729265034198761,
      "step": 574
    },
    {
      "epoch": 0.047088208498397975,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3962757317139215e-05,
      "loss": 0.6542950868606567,
      "step": 575
    },
    {
      "epoch": 0.04717010103491693,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3962560107923165e-05,
      "loss": 0.5564326047897339,
      "step": 576
    },
    {
      "epoch": 0.047251993571435884,
      "grad_norm": 0.2294921875,
      "learning_rate": 2.3962362378764977e-05,
      "loss": 0.9876377582550049,
      "step": 577
    },
    {
      "epoch": 0.04733388610795484,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3962164129673257e-05,
      "loss": 0.9433709383010864,
      "step": 578
    },
    {
      "epoch": 0.04741577864447379,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3961965360656613e-05,
      "loss": 0.6138058304786682,
      "step": 579
    },
    {
      "epoch": 0.04749767118099274,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3961766071723686e-05,
      "loss": 1.362326741218567,
      "step": 580
    },
    {
      "epoch": 0.047579563717511694,
      "grad_norm": 0.169921875,
      "learning_rate": 2.396156626288314e-05,
      "loss": 0.7150259613990784,
      "step": 581
    },
    {
      "epoch": 0.04766145625403065,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.396136593414366e-05,
      "loss": 0.6978367567062378,
      "step": 582
    },
    {
      "epoch": 0.0477433487905496,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3961165085513947e-05,
      "loss": 0.8931869268417358,
      "step": 583
    },
    {
      "epoch": 0.04782524132706856,
      "grad_norm": 0.169921875,
      "learning_rate": 2.396096371700274e-05,
      "loss": 0.6849859356880188,
      "step": 584
    },
    {
      "epoch": 0.047907133863587505,
      "grad_norm": 0.25390625,
      "learning_rate": 2.3960761828618782e-05,
      "loss": 1.1862263679504395,
      "step": 585
    },
    {
      "epoch": 0.04798902640010646,
      "grad_norm": 0.228515625,
      "learning_rate": 2.3960559420370854e-05,
      "loss": 1.0873479843139648,
      "step": 586
    },
    {
      "epoch": 0.04807091893662541,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.396035649226775e-05,
      "loss": 0.8785818219184875,
      "step": 587
    },
    {
      "epoch": 0.04815281147314437,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3960153044318295e-05,
      "loss": 0.5999451279640198,
      "step": 588
    },
    {
      "epoch": 0.04823470400966332,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3959949076531326e-05,
      "loss": 0.8909758925437927,
      "step": 589
    },
    {
      "epoch": 0.04831659654618227,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3959744588915714e-05,
      "loss": 0.7863945364952087,
      "step": 590
    },
    {
      "epoch": 0.048398489082701224,
      "grad_norm": 0.16796875,
      "learning_rate": 2.395953958148034e-05,
      "loss": 0.5463456511497498,
      "step": 591
    },
    {
      "epoch": 0.04848038161922018,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.3959334054234116e-05,
      "loss": 0.6835671067237854,
      "step": 592
    },
    {
      "epoch": 0.04856227415573913,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.395912800718598e-05,
      "loss": 1.0347270965576172,
      "step": 593
    },
    {
      "epoch": 0.04864416669225809,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3958921440344887e-05,
      "loss": 0.8445790410041809,
      "step": 594
    },
    {
      "epoch": 0.048726059228777034,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.395871435371981e-05,
      "loss": 0.8437964916229248,
      "step": 595
    },
    {
      "epoch": 0.04880795176529599,
      "grad_norm": 0.203125,
      "learning_rate": 2.3958506747319752e-05,
      "loss": 1.0122697353363037,
      "step": 596
    },
    {
      "epoch": 0.04888984430181494,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3958298621153738e-05,
      "loss": 0.8762169480323792,
      "step": 597
    },
    {
      "epoch": 0.0489717368383339,
      "grad_norm": 0.205078125,
      "learning_rate": 2.3958089975230815e-05,
      "loss": 0.9854332804679871,
      "step": 598
    },
    {
      "epoch": 0.04905362937485285,
      "grad_norm": 0.162109375,
      "learning_rate": 2.395788080956005e-05,
      "loss": 0.8069448471069336,
      "step": 599
    },
    {
      "epoch": 0.0491355219113718,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.395767112415053e-05,
      "loss": 1.1460633277893066,
      "step": 600
    },
    {
      "epoch": 0.04921741444789075,
      "grad_norm": 0.173828125,
      "learning_rate": 2.395746091901138e-05,
      "loss": 0.9800331592559814,
      "step": 601
    },
    {
      "epoch": 0.04929930698440971,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3957250194151724e-05,
      "loss": 0.7255781888961792,
      "step": 602
    },
    {
      "epoch": 0.04938119952092866,
      "grad_norm": 0.1796875,
      "learning_rate": 2.395703894958073e-05,
      "loss": 0.889350950717926,
      "step": 603
    },
    {
      "epoch": 0.049463092057447616,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3956827185307574e-05,
      "loss": 0.8888030052185059,
      "step": 604
    },
    {
      "epoch": 0.04954498459396657,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.395661490134146e-05,
      "loss": 0.9086771011352539,
      "step": 605
    },
    {
      "epoch": 0.04962687713048552,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.395640209769162e-05,
      "loss": 0.7518740296363831,
      "step": 606
    },
    {
      "epoch": 0.04970876966700447,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3956188774367295e-05,
      "loss": 0.9608128666877747,
      "step": 607
    },
    {
      "epoch": 0.04979066220352343,
      "grad_norm": 0.255859375,
      "learning_rate": 2.3955974931377765e-05,
      "loss": 0.8345894813537598,
      "step": 608
    },
    {
      "epoch": 0.04987255474004238,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3955760568732323e-05,
      "loss": 0.9964596629142761,
      "step": 609
    },
    {
      "epoch": 0.049954447276561335,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3955545686440282e-05,
      "loss": 0.8712890148162842,
      "step": 610
    },
    {
      "epoch": 0.05003633981308028,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3955330284510987e-05,
      "loss": 1.1242917776107788,
      "step": 611
    },
    {
      "epoch": 0.05011823234959924,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.3955114362953793e-05,
      "loss": 0.6739065051078796,
      "step": 612
    },
    {
      "epoch": 0.05020012488611819,
      "grad_norm": 0.1875,
      "learning_rate": 2.3954897921778094e-05,
      "loss": 1.101578712463379,
      "step": 613
    },
    {
      "epoch": 0.050282017422637146,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3954680960993293e-05,
      "loss": 0.9087690711021423,
      "step": 614
    },
    {
      "epoch": 0.0503639099591561,
      "grad_norm": 0.1875,
      "learning_rate": 2.3954463480608816e-05,
      "loss": 0.6032750606536865,
      "step": 615
    },
    {
      "epoch": 0.05044580249567505,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3954245480634123e-05,
      "loss": 0.7097981572151184,
      "step": 616
    },
    {
      "epoch": 0.050527695032194,
      "grad_norm": 0.16015625,
      "learning_rate": 2.395402696107868e-05,
      "loss": 0.7773026823997498,
      "step": 617
    },
    {
      "epoch": 0.050609587568712956,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.3953807921951992e-05,
      "loss": 0.8052107691764832,
      "step": 618
    },
    {
      "epoch": 0.05069148010523191,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3953588363263582e-05,
      "loss": 0.8313415050506592,
      "step": 619
    },
    {
      "epoch": 0.050773372641750865,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3953368285022984e-05,
      "loss": 0.7332470417022705,
      "step": 620
    },
    {
      "epoch": 0.05085526517826981,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.395314768723977e-05,
      "loss": 0.8336531519889832,
      "step": 621
    },
    {
      "epoch": 0.05093715771478877,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3952926569923527e-05,
      "loss": 0.8546928763389587,
      "step": 622
    },
    {
      "epoch": 0.05101905025130772,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3952704933083864e-05,
      "loss": 0.8728529810905457,
      "step": 623
    },
    {
      "epoch": 0.051100942787826675,
      "grad_norm": 0.17578125,
      "learning_rate": 2.395248277673041e-05,
      "loss": 1.0662384033203125,
      "step": 624
    },
    {
      "epoch": 0.05118283532434563,
      "grad_norm": 0.2294921875,
      "learning_rate": 2.3952260100872833e-05,
      "loss": 0.9664002060890198,
      "step": 625
    },
    {
      "epoch": 0.05126472786086458,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.39520369055208e-05,
      "loss": 0.8857104182243347,
      "step": 626
    },
    {
      "epoch": 0.05134662039738353,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3951813190684022e-05,
      "loss": 0.742447555065155,
      "step": 627
    },
    {
      "epoch": 0.051428512933902486,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.395158895637221e-05,
      "loss": 0.929764986038208,
      "step": 628
    },
    {
      "epoch": 0.05151040547042144,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3951364202595123e-05,
      "loss": 0.9282650947570801,
      "step": 629
    },
    {
      "epoch": 0.051592298006940394,
      "grad_norm": 0.212890625,
      "learning_rate": 2.395113892936252e-05,
      "loss": 0.7192749381065369,
      "step": 630
    },
    {
      "epoch": 0.05167419054345935,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3950913136684196e-05,
      "loss": 0.9937841892242432,
      "step": 631
    },
    {
      "epoch": 0.051756083079978296,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.395068682456996e-05,
      "loss": 1.1982215642929077,
      "step": 632
    },
    {
      "epoch": 0.05183797561649725,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3950459993029662e-05,
      "loss": 0.6705058217048645,
      "step": 633
    },
    {
      "epoch": 0.051919868153016205,
      "grad_norm": 0.134765625,
      "learning_rate": 2.395023264207315e-05,
      "loss": 0.8769830465316772,
      "step": 634
    },
    {
      "epoch": 0.05200176068953516,
      "grad_norm": 0.1875,
      "learning_rate": 2.3950004771710306e-05,
      "loss": 0.5416737794876099,
      "step": 635
    },
    {
      "epoch": 0.052083653226054114,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.394977638195104e-05,
      "loss": 0.9033718705177307,
      "step": 636
    },
    {
      "epoch": 0.05216554576257306,
      "grad_norm": 0.1875,
      "learning_rate": 2.3949547472805267e-05,
      "loss": 1.1842782497406006,
      "step": 637
    },
    {
      "epoch": 0.052247438299092015,
      "grad_norm": 0.171875,
      "learning_rate": 2.3949318044282954e-05,
      "loss": 0.9973214268684387,
      "step": 638
    },
    {
      "epoch": 0.05232933083561097,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.394908809639406e-05,
      "loss": 0.8814473152160645,
      "step": 639
    },
    {
      "epoch": 0.052411223372129924,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3948857629148578e-05,
      "loss": 0.5170973539352417,
      "step": 640
    },
    {
      "epoch": 0.05249311590864888,
      "grad_norm": 0.220703125,
      "learning_rate": 2.3948626642556535e-05,
      "loss": 1.0104471445083618,
      "step": 641
    },
    {
      "epoch": 0.052575008445167826,
      "grad_norm": 0.142578125,
      "learning_rate": 2.394839513662796e-05,
      "loss": 0.7293509840965271,
      "step": 642
    },
    {
      "epoch": 0.05265690098168678,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3948163111372928e-05,
      "loss": 0.982900857925415,
      "step": 643
    },
    {
      "epoch": 0.052738793518205734,
      "grad_norm": 0.1875,
      "learning_rate": 2.394793056680151e-05,
      "loss": 0.9158048033714294,
      "step": 644
    },
    {
      "epoch": 0.05282068605472469,
      "grad_norm": 0.212890625,
      "learning_rate": 2.394769750292382e-05,
      "loss": 0.7885652184486389,
      "step": 645
    },
    {
      "epoch": 0.05290257859124364,
      "grad_norm": 0.140625,
      "learning_rate": 2.394746391974999e-05,
      "loss": 0.6910034418106079,
      "step": 646
    },
    {
      "epoch": 0.05298447112776259,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3947229817290167e-05,
      "loss": 0.9547855257987976,
      "step": 647
    },
    {
      "epoch": 0.053066363664281545,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3946995195554533e-05,
      "loss": 0.8249313831329346,
      "step": 648
    },
    {
      "epoch": 0.0531482562008005,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3946760054553282e-05,
      "loss": 0.5903275012969971,
      "step": 649
    },
    {
      "epoch": 0.05323014873731945,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3946524394296633e-05,
      "loss": 0.8317531943321228,
      "step": 650
    },
    {
      "epoch": 0.05331204127383841,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3946288214794828e-05,
      "loss": 0.6406201124191284,
      "step": 651
    },
    {
      "epoch": 0.053393933810357355,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3946051516058134e-05,
      "loss": 0.9876687526702881,
      "step": 652
    },
    {
      "epoch": 0.05347582634687631,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3945814298096842e-05,
      "loss": 0.7347517609596252,
      "step": 653
    },
    {
      "epoch": 0.053557718883395264,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3945576560921255e-05,
      "loss": 1.0370835065841675,
      "step": 654
    },
    {
      "epoch": 0.05363961141991422,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3945338304541714e-05,
      "loss": 1.0775206089019775,
      "step": 655
    },
    {
      "epoch": 0.05372150395643317,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3945099528968568e-05,
      "loss": 0.9059051871299744,
      "step": 656
    },
    {
      "epoch": 0.05380339649295213,
      "grad_norm": 0.171875,
      "learning_rate": 2.39448602342122e-05,
      "loss": 0.954623818397522,
      "step": 657
    },
    {
      "epoch": 0.053885289029471074,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3944620420283013e-05,
      "loss": 1.0121254920959473,
      "step": 658
    },
    {
      "epoch": 0.05396718156599003,
      "grad_norm": 0.169921875,
      "learning_rate": 2.394438008719142e-05,
      "loss": 0.888215184211731,
      "step": 659
    },
    {
      "epoch": 0.05404907410250898,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3944139234947878e-05,
      "loss": 0.7328816056251526,
      "step": 660
    },
    {
      "epoch": 0.05413096663902794,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3943897863562844e-05,
      "loss": 0.934559166431427,
      "step": 661
    },
    {
      "epoch": 0.05421285917554689,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.394365597304682e-05,
      "loss": 1.2699017524719238,
      "step": 662
    },
    {
      "epoch": 0.05429475171206584,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3943413563410315e-05,
      "loss": 0.9635473489761353,
      "step": 663
    },
    {
      "epoch": 0.05437664424858479,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3943170634663867e-05,
      "loss": 0.8488468527793884,
      "step": 664
    },
    {
      "epoch": 0.05445853678510375,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3942927186818027e-05,
      "loss": 0.7763572931289673,
      "step": 665
    },
    {
      "epoch": 0.0545404293216227,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3942683219883387e-05,
      "loss": 0.8804049491882324,
      "step": 666
    },
    {
      "epoch": 0.054622321858141656,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3942438733870545e-05,
      "loss": 1.2234079837799072,
      "step": 667
    },
    {
      "epoch": 0.054704214394660604,
      "grad_norm": 0.181640625,
      "learning_rate": 2.394219372879013e-05,
      "loss": 0.9823880195617676,
      "step": 668
    },
    {
      "epoch": 0.05478610693117956,
      "grad_norm": 0.171875,
      "learning_rate": 2.3941948204652786e-05,
      "loss": 0.5647228360176086,
      "step": 669
    },
    {
      "epoch": 0.05486799946769851,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.394170216146919e-05,
      "loss": 0.7954634428024292,
      "step": 670
    },
    {
      "epoch": 0.05494989200421747,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.394145559925003e-05,
      "loss": 0.5313792824745178,
      "step": 671
    },
    {
      "epoch": 0.05503178454073642,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.394120851800603e-05,
      "loss": 0.9468759298324585,
      "step": 672
    },
    {
      "epoch": 0.05511367707725537,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3940960917747924e-05,
      "loss": 0.6649528741836548,
      "step": 673
    },
    {
      "epoch": 0.05519556961377432,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3940712798486476e-05,
      "loss": 0.980419933795929,
      "step": 674
    },
    {
      "epoch": 0.05527746215029328,
      "grad_norm": 0.169921875,
      "learning_rate": 2.394046416023247e-05,
      "loss": 0.8429052233695984,
      "step": 675
    },
    {
      "epoch": 0.05535935468681223,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3940215002996714e-05,
      "loss": 0.770273745059967,
      "step": 676
    },
    {
      "epoch": 0.055441247223331186,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.3939965326790036e-05,
      "loss": 1.0326683521270752,
      "step": 677
    },
    {
      "epoch": 0.05552313975985013,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3939715131623284e-05,
      "loss": 0.9016773700714111,
      "step": 678
    },
    {
      "epoch": 0.05560503229636909,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3939464417507336e-05,
      "loss": 1.0948452949523926,
      "step": 679
    },
    {
      "epoch": 0.05568692483288804,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.3939213184453095e-05,
      "loss": 0.5703760385513306,
      "step": 680
    },
    {
      "epoch": 0.055768817369406996,
      "grad_norm": 0.14453125,
      "learning_rate": 2.393896143247147e-05,
      "loss": 0.8222994804382324,
      "step": 681
    },
    {
      "epoch": 0.05585070990592595,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.393870916157341e-05,
      "loss": 0.7562463879585266,
      "step": 682
    },
    {
      "epoch": 0.055932602442444905,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3938456371769877e-05,
      "loss": 1.0128675699234009,
      "step": 683
    },
    {
      "epoch": 0.05601449497896385,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.393820306307186e-05,
      "loss": 0.7738261222839355,
      "step": 684
    },
    {
      "epoch": 0.05609638751548281,
      "grad_norm": 0.263671875,
      "learning_rate": 2.393794923549037e-05,
      "loss": 1.0237443447113037,
      "step": 685
    },
    {
      "epoch": 0.05617828005200176,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3937694889036433e-05,
      "loss": 0.9787827134132385,
      "step": 686
    },
    {
      "epoch": 0.056260172588520715,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3937440023721112e-05,
      "loss": 0.935725748538971,
      "step": 687
    },
    {
      "epoch": 0.05634206512503967,
      "grad_norm": 0.2294921875,
      "learning_rate": 2.3937184639555477e-05,
      "loss": 1.0377476215362549,
      "step": 688
    },
    {
      "epoch": 0.05642395766155862,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3936928736550636e-05,
      "loss": 0.9346053600311279,
      "step": 689
    },
    {
      "epoch": 0.05650585019807757,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.3936672314717705e-05,
      "loss": 0.8664288520812988,
      "step": 690
    },
    {
      "epoch": 0.056587742734596526,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3936415374067833e-05,
      "loss": 0.8344204425811768,
      "step": 691
    },
    {
      "epoch": 0.05666963527111548,
      "grad_norm": 0.15625,
      "learning_rate": 2.3936157914612182e-05,
      "loss": 0.6864413022994995,
      "step": 692
    },
    {
      "epoch": 0.056751527807634435,
      "grad_norm": 0.20703125,
      "learning_rate": 2.3935899936361956e-05,
      "loss": 1.0343800783157349,
      "step": 693
    },
    {
      "epoch": 0.05683342034415338,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3935641439328353e-05,
      "loss": 1.025879144668579,
      "step": 694
    },
    {
      "epoch": 0.056915312880672336,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.393538242352261e-05,
      "loss": 0.9210976362228394,
      "step": 695
    },
    {
      "epoch": 0.05699720541719129,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3935122888955993e-05,
      "loss": 0.7017812132835388,
      "step": 696
    },
    {
      "epoch": 0.057079097953710245,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3934862835639778e-05,
      "loss": 1.1410526037216187,
      "step": 697
    },
    {
      "epoch": 0.0571609904902292,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3934602263585267e-05,
      "loss": 0.8802720904350281,
      "step": 698
    },
    {
      "epoch": 0.05724288302674815,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.393434117280379e-05,
      "loss": 0.6988977789878845,
      "step": 699
    },
    {
      "epoch": 0.0573247755632671,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3934079563306686e-05,
      "loss": 0.7477326989173889,
      "step": 700
    },
    {
      "epoch": 0.057406668099786055,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3933817435105335e-05,
      "loss": 0.8691118359565735,
      "step": 701
    },
    {
      "epoch": 0.05748856063630501,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3933554788211125e-05,
      "loss": 1.1438881158828735,
      "step": 702
    },
    {
      "epoch": 0.057570453172823964,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3933291622635473e-05,
      "loss": 0.8003643751144409,
      "step": 703
    },
    {
      "epoch": 0.05765234570934291,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3933027938389818e-05,
      "loss": 0.9975863695144653,
      "step": 704
    },
    {
      "epoch": 0.057734238245861866,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.393276373548562e-05,
      "loss": 0.8926045298576355,
      "step": 705
    },
    {
      "epoch": 0.05781613078238082,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3932499013934365e-05,
      "loss": 0.9769608378410339,
      "step": 706
    },
    {
      "epoch": 0.057898023318899774,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3932233773747556e-05,
      "loss": 0.958143413066864,
      "step": 707
    },
    {
      "epoch": 0.05797991585541873,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3931968014936718e-05,
      "loss": 1.0230159759521484,
      "step": 708
    },
    {
      "epoch": 0.05806180839193768,
      "grad_norm": 0.1640625,
      "learning_rate": 2.393170173751341e-05,
      "loss": 0.9981197118759155,
      "step": 709
    },
    {
      "epoch": 0.05814370092845663,
      "grad_norm": 0.140625,
      "learning_rate": 2.39314349414892e-05,
      "loss": 0.7228063344955444,
      "step": 710
    },
    {
      "epoch": 0.058225593464975585,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3931167626875683e-05,
      "loss": 0.966167688369751,
      "step": 711
    },
    {
      "epoch": 0.05830748600149454,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3930899793684483e-05,
      "loss": 0.706882894039154,
      "step": 712
    },
    {
      "epoch": 0.058389378538013494,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3930631441927235e-05,
      "loss": 0.6703528165817261,
      "step": 713
    },
    {
      "epoch": 0.05847127107453245,
      "grad_norm": 0.17578125,
      "learning_rate": 2.393036257161561e-05,
      "loss": 0.7549928426742554,
      "step": 714
    },
    {
      "epoch": 0.058553163611051395,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3930093182761288e-05,
      "loss": 0.7933093309402466,
      "step": 715
    },
    {
      "epoch": 0.05863505614757035,
      "grad_norm": 0.15625,
      "learning_rate": 2.3929823275375983e-05,
      "loss": 0.7844583988189697,
      "step": 716
    },
    {
      "epoch": 0.058716948684089304,
      "grad_norm": 0.1328125,
      "learning_rate": 2.392955284947142e-05,
      "loss": 0.3895062208175659,
      "step": 717
    },
    {
      "epoch": 0.05879884122060826,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3929281905059355e-05,
      "loss": 1.131768822669983,
      "step": 718
    },
    {
      "epoch": 0.05888073375712721,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.392901044215157e-05,
      "loss": 0.6950779557228088,
      "step": 719
    },
    {
      "epoch": 0.05896262629364616,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3928738460759852e-05,
      "loss": 0.7448866367340088,
      "step": 720
    },
    {
      "epoch": 0.059044518830165114,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3928465960896036e-05,
      "loss": 0.8207625150680542,
      "step": 721
    },
    {
      "epoch": 0.05912641136668407,
      "grad_norm": 0.181640625,
      "learning_rate": 2.392819294257196e-05,
      "loss": 0.8510642647743225,
      "step": 722
    },
    {
      "epoch": 0.05920830390320302,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3927919405799486e-05,
      "loss": 1.0432260036468506,
      "step": 723
    },
    {
      "epoch": 0.05929019643972198,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3927645350590505e-05,
      "loss": 1.0727624893188477,
      "step": 724
    },
    {
      "epoch": 0.059372088976240925,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3927370776956938e-05,
      "loss": 0.7621020674705505,
      "step": 725
    },
    {
      "epoch": 0.05945398151275988,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.3927095684910705e-05,
      "loss": 1.1959811449050903,
      "step": 726
    },
    {
      "epoch": 0.059535874049278834,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.392682007446377e-05,
      "loss": 0.7877959609031677,
      "step": 727
    },
    {
      "epoch": 0.05961776658579779,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3926543945628117e-05,
      "loss": 1.078640341758728,
      "step": 728
    },
    {
      "epoch": 0.05969965912231674,
      "grad_norm": 0.166015625,
      "learning_rate": 2.392626729841574e-05,
      "loss": 1.136202096939087,
      "step": 729
    },
    {
      "epoch": 0.05978155165883569,
      "grad_norm": 0.171875,
      "learning_rate": 2.3925990132838663e-05,
      "loss": 0.8831365704536438,
      "step": 730
    },
    {
      "epoch": 0.059863444195354644,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3925712448908936e-05,
      "loss": 0.8536401987075806,
      "step": 731
    },
    {
      "epoch": 0.0599453367318736,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3925434246638633e-05,
      "loss": 0.6902964115142822,
      "step": 732
    },
    {
      "epoch": 0.06002722926839255,
      "grad_norm": 0.203125,
      "learning_rate": 2.3925155526039833e-05,
      "loss": 0.9232505559921265,
      "step": 733
    },
    {
      "epoch": 0.06010912180491151,
      "grad_norm": 0.205078125,
      "learning_rate": 2.392487628712466e-05,
      "loss": 0.9984489679336548,
      "step": 734
    },
    {
      "epoch": 0.06019101434143046,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.392459652990525e-05,
      "loss": 1.1143248081207275,
      "step": 735
    },
    {
      "epoch": 0.06027290687794941,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.392431625439376e-05,
      "loss": 0.8914723992347717,
      "step": 736
    },
    {
      "epoch": 0.06035479941446836,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3924035460602374e-05,
      "loss": 0.7896877527236938,
      "step": 737
    },
    {
      "epoch": 0.06043669195098732,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3923754148543294e-05,
      "loss": 0.6035254597663879,
      "step": 738
    },
    {
      "epoch": 0.06051858448750627,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.392347231822875e-05,
      "loss": 0.8128514289855957,
      "step": 739
    },
    {
      "epoch": 0.060600477024025226,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3923189969670987e-05,
      "loss": 0.5372666120529175,
      "step": 740
    },
    {
      "epoch": 0.060682369560544173,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3922907102882282e-05,
      "loss": 0.9451065063476562,
      "step": 741
    },
    {
      "epoch": 0.06076426209706313,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3922623717874927e-05,
      "loss": 0.8032661080360413,
      "step": 742
    },
    {
      "epoch": 0.06084615463358208,
      "grad_norm": 0.197265625,
      "learning_rate": 2.392233981466124e-05,
      "loss": 0.7575138807296753,
      "step": 743
    },
    {
      "epoch": 0.060928047170101036,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3922055393253558e-05,
      "loss": 0.7567515969276428,
      "step": 744
    },
    {
      "epoch": 0.06100993970661999,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.392177045366425e-05,
      "loss": 0.5902807712554932,
      "step": 745
    },
    {
      "epoch": 0.06109183224313894,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3921484995905694e-05,
      "loss": 0.9041367769241333,
      "step": 746
    },
    {
      "epoch": 0.06117372477965789,
      "grad_norm": 0.23828125,
      "learning_rate": 2.3921199019990297e-05,
      "loss": 1.1415441036224365,
      "step": 747
    },
    {
      "epoch": 0.06125561731617685,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.392091252593049e-05,
      "loss": 0.8333556652069092,
      "step": 748
    },
    {
      "epoch": 0.0613375098526958,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.392062551373873e-05,
      "loss": 0.7662279605865479,
      "step": 749
    },
    {
      "epoch": 0.061419402389214756,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3920337983427485e-05,
      "loss": 0.8412419557571411,
      "step": 750
    },
    {
      "epoch": 0.0615012949257337,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3920049935009257e-05,
      "loss": 0.8625929355621338,
      "step": 751
    },
    {
      "epoch": 0.06158318746225266,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3919761368496562e-05,
      "loss": 0.8521045446395874,
      "step": 752
    },
    {
      "epoch": 0.06166507999877161,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.391947228390194e-05,
      "loss": 0.9968081712722778,
      "step": 753
    },
    {
      "epoch": 0.061746972535290566,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3919182681237968e-05,
      "loss": 0.8399487733840942,
      "step": 754
    },
    {
      "epoch": 0.06182886507180952,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.391889256051722e-05,
      "loss": 0.8944022059440613,
      "step": 755
    },
    {
      "epoch": 0.06191075760832847,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3918601921752313e-05,
      "loss": 1.0813689231872559,
      "step": 756
    },
    {
      "epoch": 0.06199265014484742,
      "grad_norm": 0.208984375,
      "learning_rate": 2.3918310764955877e-05,
      "loss": 0.6755024194717407,
      "step": 757
    },
    {
      "epoch": 0.062074542681366376,
      "grad_norm": 0.140625,
      "learning_rate": 2.3918019090140568e-05,
      "loss": 0.8223787546157837,
      "step": 758
    },
    {
      "epoch": 0.06215643521788533,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3917726897319058e-05,
      "loss": 1.0436381101608276,
      "step": 759
    },
    {
      "epoch": 0.062238327754404285,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.3917434186504057e-05,
      "loss": 1.1123673915863037,
      "step": 760
    },
    {
      "epoch": 0.06232022029092324,
      "grad_norm": 0.18359375,
      "learning_rate": 2.391714095770828e-05,
      "loss": 0.897057294845581,
      "step": 761
    },
    {
      "epoch": 0.06240211282744219,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3916847210944475e-05,
      "loss": 0.7659156918525696,
      "step": 762
    },
    {
      "epoch": 0.06248400536396114,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3916552946225407e-05,
      "loss": 0.9436648488044739,
      "step": 763
    },
    {
      "epoch": 0.06256589790048009,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3916258163563866e-05,
      "loss": 0.8617867231369019,
      "step": 764
    },
    {
      "epoch": 0.06264779043699904,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.391596286297267e-05,
      "loss": 0.7458943724632263,
      "step": 765
    },
    {
      "epoch": 0.062729682973518,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3915667044464647e-05,
      "loss": 0.8565216064453125,
      "step": 766
    },
    {
      "epoch": 0.06281157551003695,
      "grad_norm": 0.203125,
      "learning_rate": 2.3915370708052655e-05,
      "loss": 0.6604374647140503,
      "step": 767
    },
    {
      "epoch": 0.0628934680465559,
      "grad_norm": 0.21875,
      "learning_rate": 2.3915073853749577e-05,
      "loss": 1.0307912826538086,
      "step": 768
    },
    {
      "epoch": 0.06297536058307486,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3914776481568318e-05,
      "loss": 0.5363497734069824,
      "step": 769
    },
    {
      "epoch": 0.06305725311959381,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3914478591521795e-05,
      "loss": 0.8559870719909668,
      "step": 770
    },
    {
      "epoch": 0.06313914565611277,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3914180183622968e-05,
      "loss": 0.6360553503036499,
      "step": 771
    },
    {
      "epoch": 0.06322103819263172,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3913881257884794e-05,
      "loss": 0.5966918468475342,
      "step": 772
    },
    {
      "epoch": 0.06330293072915068,
      "grad_norm": 0.18359375,
      "learning_rate": 2.391358181432027e-05,
      "loss": 1.0226327180862427,
      "step": 773
    },
    {
      "epoch": 0.06338482326566962,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3913281852942417e-05,
      "loss": 0.8963218331336975,
      "step": 774
    },
    {
      "epoch": 0.06346671580218857,
      "grad_norm": 0.1953125,
      "learning_rate": 2.391298137376426e-05,
      "loss": 0.7732521891593933,
      "step": 775
    },
    {
      "epoch": 0.06354860833870753,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.391268037679887e-05,
      "loss": 0.5948144197463989,
      "step": 776
    },
    {
      "epoch": 0.06363050087522648,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3912378862059327e-05,
      "loss": 0.7968024015426636,
      "step": 777
    },
    {
      "epoch": 0.06371239341174544,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3912076829558738e-05,
      "loss": 1.0155894756317139,
      "step": 778
    },
    {
      "epoch": 0.06379428594826439,
      "grad_norm": 0.1484375,
      "learning_rate": 2.391177427931022e-05,
      "loss": 0.8076337575912476,
      "step": 779
    },
    {
      "epoch": 0.06387617848478334,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3911471211326938e-05,
      "loss": 0.7400011420249939,
      "step": 780
    },
    {
      "epoch": 0.0639580710213023,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3911167625622054e-05,
      "loss": 1.0085256099700928,
      "step": 781
    },
    {
      "epoch": 0.06403996355782125,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3910863522208765e-05,
      "loss": 0.800769567489624,
      "step": 782
    },
    {
      "epoch": 0.06412185609434021,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3910558901100293e-05,
      "loss": 0.7707474827766418,
      "step": 783
    },
    {
      "epoch": 0.06420374863085916,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3910253762309875e-05,
      "loss": 1.0052869319915771,
      "step": 784
    },
    {
      "epoch": 0.0642856411673781,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3909948105850774e-05,
      "loss": 0.8430008888244629,
      "step": 785
    },
    {
      "epoch": 0.06436753370389706,
      "grad_norm": 0.2373046875,
      "learning_rate": 2.3909641931736274e-05,
      "loss": 0.8366373181343079,
      "step": 786
    },
    {
      "epoch": 0.06444942624041601,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.390933523997968e-05,
      "loss": 1.0782530307769775,
      "step": 787
    },
    {
      "epoch": 0.06453131877693496,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.390902803059433e-05,
      "loss": 0.7264035940170288,
      "step": 788
    },
    {
      "epoch": 0.06461321131345392,
      "grad_norm": 0.17578125,
      "learning_rate": 2.390872030359357e-05,
      "loss": 1.003858208656311,
      "step": 789
    },
    {
      "epoch": 0.06469510384997287,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.390841205899078e-05,
      "loss": 1.1298493146896362,
      "step": 790
    },
    {
      "epoch": 0.06477699638649183,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3908103296799353e-05,
      "loss": 1.0049954652786255,
      "step": 791
    },
    {
      "epoch": 0.06485888892301078,
      "grad_norm": 0.205078125,
      "learning_rate": 2.390779401703271e-05,
      "loss": 0.8537121415138245,
      "step": 792
    },
    {
      "epoch": 0.06494078145952974,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3907484219704294e-05,
      "loss": 0.6126849055290222,
      "step": 793
    },
    {
      "epoch": 0.06502267399604869,
      "grad_norm": 0.19921875,
      "learning_rate": 2.390717390482757e-05,
      "loss": 1.0838035345077515,
      "step": 794
    },
    {
      "epoch": 0.06510456653256763,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.390686307241603e-05,
      "loss": 0.6581450700759888,
      "step": 795
    },
    {
      "epoch": 0.06518645906908659,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.390655172248318e-05,
      "loss": 0.7470253705978394,
      "step": 796
    },
    {
      "epoch": 0.06526835160560554,
      "grad_norm": 0.1640625,
      "learning_rate": 2.390623985504255e-05,
      "loss": 0.6015428900718689,
      "step": 797
    },
    {
      "epoch": 0.0653502441421245,
      "grad_norm": 0.1640625,
      "learning_rate": 2.39059274701077e-05,
      "loss": 0.6400900483131409,
      "step": 798
    },
    {
      "epoch": 0.06543213667864345,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.39056145676922e-05,
      "loss": 0.8458360433578491,
      "step": 799
    },
    {
      "epoch": 0.0655140292151624,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.390530114780966e-05,
      "loss": 0.885083019733429,
      "step": 800
    },
    {
      "epoch": 0.06559592175168136,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3904987210473695e-05,
      "loss": 0.6624044179916382,
      "step": 801
    },
    {
      "epoch": 0.06567781428820031,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3904672755697956e-05,
      "loss": 0.7437598705291748,
      "step": 802
    },
    {
      "epoch": 0.06575970682471927,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3904357783496105e-05,
      "loss": 0.9272804260253906,
      "step": 803
    },
    {
      "epoch": 0.06584159936123822,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3904042293881833e-05,
      "loss": 0.8374510407447815,
      "step": 804
    },
    {
      "epoch": 0.06592349189775717,
      "grad_norm": 0.1328125,
      "learning_rate": 2.3903726286868857e-05,
      "loss": 0.8609854578971863,
      "step": 805
    },
    {
      "epoch": 0.06600538443427612,
      "grad_norm": 0.12353515625,
      "learning_rate": 2.3903409762470907e-05,
      "loss": 0.6509743928909302,
      "step": 806
    },
    {
      "epoch": 0.06608727697079507,
      "grad_norm": 0.19140625,
      "learning_rate": 2.390309272070174e-05,
      "loss": 0.9537597298622131,
      "step": 807
    },
    {
      "epoch": 0.06616916950731402,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3902775161575142e-05,
      "loss": 0.6923487782478333,
      "step": 808
    },
    {
      "epoch": 0.06625106204383298,
      "grad_norm": 0.19140625,
      "learning_rate": 2.390245708510491e-05,
      "loss": 0.62888503074646,
      "step": 809
    },
    {
      "epoch": 0.06633295458035193,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.390213849130487e-05,
      "loss": 0.9218167662620544,
      "step": 810
    },
    {
      "epoch": 0.06641484711687089,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3901819380188872e-05,
      "loss": 0.8725098371505737,
      "step": 811
    },
    {
      "epoch": 0.06649673965338984,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3901499751770782e-05,
      "loss": 0.9962688684463501,
      "step": 812
    },
    {
      "epoch": 0.0665786321899088,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.390117960606449e-05,
      "loss": 0.6602803468704224,
      "step": 813
    },
    {
      "epoch": 0.06666052472642775,
      "grad_norm": 0.21875,
      "learning_rate": 2.3900858943083924e-05,
      "loss": 1.0366820096969604,
      "step": 814
    },
    {
      "epoch": 0.0667424172629467,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3900537762843006e-05,
      "loss": 0.977749228477478,
      "step": 815
    },
    {
      "epoch": 0.06682430979946564,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3900216065355703e-05,
      "loss": 0.6417954564094543,
      "step": 816
    },
    {
      "epoch": 0.0669062023359846,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3899893850635997e-05,
      "loss": 1.2211062908172607,
      "step": 817
    },
    {
      "epoch": 0.06698809487250355,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3899571118697893e-05,
      "loss": 0.6863321661949158,
      "step": 818
    },
    {
      "epoch": 0.06706998740902251,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3899247869555416e-05,
      "loss": 0.9974377155303955,
      "step": 819
    },
    {
      "epoch": 0.06715187994554146,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3898924103222618e-05,
      "loss": 0.7611660361289978,
      "step": 820
    },
    {
      "epoch": 0.06723377248206042,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3898599819713572e-05,
      "loss": 0.8447750806808472,
      "step": 821
    },
    {
      "epoch": 0.06731566501857937,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.389827501904237e-05,
      "loss": 0.9318820834159851,
      "step": 822
    },
    {
      "epoch": 0.06739755755509833,
      "grad_norm": 0.134765625,
      "learning_rate": 2.3897949701223128e-05,
      "loss": 0.5046913623809814,
      "step": 823
    },
    {
      "epoch": 0.06747945009161728,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.389762386626999e-05,
      "loss": 0.8080394864082336,
      "step": 824
    },
    {
      "epoch": 0.06756134262813623,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3897297514197115e-05,
      "loss": 0.9606471061706543,
      "step": 825
    },
    {
      "epoch": 0.06764323516465517,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.389697064501869e-05,
      "loss": 0.6995418071746826,
      "step": 826
    },
    {
      "epoch": 0.06772512770117413,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.389664325874892e-05,
      "loss": 1.1075576543807983,
      "step": 827
    },
    {
      "epoch": 0.06780702023769308,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3896315355402032e-05,
      "loss": 0.8601889610290527,
      "step": 828
    },
    {
      "epoch": 0.06788891277421204,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3895986934992285e-05,
      "loss": 0.7834513187408447,
      "step": 829
    },
    {
      "epoch": 0.06797080531073099,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3895657997533952e-05,
      "loss": 0.8533635139465332,
      "step": 830
    },
    {
      "epoch": 0.06805269784724995,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3895328543041325e-05,
      "loss": 0.8712145090103149,
      "step": 831
    },
    {
      "epoch": 0.0681345903837689,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3894998571528727e-05,
      "loss": 1.0655306577682495,
      "step": 832
    },
    {
      "epoch": 0.06821648292028785,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.38946680830105e-05,
      "loss": 1.0326701402664185,
      "step": 833
    },
    {
      "epoch": 0.06829837545680681,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3894337077501005e-05,
      "loss": 1.2634947299957275,
      "step": 834
    },
    {
      "epoch": 0.06838026799332576,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3894005555014634e-05,
      "loss": 0.8876171112060547,
      "step": 835
    },
    {
      "epoch": 0.06846216052984472,
      "grad_norm": 0.1328125,
      "learning_rate": 2.389367351556579e-05,
      "loss": 0.7812813520431519,
      "step": 836
    },
    {
      "epoch": 0.06854405306636366,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.389334095916891e-05,
      "loss": 0.7355129718780518,
      "step": 837
    },
    {
      "epoch": 0.06862594560288261,
      "grad_norm": 0.146484375,
      "learning_rate": 2.389300788583845e-05,
      "loss": 0.6984882950782776,
      "step": 838
    },
    {
      "epoch": 0.06870783813940157,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3892674295588882e-05,
      "loss": 0.7929954528808594,
      "step": 839
    },
    {
      "epoch": 0.06878973067592052,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3892340188434705e-05,
      "loss": 0.8127511739730835,
      "step": 840
    },
    {
      "epoch": 0.06887162321243948,
      "grad_norm": 0.228515625,
      "learning_rate": 2.3892005564390444e-05,
      "loss": 0.9786326885223389,
      "step": 841
    },
    {
      "epoch": 0.06895351574895843,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3891670423470644e-05,
      "loss": 0.39185306429862976,
      "step": 842
    },
    {
      "epoch": 0.06903540828547738,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3891334765689862e-05,
      "loss": 0.89034503698349,
      "step": 843
    },
    {
      "epoch": 0.06911730082199634,
      "grad_norm": 0.2373046875,
      "learning_rate": 2.3890998591062704e-05,
      "loss": 1.085848331451416,
      "step": 844
    },
    {
      "epoch": 0.06919919335851529,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3890661899603764e-05,
      "loss": 0.8302363157272339,
      "step": 845
    },
    {
      "epoch": 0.06928108589503425,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3890324691327685e-05,
      "loss": 1.2554987668991089,
      "step": 846
    },
    {
      "epoch": 0.06936297843155319,
      "grad_norm": 0.140625,
      "learning_rate": 2.3889986966249126e-05,
      "loss": 0.6674655079841614,
      "step": 847
    },
    {
      "epoch": 0.06944487096807214,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.388964872438276e-05,
      "loss": 0.6139657497406006,
      "step": 848
    },
    {
      "epoch": 0.0695267635045911,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.388930996574329e-05,
      "loss": 0.7831220626831055,
      "step": 849
    },
    {
      "epoch": 0.06960865604111005,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3888970690345437e-05,
      "loss": 0.8593570590019226,
      "step": 850
    },
    {
      "epoch": 0.069690548577629,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3888630898203958e-05,
      "loss": 0.6840532422065735,
      "step": 851
    },
    {
      "epoch": 0.06977244111414796,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.388829058933361e-05,
      "loss": 0.8797383308410645,
      "step": 852
    },
    {
      "epoch": 0.06985433365066691,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.388794976374919e-05,
      "loss": 1.1468734741210938,
      "step": 853
    },
    {
      "epoch": 0.06993622618718587,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.388760842146551e-05,
      "loss": 0.8519679307937622,
      "step": 854
    },
    {
      "epoch": 0.07001811872370482,
      "grad_norm": 0.171875,
      "learning_rate": 2.3887266562497407e-05,
      "loss": 0.9134557843208313,
      "step": 855
    },
    {
      "epoch": 0.07010001126022378,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.388692418685974e-05,
      "loss": 0.7357765436172485,
      "step": 856
    },
    {
      "epoch": 0.07018190379674273,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3886581294567387e-05,
      "loss": 1.0959980487823486,
      "step": 857
    },
    {
      "epoch": 0.07026379633326167,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3886237885635254e-05,
      "loss": 0.9201380014419556,
      "step": 858
    },
    {
      "epoch": 0.07034568886978063,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.3885893960078268e-05,
      "loss": 0.7882974147796631,
      "step": 859
    },
    {
      "epoch": 0.07042758140629958,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3885549517911375e-05,
      "loss": 0.9611315131187439,
      "step": 860
    },
    {
      "epoch": 0.07050947394281853,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.388520455914955e-05,
      "loss": 0.9144834876060486,
      "step": 861
    },
    {
      "epoch": 0.07059136647933749,
      "grad_norm": 0.17578125,
      "learning_rate": 2.388485908380778e-05,
      "loss": 0.5752807855606079,
      "step": 862
    },
    {
      "epoch": 0.07067325901585644,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3884513091901083e-05,
      "loss": 0.6473537683486938,
      "step": 863
    },
    {
      "epoch": 0.0707551515523754,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.38841665834445e-05,
      "loss": 0.733709990978241,
      "step": 864
    },
    {
      "epoch": 0.07083704408889435,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.388381955845309e-05,
      "loss": 0.9901431798934937,
      "step": 865
    },
    {
      "epoch": 0.0709189366254133,
      "grad_norm": 0.171875,
      "learning_rate": 2.3883472016941936e-05,
      "loss": 0.6014519333839417,
      "step": 866
    },
    {
      "epoch": 0.07100082916193226,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3883123958926143e-05,
      "loss": 0.9459564685821533,
      "step": 867
    },
    {
      "epoch": 0.0710827216984512,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.388277538442084e-05,
      "loss": 0.7435308694839478,
      "step": 868
    },
    {
      "epoch": 0.07116461423497016,
      "grad_norm": 0.224609375,
      "learning_rate": 2.3882426293441183e-05,
      "loss": 1.2223165035247803,
      "step": 869
    },
    {
      "epoch": 0.07124650677148911,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3882076686002332e-05,
      "loss": 0.7333842515945435,
      "step": 870
    },
    {
      "epoch": 0.07132839930800806,
      "grad_norm": 0.158203125,
      "learning_rate": 2.388172656211949e-05,
      "loss": 0.7209481000900269,
      "step": 871
    },
    {
      "epoch": 0.07141029184452702,
      "grad_norm": 0.1796875,
      "learning_rate": 2.388137592180788e-05,
      "loss": 0.8283629417419434,
      "step": 872
    },
    {
      "epoch": 0.07149218438104597,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.388102476508273e-05,
      "loss": 0.9210089445114136,
      "step": 873
    },
    {
      "epoch": 0.07157407691756493,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3880673091959314e-05,
      "loss": 0.823883593082428,
      "step": 874
    },
    {
      "epoch": 0.07165596945408388,
      "grad_norm": 0.40625,
      "learning_rate": 2.3880320902452907e-05,
      "loss": 0.8558413982391357,
      "step": 875
    },
    {
      "epoch": 0.07173786199060284,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3879968196578828e-05,
      "loss": 0.925654411315918,
      "step": 876
    },
    {
      "epoch": 0.07181975452712179,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.38796149743524e-05,
      "loss": 1.2575734853744507,
      "step": 877
    },
    {
      "epoch": 0.07190164706364073,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3879261235788975e-05,
      "loss": 1.1645255088806152,
      "step": 878
    },
    {
      "epoch": 0.07198353960015969,
      "grad_norm": 0.19921875,
      "learning_rate": 2.387890698090393e-05,
      "loss": 0.667198121547699,
      "step": 879
    },
    {
      "epoch": 0.07206543213667864,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.387855220971266e-05,
      "loss": 0.9587016105651855,
      "step": 880
    },
    {
      "epoch": 0.0721473246731976,
      "grad_norm": 0.166015625,
      "learning_rate": 2.387819692223059e-05,
      "loss": 0.8451201915740967,
      "step": 881
    },
    {
      "epoch": 0.07222921720971655,
      "grad_norm": 0.17578125,
      "learning_rate": 2.387784111847316e-05,
      "loss": 0.7909094095230103,
      "step": 882
    },
    {
      "epoch": 0.0723111097462355,
      "grad_norm": 0.15625,
      "learning_rate": 2.387748479845583e-05,
      "loss": 0.6922479271888733,
      "step": 883
    },
    {
      "epoch": 0.07239300228275446,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3877127962194093e-05,
      "loss": 0.7669557929039001,
      "step": 884
    },
    {
      "epoch": 0.07247489481927341,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3876770609703456e-05,
      "loss": 0.8714954853057861,
      "step": 885
    },
    {
      "epoch": 0.07255678735579237,
      "grad_norm": 0.216796875,
      "learning_rate": 2.3876412740999452e-05,
      "loss": 0.9960775375366211,
      "step": 886
    },
    {
      "epoch": 0.07263867989231132,
      "grad_norm": 0.20703125,
      "learning_rate": 2.3876054356097632e-05,
      "loss": 0.9100107550621033,
      "step": 887
    },
    {
      "epoch": 0.07272057242883027,
      "grad_norm": 0.185546875,
      "learning_rate": 2.387569545501358e-05,
      "loss": 1.038118839263916,
      "step": 888
    },
    {
      "epoch": 0.07280246496534921,
      "grad_norm": 0.150390625,
      "learning_rate": 2.387533603776289e-05,
      "loss": 0.5999362468719482,
      "step": 889
    },
    {
      "epoch": 0.07288435750186817,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3874976104361187e-05,
      "loss": 0.780677318572998,
      "step": 890
    },
    {
      "epoch": 0.07296625003838712,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.387461565482411e-05,
      "loss": 0.7013919353485107,
      "step": 891
    },
    {
      "epoch": 0.07304814257490608,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3874254689167324e-05,
      "loss": 0.9123063683509827,
      "step": 892
    },
    {
      "epoch": 0.07313003511142503,
      "grad_norm": 0.16015625,
      "learning_rate": 2.387389320740653e-05,
      "loss": 1.3979144096374512,
      "step": 893
    },
    {
      "epoch": 0.07321192764794399,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.387353120955743e-05,
      "loss": 0.8316360712051392,
      "step": 894
    },
    {
      "epoch": 0.07329382018446294,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.387316869563576e-05,
      "loss": 0.9191043972969055,
      "step": 895
    },
    {
      "epoch": 0.0733757127209819,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3872805665657274e-05,
      "loss": 0.7610026001930237,
      "step": 896
    },
    {
      "epoch": 0.07345760525750085,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3872442119637756e-05,
      "loss": 0.5807282328605652,
      "step": 897
    },
    {
      "epoch": 0.0735394977940198,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3872078057593007e-05,
      "loss": 1.0620054006576538,
      "step": 898
    },
    {
      "epoch": 0.07362139033053874,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3871713479538846e-05,
      "loss": 0.9009345173835754,
      "step": 899
    },
    {
      "epoch": 0.0737032828670577,
      "grad_norm": 0.294921875,
      "learning_rate": 2.387134838549112e-05,
      "loss": 0.8334758877754211,
      "step": 900
    },
    {
      "epoch": 0.07378517540357665,
      "grad_norm": 0.185546875,
      "learning_rate": 2.38709827754657e-05,
      "loss": 0.8208713531494141,
      "step": 901
    },
    {
      "epoch": 0.07386706794009561,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3870616649478472e-05,
      "loss": 1.0119149684906006,
      "step": 902
    },
    {
      "epoch": 0.07394896047661456,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.387025000754536e-05,
      "loss": 1.1852924823760986,
      "step": 903
    },
    {
      "epoch": 0.07403085301313352,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3869882849682287e-05,
      "loss": 0.9489538073539734,
      "step": 904
    },
    {
      "epoch": 0.07411274554965247,
      "grad_norm": 0.16015625,
      "learning_rate": 2.386951517590522e-05,
      "loss": 0.8638594150543213,
      "step": 905
    },
    {
      "epoch": 0.07419463808617142,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3869146986230134e-05,
      "loss": 1.0569462776184082,
      "step": 906
    },
    {
      "epoch": 0.07427653062269038,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3868778280673035e-05,
      "loss": 0.7985172271728516,
      "step": 907
    },
    {
      "epoch": 0.07435842315920933,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.386840905924995e-05,
      "loss": 0.7662058472633362,
      "step": 908
    },
    {
      "epoch": 0.07444031569572829,
      "grad_norm": 0.177734375,
      "learning_rate": 2.386803932197692e-05,
      "loss": 0.875137209892273,
      "step": 909
    },
    {
      "epoch": 0.07452220823224723,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3867669068870027e-05,
      "loss": 0.6429243683815002,
      "step": 910
    },
    {
      "epoch": 0.07460410076876618,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3867298299945353e-05,
      "loss": 0.6278123259544373,
      "step": 911
    },
    {
      "epoch": 0.07468599330528514,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3866927015219018e-05,
      "loss": 0.5837377905845642,
      "step": 912
    },
    {
      "epoch": 0.07476788584180409,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3866555214707158e-05,
      "loss": 0.752549946308136,
      "step": 913
    },
    {
      "epoch": 0.07484977837832305,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3866182898425935e-05,
      "loss": 0.6466978192329407,
      "step": 914
    },
    {
      "epoch": 0.074931670914842,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.386581006639153e-05,
      "loss": 1.0016194581985474,
      "step": 915
    },
    {
      "epoch": 0.07501356345136095,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3865436718620146e-05,
      "loss": 0.9586094617843628,
      "step": 916
    },
    {
      "epoch": 0.07509545598787991,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3865062855128013e-05,
      "loss": 0.7151205539703369,
      "step": 917
    },
    {
      "epoch": 0.07517734852439886,
      "grad_norm": 0.25390625,
      "learning_rate": 2.386468847593138e-05,
      "loss": 0.8532623052597046,
      "step": 918
    },
    {
      "epoch": 0.07525924106091782,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3864313581046516e-05,
      "loss": 1.0382089614868164,
      "step": 919
    },
    {
      "epoch": 0.07534113359743676,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.386393817048972e-05,
      "loss": 0.9049426913261414,
      "step": 920
    },
    {
      "epoch": 0.07542302613395571,
      "grad_norm": 0.19140625,
      "learning_rate": 2.386356224427731e-05,
      "loss": 0.6633367538452148,
      "step": 921
    },
    {
      "epoch": 0.07550491867047467,
      "grad_norm": 0.185546875,
      "learning_rate": 2.386318580242562e-05,
      "loss": 0.6424201726913452,
      "step": 922
    },
    {
      "epoch": 0.07558681120699362,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3862808844951015e-05,
      "loss": 0.681705117225647,
      "step": 923
    },
    {
      "epoch": 0.07566870374351257,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.3862431371869875e-05,
      "loss": 1.1234625577926636,
      "step": 924
    },
    {
      "epoch": 0.07575059628003153,
      "grad_norm": 0.21875,
      "learning_rate": 2.3862053383198615e-05,
      "loss": 1.153337836265564,
      "step": 925
    },
    {
      "epoch": 0.07583248881655048,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3861674878953656e-05,
      "loss": 0.7771589756011963,
      "step": 926
    },
    {
      "epoch": 0.07591438135306944,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3861295859151452e-05,
      "loss": 0.824066162109375,
      "step": 927
    },
    {
      "epoch": 0.07599627388958839,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3860916323808476e-05,
      "loss": 0.779680073261261,
      "step": 928
    },
    {
      "epoch": 0.07607816642610735,
      "grad_norm": 0.15625,
      "learning_rate": 2.3860536272941227e-05,
      "loss": 0.7602167129516602,
      "step": 929
    },
    {
      "epoch": 0.07616005896262629,
      "grad_norm": 0.228515625,
      "learning_rate": 2.386015570656622e-05,
      "loss": 1.1875331401824951,
      "step": 930
    },
    {
      "epoch": 0.07624195149914524,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.38597746247e-05,
      "loss": 1.1889960765838623,
      "step": 931
    },
    {
      "epoch": 0.0763238440356642,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3859393027359125e-05,
      "loss": 0.7926914095878601,
      "step": 932
    },
    {
      "epoch": 0.07640573657218315,
      "grad_norm": 0.220703125,
      "learning_rate": 2.3859010914560186e-05,
      "loss": 0.5850602984428406,
      "step": 933
    },
    {
      "epoch": 0.0764876291087021,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3858628286319788e-05,
      "loss": 0.9435917735099792,
      "step": 934
    },
    {
      "epoch": 0.07656952164522106,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3858245142654563e-05,
      "loss": 0.6421206593513489,
      "step": 935
    },
    {
      "epoch": 0.07665141418174001,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3857861483581163e-05,
      "loss": 0.9105852246284485,
      "step": 936
    },
    {
      "epoch": 0.07673330671825897,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3857477309116265e-05,
      "loss": 1.106433391571045,
      "step": 937
    },
    {
      "epoch": 0.07681519925477792,
      "grad_norm": 0.20703125,
      "learning_rate": 2.3857092619276566e-05,
      "loss": 1.0934832096099854,
      "step": 938
    },
    {
      "epoch": 0.07689709179129688,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3856707414078782e-05,
      "loss": 1.102327823638916,
      "step": 939
    },
    {
      "epoch": 0.07697898432781583,
      "grad_norm": 0.13671875,
      "learning_rate": 2.3856321693539662e-05,
      "loss": 0.7117345929145813,
      "step": 940
    },
    {
      "epoch": 0.07706087686433477,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.385593545767597e-05,
      "loss": 1.048274040222168,
      "step": 941
    },
    {
      "epoch": 0.07714276940085373,
      "grad_norm": 0.2314453125,
      "learning_rate": 2.3855548706504493e-05,
      "loss": 0.7626367211341858,
      "step": 942
    },
    {
      "epoch": 0.07722466193737268,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3855161440042037e-05,
      "loss": 0.9599903225898743,
      "step": 943
    },
    {
      "epoch": 0.07730655447389163,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3854773658305438e-05,
      "loss": 0.9171708822250366,
      "step": 944
    },
    {
      "epoch": 0.07738844701041059,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.385438536131155e-05,
      "loss": 0.8478205800056458,
      "step": 945
    },
    {
      "epoch": 0.07747033954692954,
      "grad_norm": 0.208984375,
      "learning_rate": 2.385399654907725e-05,
      "loss": 1.2198203802108765,
      "step": 946
    },
    {
      "epoch": 0.0775522320834485,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3853607221619436e-05,
      "loss": 0.9292618632316589,
      "step": 947
    },
    {
      "epoch": 0.07763412461996745,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.385321737895503e-05,
      "loss": 0.9733217358589172,
      "step": 948
    },
    {
      "epoch": 0.0777160171564864,
      "grad_norm": 0.169921875,
      "learning_rate": 2.385282702110098e-05,
      "loss": 0.9776267409324646,
      "step": 949
    },
    {
      "epoch": 0.07779790969300536,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3852436148074245e-05,
      "loss": 0.7498472332954407,
      "step": 950
    },
    {
      "epoch": 0.0778798022295243,
      "grad_norm": 0.18359375,
      "learning_rate": 2.385204475989182e-05,
      "loss": 0.9233551621437073,
      "step": 951
    },
    {
      "epoch": 0.07796169476604325,
      "grad_norm": 0.287109375,
      "learning_rate": 2.3851652856570717e-05,
      "loss": 0.9215163588523865,
      "step": 952
    },
    {
      "epoch": 0.07804358730256221,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3851260438127965e-05,
      "loss": 0.8095278739929199,
      "step": 953
    },
    {
      "epoch": 0.07812547983908116,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3850867504580622e-05,
      "loss": 0.755857527256012,
      "step": 954
    },
    {
      "epoch": 0.07820737237560012,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3850474055945768e-05,
      "loss": 0.8303584456443787,
      "step": 955
    },
    {
      "epoch": 0.07828926491211907,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3850080092240505e-05,
      "loss": 0.8198450207710266,
      "step": 956
    },
    {
      "epoch": 0.07837115744863803,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3849685613481954e-05,
      "loss": 0.9797332882881165,
      "step": 957
    },
    {
      "epoch": 0.07845304998515698,
      "grad_norm": 0.234375,
      "learning_rate": 2.3849290619687256e-05,
      "loss": 0.7985044717788696,
      "step": 958
    },
    {
      "epoch": 0.07853494252167594,
      "grad_norm": 0.20703125,
      "learning_rate": 2.3848895110873588e-05,
      "loss": 0.9533881545066833,
      "step": 959
    },
    {
      "epoch": 0.07861683505819489,
      "grad_norm": 0.1875,
      "learning_rate": 2.384849908705814e-05,
      "loss": 0.9056169390678406,
      "step": 960
    },
    {
      "epoch": 0.07869872759471383,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3848102548258116e-05,
      "loss": 0.8116352558135986,
      "step": 961
    },
    {
      "epoch": 0.07878062013123278,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.384770549449076e-05,
      "loss": 0.7232645750045776,
      "step": 962
    },
    {
      "epoch": 0.07886251266775174,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3847307925773323e-05,
      "loss": 0.8241935968399048,
      "step": 963
    },
    {
      "epoch": 0.07894440520427069,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3846909842123085e-05,
      "loss": 0.7238634824752808,
      "step": 964
    },
    {
      "epoch": 0.07902629774078965,
      "grad_norm": 0.19921875,
      "learning_rate": 2.384651124355736e-05,
      "loss": 0.8749205470085144,
      "step": 965
    },
    {
      "epoch": 0.0791081902773086,
      "grad_norm": 0.150390625,
      "learning_rate": 2.384611213009346e-05,
      "loss": 0.6803001165390015,
      "step": 966
    },
    {
      "epoch": 0.07919008281382756,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3845712501748735e-05,
      "loss": 0.7880287766456604,
      "step": 967
    },
    {
      "epoch": 0.07927197535034651,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3845312358540556e-05,
      "loss": 0.7550244927406311,
      "step": 968
    },
    {
      "epoch": 0.07935386788686546,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3844911700486316e-05,
      "loss": 0.7586873173713684,
      "step": 969
    },
    {
      "epoch": 0.07943576042338442,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3844510527603425e-05,
      "loss": 0.6517073512077332,
      "step": 970
    },
    {
      "epoch": 0.07951765295990337,
      "grad_norm": 0.205078125,
      "learning_rate": 2.384410883990933e-05,
      "loss": 0.7106707096099854,
      "step": 971
    },
    {
      "epoch": 0.07959954549642231,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3843706637421475e-05,
      "loss": 0.7612528800964355,
      "step": 972
    },
    {
      "epoch": 0.07968143803294127,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.384330392015735e-05,
      "loss": 0.9213825464248657,
      "step": 973
    },
    {
      "epoch": 0.07976333056946022,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3842900688134457e-05,
      "loss": 0.6322633028030396,
      "step": 974
    },
    {
      "epoch": 0.07984522310597918,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3842496941370327e-05,
      "loss": 0.8157610893249512,
      "step": 975
    },
    {
      "epoch": 0.07992711564249813,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3842092679882503e-05,
      "loss": 0.8752837777137756,
      "step": 976
    },
    {
      "epoch": 0.08000900817901709,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3841687903688555e-05,
      "loss": 0.6833584308624268,
      "step": 977
    },
    {
      "epoch": 0.08009090071553604,
      "grad_norm": 0.18359375,
      "learning_rate": 2.384128261280608e-05,
      "loss": 0.7934967279434204,
      "step": 978
    },
    {
      "epoch": 0.080172793252055,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.384087680725269e-05,
      "loss": 0.8384382128715515,
      "step": 979
    },
    {
      "epoch": 0.08025468578857395,
      "grad_norm": 0.15234375,
      "learning_rate": 2.384047048704603e-05,
      "loss": 0.916492223739624,
      "step": 980
    },
    {
      "epoch": 0.0803365783250929,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.384006365220375e-05,
      "loss": 1.0703872442245483,
      "step": 981
    },
    {
      "epoch": 0.08041847086161184,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.383965630274354e-05,
      "loss": 1.0057543516159058,
      "step": 982
    },
    {
      "epoch": 0.0805003633981308,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3839248438683102e-05,
      "loss": 0.7645152807235718,
      "step": 983
    },
    {
      "epoch": 0.08058225593464975,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3838840060040164e-05,
      "loss": 0.6167287826538086,
      "step": 984
    },
    {
      "epoch": 0.0806641484711687,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3838431166832477e-05,
      "loss": 0.6900557279586792,
      "step": 985
    },
    {
      "epoch": 0.08074604100768766,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3838021759077814e-05,
      "loss": 0.6050208806991577,
      "step": 986
    },
    {
      "epoch": 0.08082793354420661,
      "grad_norm": 0.185546875,
      "learning_rate": 2.383761183679397e-05,
      "loss": 0.8370410799980164,
      "step": 987
    },
    {
      "epoch": 0.08090982608072557,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.383720139999876e-05,
      "loss": 0.9643905758857727,
      "step": 988
    },
    {
      "epoch": 0.08099171861724452,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3836790448710016e-05,
      "loss": 0.9964718222618103,
      "step": 989
    },
    {
      "epoch": 0.08107361115376348,
      "grad_norm": 0.13671875,
      "learning_rate": 2.3836378982945615e-05,
      "loss": 0.891491711139679,
      "step": 990
    },
    {
      "epoch": 0.08115550369028243,
      "grad_norm": 0.1875,
      "learning_rate": 2.3835967002723428e-05,
      "loss": 1.0418084859848022,
      "step": 991
    },
    {
      "epoch": 0.08123739622680139,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3835554508061367e-05,
      "loss": 0.9015802145004272,
      "step": 992
    },
    {
      "epoch": 0.08131928876332033,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3835141498977362e-05,
      "loss": 1.27971613407135,
      "step": 993
    },
    {
      "epoch": 0.08140118129983928,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.383472797548936e-05,
      "loss": 0.9369308948516846,
      "step": 994
    },
    {
      "epoch": 0.08148307383635824,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3834313937615337e-05,
      "loss": 0.7776637673377991,
      "step": 995
    },
    {
      "epoch": 0.08156496637287719,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.383389938537329e-05,
      "loss": 1.026078224182129,
      "step": 996
    },
    {
      "epoch": 0.08164685890939614,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3833484318781234e-05,
      "loss": 1.4209386110305786,
      "step": 997
    },
    {
      "epoch": 0.0817287514459151,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3833068737857214e-05,
      "loss": 0.8527098894119263,
      "step": 998
    },
    {
      "epoch": 0.08181064398243405,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3832652642619284e-05,
      "loss": 0.8481400012969971,
      "step": 999
    },
    {
      "epoch": 0.08189253651895301,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3832236033085538e-05,
      "loss": 1.1074132919311523,
      "step": 1000
    },
    {
      "epoch": 0.08197442905547196,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3831818909274084e-05,
      "loss": 0.5550487637519836,
      "step": 1001
    },
    {
      "epoch": 0.08205632159199092,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3831401271203043e-05,
      "loss": 1.1143569946289062,
      "step": 1002
    },
    {
      "epoch": 0.08213821412850986,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3830983118890574e-05,
      "loss": 0.7879875898361206,
      "step": 1003
    },
    {
      "epoch": 0.08222010666502881,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3830564452354852e-05,
      "loss": 0.770058810710907,
      "step": 1004
    },
    {
      "epoch": 0.08230199920154777,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.383014527161407e-05,
      "loss": 1.0728634595870972,
      "step": 1005
    },
    {
      "epoch": 0.08238389173806672,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3829725576686456e-05,
      "loss": 0.9467970132827759,
      "step": 1006
    },
    {
      "epoch": 0.08246578427458567,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3829305367590242e-05,
      "loss": 1.0244967937469482,
      "step": 1007
    },
    {
      "epoch": 0.08254767681110463,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3828884644343693e-05,
      "loss": 0.8292291164398193,
      "step": 1008
    },
    {
      "epoch": 0.08262956934762358,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3828463406965097e-05,
      "loss": 0.9327377080917358,
      "step": 1009
    },
    {
      "epoch": 0.08271146188414254,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3828041655472765e-05,
      "loss": 0.5960700511932373,
      "step": 1010
    },
    {
      "epoch": 0.08279335442066149,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3827619389885028e-05,
      "loss": 0.7657983899116516,
      "step": 1011
    },
    {
      "epoch": 0.08287524695718045,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.382719661022024e-05,
      "loss": 0.9336369633674622,
      "step": 1012
    },
    {
      "epoch": 0.08295713949369939,
      "grad_norm": 0.166015625,
      "learning_rate": 2.382677331649677e-05,
      "loss": 0.8051844835281372,
      "step": 1013
    },
    {
      "epoch": 0.08303903203021834,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3826349508733023e-05,
      "loss": 0.6600877642631531,
      "step": 1014
    },
    {
      "epoch": 0.0831209245667373,
      "grad_norm": 0.208984375,
      "learning_rate": 2.3825925186947415e-05,
      "loss": 0.9509165287017822,
      "step": 1015
    },
    {
      "epoch": 0.08320281710325625,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3825500351158394e-05,
      "loss": 0.8809319734573364,
      "step": 1016
    },
    {
      "epoch": 0.0832847096397752,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.382507500138442e-05,
      "loss": 1.0736415386199951,
      "step": 1017
    },
    {
      "epoch": 0.08336660217629416,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3824649137643982e-05,
      "loss": 0.7450970411300659,
      "step": 1018
    },
    {
      "epoch": 0.08344849471281311,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3824222759955594e-05,
      "loss": 0.7833691835403442,
      "step": 1019
    },
    {
      "epoch": 0.08353038724933207,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.382379586833778e-05,
      "loss": 1.0517538785934448,
      "step": 1020
    },
    {
      "epoch": 0.08361227978585102,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3823368462809104e-05,
      "loss": 0.8261633515357971,
      "step": 1021
    },
    {
      "epoch": 0.08369417232236998,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3822940543388134e-05,
      "loss": 0.952375054359436,
      "step": 1022
    },
    {
      "epoch": 0.08377606485888893,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3822512110093474e-05,
      "loss": 0.7037366628646851,
      "step": 1023
    },
    {
      "epoch": 0.08385795739540787,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3822083162943745e-05,
      "loss": 0.8279221653938293,
      "step": 1024
    },
    {
      "epoch": 0.08393984993192682,
      "grad_norm": 0.125,
      "learning_rate": 2.3821653701957587e-05,
      "loss": 0.7880839109420776,
      "step": 1025
    },
    {
      "epoch": 0.08402174246844578,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3821223727153673e-05,
      "loss": 0.8539780974388123,
      "step": 1026
    },
    {
      "epoch": 0.08410363500496473,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3820793238550683e-05,
      "loss": 1.090681791305542,
      "step": 1027
    },
    {
      "epoch": 0.08418552754148369,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3820362236167337e-05,
      "loss": 0.6790997982025146,
      "step": 1028
    },
    {
      "epoch": 0.08426742007800264,
      "grad_norm": 0.15625,
      "learning_rate": 2.3819930720022363e-05,
      "loss": 0.7699350118637085,
      "step": 1029
    },
    {
      "epoch": 0.0843493126145216,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3819498690134516e-05,
      "loss": 0.9489578008651733,
      "step": 1030
    },
    {
      "epoch": 0.08443120515104055,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.381906614652258e-05,
      "loss": 0.7199457883834839,
      "step": 1031
    },
    {
      "epoch": 0.0845130976875595,
      "grad_norm": 0.259765625,
      "learning_rate": 2.3818633089205342e-05,
      "loss": 0.8059130311012268,
      "step": 1032
    },
    {
      "epoch": 0.08459499022407846,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3818199518201637e-05,
      "loss": 0.7351808547973633,
      "step": 1033
    },
    {
      "epoch": 0.0846768827605974,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3817765433530302e-05,
      "loss": 0.9748575091362,
      "step": 1034
    },
    {
      "epoch": 0.08475877529711635,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.381733083521021e-05,
      "loss": 1.220763087272644,
      "step": 1035
    },
    {
      "epoch": 0.08484066783363531,
      "grad_norm": 0.173828125,
      "learning_rate": 2.381689572326025e-05,
      "loss": 0.856031596660614,
      "step": 1036
    },
    {
      "epoch": 0.08492256037015426,
      "grad_norm": 0.154296875,
      "learning_rate": 2.381646009769933e-05,
      "loss": 0.9350159168243408,
      "step": 1037
    },
    {
      "epoch": 0.08500445290667322,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3816023958546383e-05,
      "loss": 0.892288088798523,
      "step": 1038
    },
    {
      "epoch": 0.08508634544319217,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3815587305820372e-05,
      "loss": 0.79413241147995,
      "step": 1039
    },
    {
      "epoch": 0.08516823797971113,
      "grad_norm": 0.1796875,
      "learning_rate": 2.381515013954027e-05,
      "loss": 0.9142088890075684,
      "step": 1040
    },
    {
      "epoch": 0.08525013051623008,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.381471245972508e-05,
      "loss": 0.7598832249641418,
      "step": 1041
    },
    {
      "epoch": 0.08533202305274903,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3814274266393826e-05,
      "loss": 1.1087967157363892,
      "step": 1042
    },
    {
      "epoch": 0.08541391558926799,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3813835559565552e-05,
      "loss": 0.9357991814613342,
      "step": 1043
    },
    {
      "epoch": 0.08549580812578694,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.381339633925933e-05,
      "loss": 1.114556908607483,
      "step": 1044
    },
    {
      "epoch": 0.08557770066230588,
      "grad_norm": 0.13671875,
      "learning_rate": 2.3812956605494242e-05,
      "loss": 0.8096150755882263,
      "step": 1045
    },
    {
      "epoch": 0.08565959319882484,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.381251635828941e-05,
      "loss": 0.8399940133094788,
      "step": 1046
    },
    {
      "epoch": 0.08574148573534379,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3812075597663964e-05,
      "loss": 0.6026678085327148,
      "step": 1047
    },
    {
      "epoch": 0.08582337827186275,
      "grad_norm": 0.17578125,
      "learning_rate": 2.381163432363706e-05,
      "loss": 0.8701160550117493,
      "step": 1048
    },
    {
      "epoch": 0.0859052708083817,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3811192536227883e-05,
      "loss": 1.0040497779846191,
      "step": 1049
    },
    {
      "epoch": 0.08598716334490066,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.381075023545563e-05,
      "loss": 1.1004680395126343,
      "step": 1050
    },
    {
      "epoch": 0.08606905588141961,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3810307421339525e-05,
      "loss": 0.7033898234367371,
      "step": 1051
    },
    {
      "epoch": 0.08615094841793856,
      "grad_norm": 0.1640625,
      "learning_rate": 2.380986409389882e-05,
      "loss": 0.9362397789955139,
      "step": 1052
    },
    {
      "epoch": 0.08623284095445752,
      "grad_norm": 0.146484375,
      "learning_rate": 2.380942025315278e-05,
      "loss": 0.812309205532074,
      "step": 1053
    },
    {
      "epoch": 0.08631473349097647,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3808975899120693e-05,
      "loss": 0.8131587505340576,
      "step": 1054
    },
    {
      "epoch": 0.08639662602749541,
      "grad_norm": 0.171875,
      "learning_rate": 2.3808531031821875e-05,
      "loss": 0.8315556049346924,
      "step": 1055
    },
    {
      "epoch": 0.08647851856401437,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3808085651275667e-05,
      "loss": 0.621181070804596,
      "step": 1056
    },
    {
      "epoch": 0.08656041110053332,
      "grad_norm": 0.162109375,
      "learning_rate": 2.380763975750142e-05,
      "loss": 0.7798931002616882,
      "step": 1057
    },
    {
      "epoch": 0.08664230363705228,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.380719335051852e-05,
      "loss": 0.7550495266914368,
      "step": 1058
    },
    {
      "epoch": 0.08672419617357123,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.380674643034636e-05,
      "loss": 0.8598673343658447,
      "step": 1059
    },
    {
      "epoch": 0.08680608871009018,
      "grad_norm": 0.2421875,
      "learning_rate": 2.380629899700438e-05,
      "loss": 1.1551227569580078,
      "step": 1060
    },
    {
      "epoch": 0.08688798124660914,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3805851050512018e-05,
      "loss": 0.6638979315757751,
      "step": 1061
    },
    {
      "epoch": 0.0869698737831281,
      "grad_norm": 0.15625,
      "learning_rate": 2.380540259088874e-05,
      "loss": 0.8854860067367554,
      "step": 1062
    },
    {
      "epoch": 0.08705176631964705,
      "grad_norm": 0.146484375,
      "learning_rate": 2.3804953618154045e-05,
      "loss": 0.9188386797904968,
      "step": 1063
    },
    {
      "epoch": 0.087133658856166,
      "grad_norm": 0.12890625,
      "learning_rate": 2.3804504132327444e-05,
      "loss": 0.6485326290130615,
      "step": 1064
    },
    {
      "epoch": 0.08721555139268494,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3804054133428474e-05,
      "loss": 0.8500991463661194,
      "step": 1065
    },
    {
      "epoch": 0.0872974439292039,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3803603621476695e-05,
      "loss": 1.025068998336792,
      "step": 1066
    },
    {
      "epoch": 0.08737933646572285,
      "grad_norm": 0.162109375,
      "learning_rate": 2.380315259649169e-05,
      "loss": 0.7836132645606995,
      "step": 1067
    },
    {
      "epoch": 0.0874612290022418,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3802701058493054e-05,
      "loss": 0.718864917755127,
      "step": 1068
    },
    {
      "epoch": 0.08754312153876076,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.380224900750042e-05,
      "loss": 1.11733877658844,
      "step": 1069
    },
    {
      "epoch": 0.08762501407527971,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3801796443533436e-05,
      "loss": 1.039713978767395,
      "step": 1070
    },
    {
      "epoch": 0.08770690661179867,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3801343366611766e-05,
      "loss": 0.887320339679718,
      "step": 1071
    },
    {
      "epoch": 0.08778879914831762,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3800889776755113e-05,
      "loss": 0.954429030418396,
      "step": 1072
    },
    {
      "epoch": 0.08787069168483658,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3800435673983183e-05,
      "loss": 0.8651047945022583,
      "step": 1073
    },
    {
      "epoch": 0.08795258422135553,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3799981058315715e-05,
      "loss": 0.969821572303772,
      "step": 1074
    },
    {
      "epoch": 0.08803447675787449,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3799525929772474e-05,
      "loss": 1.157622218132019,
      "step": 1075
    },
    {
      "epoch": 0.08811636929439343,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3799070288373233e-05,
      "loss": 0.49655666947364807,
      "step": 1076
    },
    {
      "epoch": 0.08819826183091238,
      "grad_norm": 0.21875,
      "learning_rate": 2.3798614134137802e-05,
      "loss": 0.9968538284301758,
      "step": 1077
    },
    {
      "epoch": 0.08828015436743133,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3798157467086003e-05,
      "loss": 1.0708034038543701,
      "step": 1078
    },
    {
      "epoch": 0.08836204690395029,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3797700287237696e-05,
      "loss": 0.873045802116394,
      "step": 1079
    },
    {
      "epoch": 0.08844393944046924,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3797242594612736e-05,
      "loss": 0.841846227645874,
      "step": 1080
    },
    {
      "epoch": 0.0885258319769882,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3796784389231025e-05,
      "loss": 0.7304664850234985,
      "step": 1081
    },
    {
      "epoch": 0.08860772451350715,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3796325671112477e-05,
      "loss": 0.933490514755249,
      "step": 1082
    },
    {
      "epoch": 0.0886896170500261,
      "grad_norm": 0.177734375,
      "learning_rate": 2.379586644027703e-05,
      "loss": 1.2038507461547852,
      "step": 1083
    },
    {
      "epoch": 0.08877150958654506,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3795406696744646e-05,
      "loss": 0.8266881704330444,
      "step": 1084
    },
    {
      "epoch": 0.08885340212306402,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3794946440535304e-05,
      "loss": 0.8591686487197876,
      "step": 1085
    },
    {
      "epoch": 0.08893529465958296,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3794485671669005e-05,
      "loss": 0.9704241752624512,
      "step": 1086
    },
    {
      "epoch": 0.08901718719610191,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3794024390165786e-05,
      "loss": 0.8588113784790039,
      "step": 1087
    },
    {
      "epoch": 0.08909907973262086,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.379356259604569e-05,
      "loss": 0.7976828217506409,
      "step": 1088
    },
    {
      "epoch": 0.08918097226913982,
      "grad_norm": 0.17578125,
      "learning_rate": 2.379310028932879e-05,
      "loss": 0.9168569445610046,
      "step": 1089
    },
    {
      "epoch": 0.08926286480565877,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3792637470035175e-05,
      "loss": 0.6929162740707397,
      "step": 1090
    },
    {
      "epoch": 0.08934475734217773,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3792174138184973e-05,
      "loss": 1.091790795326233,
      "step": 1091
    },
    {
      "epoch": 0.08942664987869668,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.379171029379831e-05,
      "loss": 0.8192435503005981,
      "step": 1092
    },
    {
      "epoch": 0.08950854241521564,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.379124593689535e-05,
      "loss": 0.9715936779975891,
      "step": 1093
    },
    {
      "epoch": 0.08959043495173459,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3790781067496282e-05,
      "loss": 1.0156961679458618,
      "step": 1094
    },
    {
      "epoch": 0.08967232748825354,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.37903156856213e-05,
      "loss": 0.8466988205909729,
      "step": 1095
    },
    {
      "epoch": 0.0897542200247725,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3789849791290642e-05,
      "loss": 0.5893105864524841,
      "step": 1096
    },
    {
      "epoch": 0.08983611256129144,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.378938338452455e-05,
      "loss": 0.7864755392074585,
      "step": 1097
    },
    {
      "epoch": 0.0899180050978104,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.3788916465343302e-05,
      "loss": 0.7493928670883179,
      "step": 1098
    },
    {
      "epoch": 0.08999989763432935,
      "grad_norm": 0.1875,
      "learning_rate": 2.3788449033767187e-05,
      "loss": 1.1909775733947754,
      "step": 1099
    },
    {
      "epoch": 0.0900817901708483,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3787981089816526e-05,
      "loss": 0.8320218324661255,
      "step": 1100
    },
    {
      "epoch": 0.09016368270736726,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3787512633511654e-05,
      "loss": 0.8960492610931396,
      "step": 1101
    },
    {
      "epoch": 0.09024557524388621,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3787043664872934e-05,
      "loss": 0.8856436014175415,
      "step": 1102
    },
    {
      "epoch": 0.09032746778040517,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3786574183920752e-05,
      "loss": 0.8444043397903442,
      "step": 1103
    },
    {
      "epoch": 0.09040936031692412,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3786104190675506e-05,
      "loss": 0.9387022256851196,
      "step": 1104
    },
    {
      "epoch": 0.09049125285344307,
      "grad_norm": 0.173828125,
      "learning_rate": 2.378563368515763e-05,
      "loss": 1.1504004001617432,
      "step": 1105
    },
    {
      "epoch": 0.09057314538996203,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3785162667387576e-05,
      "loss": 0.7708770632743835,
      "step": 1106
    },
    {
      "epoch": 0.09065503792648097,
      "grad_norm": 0.15234375,
      "learning_rate": 2.378469113738581e-05,
      "loss": 0.6220785975456238,
      "step": 1107
    },
    {
      "epoch": 0.09073693046299992,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.378421909517283e-05,
      "loss": 0.8233803510665894,
      "step": 1108
    },
    {
      "epoch": 0.09081882299951888,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.378374654076915e-05,
      "loss": 1.0616023540496826,
      "step": 1109
    },
    {
      "epoch": 0.09090071553603783,
      "grad_norm": 0.205078125,
      "learning_rate": 2.3783273474195315e-05,
      "loss": 0.6827281713485718,
      "step": 1110
    },
    {
      "epoch": 0.09098260807255679,
      "grad_norm": 0.15234375,
      "learning_rate": 2.378279989547188e-05,
      "loss": 0.7312248349189758,
      "step": 1111
    },
    {
      "epoch": 0.09106450060907574,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3782325804619435e-05,
      "loss": 0.9664647579193115,
      "step": 1112
    },
    {
      "epoch": 0.0911463931455947,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.378185120165858e-05,
      "loss": 0.8721472024917603,
      "step": 1113
    },
    {
      "epoch": 0.09122828568211365,
      "grad_norm": 0.166015625,
      "learning_rate": 2.378137608660995e-05,
      "loss": 0.8965354561805725,
      "step": 1114
    },
    {
      "epoch": 0.0913101782186326,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.378090045949419e-05,
      "loss": 1.1307601928710938,
      "step": 1115
    },
    {
      "epoch": 0.09139207075515156,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.378042432033197e-05,
      "loss": 0.9814367294311523,
      "step": 1116
    },
    {
      "epoch": 0.0914739632916705,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.377994766914399e-05,
      "loss": 1.02003812789917,
      "step": 1117
    },
    {
      "epoch": 0.09155585582818945,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.377947050595097e-05,
      "loss": 1.2087690830230713,
      "step": 1118
    },
    {
      "epoch": 0.09163774836470841,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3778992830773644e-05,
      "loss": 0.6142762899398804,
      "step": 1119
    },
    {
      "epoch": 0.09171964090122736,
      "grad_norm": 0.16015625,
      "learning_rate": 2.377851464363278e-05,
      "loss": 0.8210044503211975,
      "step": 1120
    },
    {
      "epoch": 0.09180153343774632,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3778035944549148e-05,
      "loss": 0.7467665076255798,
      "step": 1121
    },
    {
      "epoch": 0.09188342597426527,
      "grad_norm": 0.18359375,
      "learning_rate": 2.377755673354357e-05,
      "loss": 0.7314620018005371,
      "step": 1122
    },
    {
      "epoch": 0.09196531851078422,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3777077010636864e-05,
      "loss": 0.8658515214920044,
      "step": 1123
    },
    {
      "epoch": 0.09204721104730318,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.377659677584989e-05,
      "loss": 0.8021686673164368,
      "step": 1124
    },
    {
      "epoch": 0.09212910358382213,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3776116029203512e-05,
      "loss": 0.7749435901641846,
      "step": 1125
    },
    {
      "epoch": 0.09221099612034109,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3775634770718632e-05,
      "loss": 0.8898816704750061,
      "step": 1126
    },
    {
      "epoch": 0.09229288865686004,
      "grad_norm": 0.2109375,
      "learning_rate": 2.3775153000416165e-05,
      "loss": 1.1790481805801392,
      "step": 1127
    },
    {
      "epoch": 0.09237478119337898,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3774670718317044e-05,
      "loss": 0.62465500831604,
      "step": 1128
    },
    {
      "epoch": 0.09245667372989794,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.3774187924442243e-05,
      "loss": 0.9543071985244751,
      "step": 1129
    },
    {
      "epoch": 0.09253856626641689,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3773704618812737e-05,
      "loss": 0.697795569896698,
      "step": 1130
    },
    {
      "epoch": 0.09262045880293585,
      "grad_norm": 0.138671875,
      "learning_rate": 2.377322080144954e-05,
      "loss": 0.6742991209030151,
      "step": 1131
    },
    {
      "epoch": 0.0927023513394548,
      "grad_norm": 0.31640625,
      "learning_rate": 2.3772736472373675e-05,
      "loss": 1.0602946281433105,
      "step": 1132
    },
    {
      "epoch": 0.09278424387597375,
      "grad_norm": 0.15625,
      "learning_rate": 2.3772251631606192e-05,
      "loss": 0.9222228527069092,
      "step": 1133
    },
    {
      "epoch": 0.09286613641249271,
      "grad_norm": 0.2109375,
      "learning_rate": 2.377176627916817e-05,
      "loss": 0.825689435005188,
      "step": 1134
    },
    {
      "epoch": 0.09294802894901166,
      "grad_norm": 0.16796875,
      "learning_rate": 2.37712804150807e-05,
      "loss": 0.8277634382247925,
      "step": 1135
    },
    {
      "epoch": 0.09302992148553062,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.37707940393649e-05,
      "loss": 0.9808722138404846,
      "step": 1136
    },
    {
      "epoch": 0.09311181402204957,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3770307152041913e-05,
      "loss": 0.6830477118492126,
      "step": 1137
    },
    {
      "epoch": 0.09319370655856851,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3769819753132897e-05,
      "loss": 0.9099923372268677,
      "step": 1138
    },
    {
      "epoch": 0.09327559909508747,
      "grad_norm": 0.16015625,
      "learning_rate": 2.376933184265904e-05,
      "loss": 0.8088845014572144,
      "step": 1139
    },
    {
      "epoch": 0.09335749163160642,
      "grad_norm": 0.205078125,
      "learning_rate": 2.3768843420641545e-05,
      "loss": 0.8877663612365723,
      "step": 1140
    },
    {
      "epoch": 0.09343938416812538,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3768354487101646e-05,
      "loss": 1.0679798126220703,
      "step": 1141
    },
    {
      "epoch": 0.09352127670464433,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3767865042060584e-05,
      "loss": 1.0294158458709717,
      "step": 1142
    },
    {
      "epoch": 0.09360316924116328,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3767375085539645e-05,
      "loss": 0.6938620209693909,
      "step": 1143
    },
    {
      "epoch": 0.09368506177768224,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3766884617560115e-05,
      "loss": 1.0503660440444946,
      "step": 1144
    },
    {
      "epoch": 0.09376695431420119,
      "grad_norm": 0.140625,
      "learning_rate": 2.3766393638143317e-05,
      "loss": 0.6979154944419861,
      "step": 1145
    },
    {
      "epoch": 0.09384884685072015,
      "grad_norm": 0.185546875,
      "learning_rate": 2.376590214731059e-05,
      "loss": 1.1186442375183105,
      "step": 1146
    },
    {
      "epoch": 0.0939307393872391,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3765410145083295e-05,
      "loss": 1.0059125423431396,
      "step": 1147
    },
    {
      "epoch": 0.09401263192375804,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3764917631482815e-05,
      "loss": 0.8100786209106445,
      "step": 1148
    },
    {
      "epoch": 0.094094524460277,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.376442460653056e-05,
      "loss": 1.0981206893920898,
      "step": 1149
    },
    {
      "epoch": 0.09417641699679595,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.376393107024796e-05,
      "loss": 1.0711621046066284,
      "step": 1150
    },
    {
      "epoch": 0.0942583095333149,
      "grad_norm": 0.171875,
      "learning_rate": 2.3763437022656456e-05,
      "loss": 0.9385680556297302,
      "step": 1151
    },
    {
      "epoch": 0.09434020206983386,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3762942463777536e-05,
      "loss": 1.0812402963638306,
      "step": 1152
    },
    {
      "epoch": 0.09442209460635281,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3762447393632687e-05,
      "loss": 0.9176828265190125,
      "step": 1153
    },
    {
      "epoch": 0.09450398714287177,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3761951812243427e-05,
      "loss": 0.6775470972061157,
      "step": 1154
    },
    {
      "epoch": 0.09458587967939072,
      "grad_norm": 0.251953125,
      "learning_rate": 2.3761455719631297e-05,
      "loss": 1.1069231033325195,
      "step": 1155
    },
    {
      "epoch": 0.09466777221590968,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.376095911581786e-05,
      "loss": 1.2215290069580078,
      "step": 1156
    },
    {
      "epoch": 0.09474966475242863,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3760462000824702e-05,
      "loss": 0.8767337203025818,
      "step": 1157
    },
    {
      "epoch": 0.09483155728894758,
      "grad_norm": 0.228515625,
      "learning_rate": 2.3759964374673428e-05,
      "loss": 1.3536542654037476,
      "step": 1158
    },
    {
      "epoch": 0.09491344982546653,
      "grad_norm": 0.23828125,
      "learning_rate": 2.3759466237385665e-05,
      "loss": 0.8645602464675903,
      "step": 1159
    },
    {
      "epoch": 0.09499534236198548,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3758967588983066e-05,
      "loss": 0.7537268400192261,
      "step": 1160
    },
    {
      "epoch": 0.09507723489850443,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3758468429487302e-05,
      "loss": 1.343405842781067,
      "step": 1161
    },
    {
      "epoch": 0.09515912743502339,
      "grad_norm": 0.1875,
      "learning_rate": 2.375796875892007e-05,
      "loss": 0.5895355939865112,
      "step": 1162
    },
    {
      "epoch": 0.09524101997154234,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.375746857730309e-05,
      "loss": 0.7574297785758972,
      "step": 1163
    },
    {
      "epoch": 0.0953229125080613,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3756967884658103e-05,
      "loss": 1.0528451204299927,
      "step": 1164
    },
    {
      "epoch": 0.09540480504458025,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3756466681006865e-05,
      "loss": 1.221675992012024,
      "step": 1165
    },
    {
      "epoch": 0.0954866975810992,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3755964966371164e-05,
      "loss": 0.9941505193710327,
      "step": 1166
    },
    {
      "epoch": 0.09556859011761816,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.375546274077281e-05,
      "loss": 0.8325309753417969,
      "step": 1167
    },
    {
      "epoch": 0.09565048265413711,
      "grad_norm": 0.208984375,
      "learning_rate": 2.3754960004233622e-05,
      "loss": 1.1821049451828003,
      "step": 1168
    },
    {
      "epoch": 0.09573237519065605,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3754456756775458e-05,
      "loss": 0.973341703414917,
      "step": 1169
    },
    {
      "epoch": 0.09581426772717501,
      "grad_norm": 0.22265625,
      "learning_rate": 2.3753952998420193e-05,
      "loss": 0.8207471966743469,
      "step": 1170
    },
    {
      "epoch": 0.09589616026369396,
      "grad_norm": 0.1484375,
      "learning_rate": 2.375344872918972e-05,
      "loss": 0.7534990310668945,
      "step": 1171
    },
    {
      "epoch": 0.09597805280021292,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3752943949105952e-05,
      "loss": 1.007485032081604,
      "step": 1172
    },
    {
      "epoch": 0.09605994533673187,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3752438658190838e-05,
      "loss": 0.6339713931083679,
      "step": 1173
    },
    {
      "epoch": 0.09614183787325083,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3751932856466336e-05,
      "loss": 0.8266494870185852,
      "step": 1174
    },
    {
      "epoch": 0.09622373040976978,
      "grad_norm": 0.15625,
      "learning_rate": 2.3751426543954424e-05,
      "loss": 1.0286155939102173,
      "step": 1175
    },
    {
      "epoch": 0.09630562294628874,
      "grad_norm": 0.169921875,
      "learning_rate": 2.375091972067712e-05,
      "loss": 0.9234880805015564,
      "step": 1176
    },
    {
      "epoch": 0.09638751548280769,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3750412386656444e-05,
      "loss": 0.9872547388076782,
      "step": 1177
    },
    {
      "epoch": 0.09646940801932664,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3749904541914446e-05,
      "loss": 0.8976078629493713,
      "step": 1178
    },
    {
      "epoch": 0.0965513005558456,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3749396186473205e-05,
      "loss": 1.085433006286621,
      "step": 1179
    },
    {
      "epoch": 0.09663319309236454,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3748887320354813e-05,
      "loss": 0.7641810178756714,
      "step": 1180
    },
    {
      "epoch": 0.0967150856288835,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3748377943581386e-05,
      "loss": 0.6829699277877808,
      "step": 1181
    },
    {
      "epoch": 0.09679697816540245,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.374786805617507e-05,
      "loss": 0.8693859577178955,
      "step": 1182
    },
    {
      "epoch": 0.0968788707019214,
      "grad_norm": 0.208984375,
      "learning_rate": 2.374735765815802e-05,
      "loss": 0.7369914650917053,
      "step": 1183
    },
    {
      "epoch": 0.09696076323844036,
      "grad_norm": 0.158203125,
      "learning_rate": 2.374684674955242e-05,
      "loss": 0.9452453255653381,
      "step": 1184
    },
    {
      "epoch": 0.09704265577495931,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3746335330380484e-05,
      "loss": 1.0259038209915161,
      "step": 1185
    },
    {
      "epoch": 0.09712454831147826,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3745823400664435e-05,
      "loss": 0.7812591791152954,
      "step": 1186
    },
    {
      "epoch": 0.09720644084799722,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3745310960426517e-05,
      "loss": 0.6742737293243408,
      "step": 1187
    },
    {
      "epoch": 0.09728833338451617,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3744798009689015e-05,
      "loss": 0.855651319026947,
      "step": 1188
    },
    {
      "epoch": 0.09737022592103513,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3744284548474215e-05,
      "loss": 0.994301974773407,
      "step": 1189
    },
    {
      "epoch": 0.09745211845755407,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3743770576804435e-05,
      "loss": 0.9601473808288574,
      "step": 1190
    },
    {
      "epoch": 0.09753401099407302,
      "grad_norm": 0.17578125,
      "learning_rate": 2.374325609470202e-05,
      "loss": 0.5318287014961243,
      "step": 1191
    },
    {
      "epoch": 0.09761590353059198,
      "grad_norm": 0.220703125,
      "learning_rate": 2.3742741102189332e-05,
      "loss": 1.1253520250320435,
      "step": 1192
    },
    {
      "epoch": 0.09769779606711093,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3742225599288746e-05,
      "loss": 0.7274142503738403,
      "step": 1193
    },
    {
      "epoch": 0.09777968860362989,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3741709586022673e-05,
      "loss": 0.6838446855545044,
      "step": 1194
    },
    {
      "epoch": 0.09786158114014884,
      "grad_norm": 0.2109375,
      "learning_rate": 2.374119306241354e-05,
      "loss": 1.2360278367996216,
      "step": 1195
    },
    {
      "epoch": 0.0979434736766678,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3740676028483803e-05,
      "loss": 0.8400553464889526,
      "step": 1196
    },
    {
      "epoch": 0.09802536621318675,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3740158484255924e-05,
      "loss": 1.3896855115890503,
      "step": 1197
    },
    {
      "epoch": 0.0981072587497057,
      "grad_norm": 0.21875,
      "learning_rate": 2.3739640429752406e-05,
      "loss": 0.9722265005111694,
      "step": 1198
    },
    {
      "epoch": 0.09818915128622466,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3739121864995767e-05,
      "loss": 0.5556004047393799,
      "step": 1199
    },
    {
      "epoch": 0.0982710438227436,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3738602790008533e-05,
      "loss": 0.684959888458252,
      "step": 1200
    },
    {
      "epoch": 0.09835293635926255,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.373808320481328e-05,
      "loss": 0.9369453191757202,
      "step": 1201
    },
    {
      "epoch": 0.0984348288957815,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3737563109432584e-05,
      "loss": 0.9035078287124634,
      "step": 1202
    },
    {
      "epoch": 0.09851672143230046,
      "grad_norm": 0.17578125,
      "learning_rate": 2.373704250388905e-05,
      "loss": 0.9444288611412048,
      "step": 1203
    },
    {
      "epoch": 0.09859861396881942,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.373652138820531e-05,
      "loss": 0.963218629360199,
      "step": 1204
    },
    {
      "epoch": 0.09868050650533837,
      "grad_norm": 0.162109375,
      "learning_rate": 2.373599976240401e-05,
      "loss": 0.8077481985092163,
      "step": 1205
    },
    {
      "epoch": 0.09876239904185732,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3735477626507825e-05,
      "loss": 0.8083269596099854,
      "step": 1206
    },
    {
      "epoch": 0.09884429157837628,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.373495498053944e-05,
      "loss": 0.8624137043952942,
      "step": 1207
    },
    {
      "epoch": 0.09892618411489523,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3734431824521585e-05,
      "loss": 0.7679412364959717,
      "step": 1208
    },
    {
      "epoch": 0.09900807665141419,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3733908158476996e-05,
      "loss": 0.7846046686172485,
      "step": 1209
    },
    {
      "epoch": 0.09908996918793314,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3733383982428425e-05,
      "loss": 0.846065104007721,
      "step": 1210
    },
    {
      "epoch": 0.09917186172445208,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.373285929639866e-05,
      "loss": 0.7402765154838562,
      "step": 1211
    },
    {
      "epoch": 0.09925375426097104,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3732334100410507e-05,
      "loss": 0.6380136013031006,
      "step": 1212
    },
    {
      "epoch": 0.09933564679748999,
      "grad_norm": 0.1640625,
      "learning_rate": 2.373180839448679e-05,
      "loss": 0.767514705657959,
      "step": 1213
    },
    {
      "epoch": 0.09941753933400894,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.3731282178650363e-05,
      "loss": 0.928498387336731,
      "step": 1214
    },
    {
      "epoch": 0.0994994318705279,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3730755452924096e-05,
      "loss": 0.8237026333808899,
      "step": 1215
    },
    {
      "epoch": 0.09958132440704685,
      "grad_norm": 0.16015625,
      "learning_rate": 2.373022821733088e-05,
      "loss": 0.9202016592025757,
      "step": 1216
    },
    {
      "epoch": 0.09966321694356581,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.372970047189363e-05,
      "loss": 0.7350637912750244,
      "step": 1217
    },
    {
      "epoch": 0.09974510948008476,
      "grad_norm": 0.177734375,
      "learning_rate": 2.372917221663529e-05,
      "loss": 0.7856539487838745,
      "step": 1218
    },
    {
      "epoch": 0.09982700201660372,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3728643451578813e-05,
      "loss": 0.8335817456245422,
      "step": 1219
    },
    {
      "epoch": 0.09990889455312267,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.372811417674719e-05,
      "loss": 0.7585546970367432,
      "step": 1220
    },
    {
      "epoch": 0.09999078708964161,
      "grad_norm": 0.20703125,
      "learning_rate": 2.3727584392163416e-05,
      "loss": 1.0586453676223755,
      "step": 1221
    },
    {
      "epoch": 0.10007267962616057,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3727054097850523e-05,
      "loss": 0.8208510279655457,
      "step": 1222
    },
    {
      "epoch": 0.10015457216267952,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.372652329383156e-05,
      "loss": 1.077556848526001,
      "step": 1223
    },
    {
      "epoch": 0.10023646469919847,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3725991980129597e-05,
      "loss": 1.1347525119781494,
      "step": 1224
    },
    {
      "epoch": 0.10031835723571743,
      "grad_norm": 0.20703125,
      "learning_rate": 2.3725460156767725e-05,
      "loss": 0.9661146402359009,
      "step": 1225
    },
    {
      "epoch": 0.10040024977223638,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.372492782376906e-05,
      "loss": 0.8614739179611206,
      "step": 1226
    },
    {
      "epoch": 0.10048214230875534,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.372439498115674e-05,
      "loss": 0.6785455346107483,
      "step": 1227
    },
    {
      "epoch": 0.10056403484527429,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.372386162895393e-05,
      "loss": 0.767665445804596,
      "step": 1228
    },
    {
      "epoch": 0.10064592738179325,
      "grad_norm": 0.18359375,
      "learning_rate": 2.37233277671838e-05,
      "loss": 0.7161230444908142,
      "step": 1229
    },
    {
      "epoch": 0.1007278199183122,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3722793395869566e-05,
      "loss": 1.0382130146026611,
      "step": 1230
    },
    {
      "epoch": 0.10080971245483115,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3722258515034443e-05,
      "loss": 0.7062875628471375,
      "step": 1231
    },
    {
      "epoch": 0.1008916049913501,
      "grad_norm": 0.15625,
      "learning_rate": 2.3721723124701688e-05,
      "loss": 0.8426028490066528,
      "step": 1232
    },
    {
      "epoch": 0.10097349752786905,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3721187224894566e-05,
      "loss": 0.9212101101875305,
      "step": 1233
    },
    {
      "epoch": 0.101055390064388,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3720650815636375e-05,
      "loss": 0.8535024523735046,
      "step": 1234
    },
    {
      "epoch": 0.10113728260090696,
      "grad_norm": 0.20703125,
      "learning_rate": 2.372011389695042e-05,
      "loss": 0.9604092836380005,
      "step": 1235
    },
    {
      "epoch": 0.10121917513742591,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3719576468860048e-05,
      "loss": 0.9024224281311035,
      "step": 1236
    },
    {
      "epoch": 0.10130106767394487,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.371903853138861e-05,
      "loss": 1.0288610458374023,
      "step": 1237
    },
    {
      "epoch": 0.10138296021046382,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3718500084559492e-05,
      "loss": 0.8394074440002441,
      "step": 1238
    },
    {
      "epoch": 0.10146485274698278,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3717961128396096e-05,
      "loss": 0.8096848726272583,
      "step": 1239
    },
    {
      "epoch": 0.10154674528350173,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.3717421662921844e-05,
      "loss": 0.8051522970199585,
      "step": 1240
    },
    {
      "epoch": 0.10162863782002068,
      "grad_norm": 0.228515625,
      "learning_rate": 2.371688168816019e-05,
      "loss": 1.0528494119644165,
      "step": 1241
    },
    {
      "epoch": 0.10171053035653962,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3716341204134595e-05,
      "loss": 0.7388101816177368,
      "step": 1242
    },
    {
      "epoch": 0.10179242289305858,
      "grad_norm": 0.173828125,
      "learning_rate": 2.371580021086856e-05,
      "loss": 1.0895910263061523,
      "step": 1243
    },
    {
      "epoch": 0.10187431542957753,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3715258708385592e-05,
      "loss": 0.7570978999137878,
      "step": 1244
    },
    {
      "epoch": 0.10195620796609649,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3714716696709226e-05,
      "loss": 0.8175119161605835,
      "step": 1245
    },
    {
      "epoch": 0.10203810050261544,
      "grad_norm": 0.146484375,
      "learning_rate": 2.3714174175863027e-05,
      "loss": 0.9203885197639465,
      "step": 1246
    },
    {
      "epoch": 0.1021199930391344,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.371363114587057e-05,
      "loss": 0.7850099205970764,
      "step": 1247
    },
    {
      "epoch": 0.10220188557565335,
      "grad_norm": 0.15625,
      "learning_rate": 2.3713087606755456e-05,
      "loss": 0.9579271078109741,
      "step": 1248
    },
    {
      "epoch": 0.1022837781121723,
      "grad_norm": 0.177734375,
      "learning_rate": 2.371254355854132e-05,
      "loss": 0.795863687992096,
      "step": 1249
    },
    {
      "epoch": 0.10236567064869126,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.371199900125179e-05,
      "loss": 0.7548266649246216,
      "step": 1250
    },
    {
      "epoch": 0.10244756318521021,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3711453934910552e-05,
      "loss": 0.867897093296051,
      "step": 1251
    },
    {
      "epoch": 0.10252945572172915,
      "grad_norm": 0.173828125,
      "learning_rate": 2.371090835954129e-05,
      "loss": 0.845425546169281,
      "step": 1252
    },
    {
      "epoch": 0.10261134825824811,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3710362275167715e-05,
      "loss": 0.7724940776824951,
      "step": 1253
    },
    {
      "epoch": 0.10269324079476706,
      "grad_norm": 0.142578125,
      "learning_rate": 2.3709815681813562e-05,
      "loss": 0.6816731095314026,
      "step": 1254
    },
    {
      "epoch": 0.10277513333128602,
      "grad_norm": 0.166015625,
      "learning_rate": 2.370926857950259e-05,
      "loss": 0.802532970905304,
      "step": 1255
    },
    {
      "epoch": 0.10285702586780497,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3708720968258582e-05,
      "loss": 0.8309116363525391,
      "step": 1256
    },
    {
      "epoch": 0.10293891840432393,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3708172848105336e-05,
      "loss": 0.8417888879776001,
      "step": 1257
    },
    {
      "epoch": 0.10302081094084288,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3707624219066676e-05,
      "loss": 0.7619121670722961,
      "step": 1258
    },
    {
      "epoch": 0.10310270347736183,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.3707075081166445e-05,
      "loss": 0.9109389781951904,
      "step": 1259
    },
    {
      "epoch": 0.10318459601388079,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3706525434428514e-05,
      "loss": 0.9622306227684021,
      "step": 1260
    },
    {
      "epoch": 0.10326648855039974,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.370597527887677e-05,
      "loss": 0.7471734285354614,
      "step": 1261
    },
    {
      "epoch": 0.1033483810869187,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.370542461453513e-05,
      "loss": 1.0727462768554688,
      "step": 1262
    },
    {
      "epoch": 0.10343027362343764,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3704873441427524e-05,
      "loss": 0.8078822493553162,
      "step": 1263
    },
    {
      "epoch": 0.10351216615995659,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.370432175957791e-05,
      "loss": 0.8034855723381042,
      "step": 1264
    },
    {
      "epoch": 0.10359405869647555,
      "grad_norm": 0.29296875,
      "learning_rate": 2.3703769569010266e-05,
      "loss": 0.682559072971344,
      "step": 1265
    },
    {
      "epoch": 0.1036759512329945,
      "grad_norm": 0.212890625,
      "learning_rate": 2.3703216869748593e-05,
      "loss": 0.8307250142097473,
      "step": 1266
    },
    {
      "epoch": 0.10375784376951346,
      "grad_norm": 0.17578125,
      "learning_rate": 2.370266366181691e-05,
      "loss": 0.7775280475616455,
      "step": 1267
    },
    {
      "epoch": 0.10383973630603241,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3702109945239264e-05,
      "loss": 0.8794049024581909,
      "step": 1268
    },
    {
      "epoch": 0.10392162884255136,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3701555720039724e-05,
      "loss": 0.8266291618347168,
      "step": 1269
    },
    {
      "epoch": 0.10400352137907032,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3701000986242377e-05,
      "loss": 1.2334038019180298,
      "step": 1270
    },
    {
      "epoch": 0.10408541391558927,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3700445743871335e-05,
      "loss": 0.8541483283042908,
      "step": 1271
    },
    {
      "epoch": 0.10416730645210823,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.369988999295073e-05,
      "loss": 0.869008481502533,
      "step": 1272
    },
    {
      "epoch": 0.10424919898862717,
      "grad_norm": 0.15625,
      "learning_rate": 2.369933373350472e-05,
      "loss": 0.8213968276977539,
      "step": 1273
    },
    {
      "epoch": 0.10433109152514612,
      "grad_norm": 0.171875,
      "learning_rate": 2.3698776965557477e-05,
      "loss": 1.0679702758789062,
      "step": 1274
    },
    {
      "epoch": 0.10441298406166508,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3698219689133204e-05,
      "loss": 0.9498001933097839,
      "step": 1275
    },
    {
      "epoch": 0.10449487659818403,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.369766190425612e-05,
      "loss": 0.806916356086731,
      "step": 1276
    },
    {
      "epoch": 0.10457676913470298,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3697103610950474e-05,
      "loss": 0.8300641179084778,
      "step": 1277
    },
    {
      "epoch": 0.10465866167122194,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3696544809240522e-05,
      "loss": 0.7898098826408386,
      "step": 1278
    },
    {
      "epoch": 0.1047405542077409,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3695985499150567e-05,
      "loss": 0.6283451318740845,
      "step": 1279
    },
    {
      "epoch": 0.10482244674425985,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3695425680704904e-05,
      "loss": 0.8911735415458679,
      "step": 1280
    },
    {
      "epoch": 0.1049043392807788,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3694865353927875e-05,
      "loss": 0.8601976633071899,
      "step": 1281
    },
    {
      "epoch": 0.10498623181729776,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3694304518843827e-05,
      "loss": 0.7631255388259888,
      "step": 1282
    },
    {
      "epoch": 0.10506812435381671,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3693743175477136e-05,
      "loss": 0.90514075756073,
      "step": 1283
    },
    {
      "epoch": 0.10515001689033565,
      "grad_norm": 0.18359375,
      "learning_rate": 2.369318132385221e-05,
      "loss": 0.739728569984436,
      "step": 1284
    },
    {
      "epoch": 0.1052319094268546,
      "grad_norm": 0.181640625,
      "learning_rate": 2.369261896399346e-05,
      "loss": 1.0946121215820312,
      "step": 1285
    },
    {
      "epoch": 0.10531380196337356,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3692056095925332e-05,
      "loss": 0.6571671366691589,
      "step": 1286
    },
    {
      "epoch": 0.10539569449989251,
      "grad_norm": 0.15234375,
      "learning_rate": 2.369149271967229e-05,
      "loss": 0.7671352624893188,
      "step": 1287
    },
    {
      "epoch": 0.10547758703641147,
      "grad_norm": 0.162109375,
      "learning_rate": 2.369092883525882e-05,
      "loss": 0.6566355228424072,
      "step": 1288
    },
    {
      "epoch": 0.10555947957293042,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3690364442709435e-05,
      "loss": 0.6651175022125244,
      "step": 1289
    },
    {
      "epoch": 0.10564137210944938,
      "grad_norm": 0.171875,
      "learning_rate": 2.3689799542048657e-05,
      "loss": 0.940728485584259,
      "step": 1290
    },
    {
      "epoch": 0.10572326464596833,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3689234133301048e-05,
      "loss": 0.833203911781311,
      "step": 1291
    },
    {
      "epoch": 0.10580515718248729,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3688668216491177e-05,
      "loss": 0.8607549071311951,
      "step": 1292
    },
    {
      "epoch": 0.10588704971900624,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3688101791643644e-05,
      "loss": 0.9978910088539124,
      "step": 1293
    },
    {
      "epoch": 0.10596894225552518,
      "grad_norm": 0.15234375,
      "learning_rate": 2.368753485878307e-05,
      "loss": 0.6208826899528503,
      "step": 1294
    },
    {
      "epoch": 0.10605083479204414,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3686967417934088e-05,
      "loss": 0.9551267623901367,
      "step": 1295
    },
    {
      "epoch": 0.10613272732856309,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3686399469121373e-05,
      "loss": 0.8951752185821533,
      "step": 1296
    },
    {
      "epoch": 0.10621461986508204,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.36858310123696e-05,
      "loss": 0.9852178692817688,
      "step": 1297
    },
    {
      "epoch": 0.106296512401601,
      "grad_norm": 0.181640625,
      "learning_rate": 2.368526204770348e-05,
      "loss": 0.9228842258453369,
      "step": 1298
    },
    {
      "epoch": 0.10637840493811995,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3684692575147747e-05,
      "loss": 0.8692556619644165,
      "step": 1299
    },
    {
      "epoch": 0.1064602974746389,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3684122594727148e-05,
      "loss": 0.6508516669273376,
      "step": 1300
    },
    {
      "epoch": 0.10654219001115786,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3683552106466455e-05,
      "loss": 1.052425742149353,
      "step": 1301
    },
    {
      "epoch": 0.10662408254767682,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3682981110390468e-05,
      "loss": 0.9383931756019592,
      "step": 1302
    },
    {
      "epoch": 0.10670597508419577,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3682409606524003e-05,
      "loss": 0.6776193976402283,
      "step": 1303
    },
    {
      "epoch": 0.10678786762071471,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3681837594891902e-05,
      "loss": 0.697880744934082,
      "step": 1304
    },
    {
      "epoch": 0.10686976015723366,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3681265075519022e-05,
      "loss": 0.8639987111091614,
      "step": 1305
    },
    {
      "epoch": 0.10695165269375262,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3680692048430254e-05,
      "loss": 0.8437212109565735,
      "step": 1306
    },
    {
      "epoch": 0.10703354523027157,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3680118513650496e-05,
      "loss": 0.621090829372406,
      "step": 1307
    },
    {
      "epoch": 0.10711543776679053,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3679544471204686e-05,
      "loss": 0.931085467338562,
      "step": 1308
    },
    {
      "epoch": 0.10719733030330948,
      "grad_norm": 0.1376953125,
      "learning_rate": 2.3678969921117766e-05,
      "loss": 0.8058827519416809,
      "step": 1309
    },
    {
      "epoch": 0.10727922283982844,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.3678394863414707e-05,
      "loss": 0.6842740178108215,
      "step": 1310
    },
    {
      "epoch": 0.10736111537634739,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3677819298120514e-05,
      "loss": 0.8930715918540955,
      "step": 1311
    },
    {
      "epoch": 0.10744300791286635,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3677243225260195e-05,
      "loss": 0.765418529510498,
      "step": 1312
    },
    {
      "epoch": 0.1075249004493853,
      "grad_norm": 0.150390625,
      "learning_rate": 2.367666664485879e-05,
      "loss": 0.9842696189880371,
      "step": 1313
    },
    {
      "epoch": 0.10760679298590425,
      "grad_norm": 0.189453125,
      "learning_rate": 2.367608955694136e-05,
      "loss": 0.7024298906326294,
      "step": 1314
    },
    {
      "epoch": 0.1076886855224232,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3675511961532985e-05,
      "loss": 1.0522289276123047,
      "step": 1315
    },
    {
      "epoch": 0.10777057805894215,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3674933858658776e-05,
      "loss": 0.6449146270751953,
      "step": 1316
    },
    {
      "epoch": 0.1078524705954611,
      "grad_norm": 0.1875,
      "learning_rate": 2.367435524834385e-05,
      "loss": 1.0934807062149048,
      "step": 1317
    },
    {
      "epoch": 0.10793436313198006,
      "grad_norm": 0.193359375,
      "learning_rate": 2.367377613061337e-05,
      "loss": 0.9252569079399109,
      "step": 1318
    },
    {
      "epoch": 0.10801625566849901,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3673196505492493e-05,
      "loss": 0.7038392424583435,
      "step": 1319
    },
    {
      "epoch": 0.10809814820501797,
      "grad_norm": 0.17578125,
      "learning_rate": 2.367261637300642e-05,
      "loss": 0.7427034974098206,
      "step": 1320
    },
    {
      "epoch": 0.10818004074153692,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3672035733180364e-05,
      "loss": 1.0501346588134766,
      "step": 1321
    },
    {
      "epoch": 0.10826193327805587,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.367145458603956e-05,
      "loss": 0.9486469030380249,
      "step": 1322
    },
    {
      "epoch": 0.10834382581457483,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.367087293160927e-05,
      "loss": 1.0136432647705078,
      "step": 1323
    },
    {
      "epoch": 0.10842571835109378,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3670290769914774e-05,
      "loss": 0.835607647895813,
      "step": 1324
    },
    {
      "epoch": 0.10850761088761272,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.366970810098137e-05,
      "loss": 0.6189396977424622,
      "step": 1325
    },
    {
      "epoch": 0.10858950342413168,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.366912492483439e-05,
      "loss": 0.5748917460441589,
      "step": 1326
    },
    {
      "epoch": 0.10867139596065063,
      "grad_norm": 0.1640625,
      "learning_rate": 2.366854124149918e-05,
      "loss": 0.8622034788131714,
      "step": 1327
    },
    {
      "epoch": 0.10875328849716959,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.366795705100111e-05,
      "loss": 0.7574789524078369,
      "step": 1328
    },
    {
      "epoch": 0.10883518103368854,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3667372353365567e-05,
      "loss": 0.8984616994857788,
      "step": 1329
    },
    {
      "epoch": 0.1089170735702075,
      "grad_norm": 0.1875,
      "learning_rate": 2.3666787148617968e-05,
      "loss": 1.061842679977417,
      "step": 1330
    },
    {
      "epoch": 0.10899896610672645,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.366620143678375e-05,
      "loss": 1.0523066520690918,
      "step": 1331
    },
    {
      "epoch": 0.1090808586432454,
      "grad_norm": 0.1875,
      "learning_rate": 2.3665615217888367e-05,
      "loss": 1.214852213859558,
      "step": 1332
    },
    {
      "epoch": 0.10916275117976436,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3665028491957296e-05,
      "loss": 0.6970698237419128,
      "step": 1333
    },
    {
      "epoch": 0.10924464371628331,
      "grad_norm": 0.15625,
      "learning_rate": 2.3664441259016047e-05,
      "loss": 0.4861088991165161,
      "step": 1334
    },
    {
      "epoch": 0.10932653625280227,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3663853519090136e-05,
      "loss": 0.9725649952888489,
      "step": 1335
    },
    {
      "epoch": 0.10940842878932121,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3663265272205112e-05,
      "loss": 1.0047404766082764,
      "step": 1336
    },
    {
      "epoch": 0.10949032132584016,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.366267651838654e-05,
      "loss": 0.6580833196640015,
      "step": 1337
    },
    {
      "epoch": 0.10957221386235912,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3662087257660012e-05,
      "loss": 0.9446367025375366,
      "step": 1338
    },
    {
      "epoch": 0.10965410639887807,
      "grad_norm": 0.197265625,
      "learning_rate": 2.366149749005114e-05,
      "loss": 1.0608404874801636,
      "step": 1339
    },
    {
      "epoch": 0.10973599893539702,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3660907215585557e-05,
      "loss": 1.1110117435455322,
      "step": 1340
    },
    {
      "epoch": 0.10981789147191598,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3660316434288915e-05,
      "loss": 0.9198451638221741,
      "step": 1341
    },
    {
      "epoch": 0.10989978400843493,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.36597251461869e-05,
      "loss": 0.7069704532623291,
      "step": 1342
    },
    {
      "epoch": 0.10998167654495389,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.365913335130521e-05,
      "loss": 0.931777834892273,
      "step": 1343
    },
    {
      "epoch": 0.11006356908147284,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3658541049669557e-05,
      "loss": 0.8208369016647339,
      "step": 1344
    },
    {
      "epoch": 0.1101454616179918,
      "grad_norm": 0.203125,
      "learning_rate": 2.3657948241305698e-05,
      "loss": 1.2157013416290283,
      "step": 1345
    },
    {
      "epoch": 0.11022735415451074,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3657354926239387e-05,
      "loss": 1.0143898725509644,
      "step": 1346
    },
    {
      "epoch": 0.11030924669102969,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3656761104496422e-05,
      "loss": 1.0013749599456787,
      "step": 1347
    },
    {
      "epoch": 0.11039113922754865,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3656166776102607e-05,
      "loss": 0.8227722644805908,
      "step": 1348
    },
    {
      "epoch": 0.1104730317640676,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3655571941083777e-05,
      "loss": 0.7714188098907471,
      "step": 1349
    },
    {
      "epoch": 0.11055492430058655,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3654976599465783e-05,
      "loss": 1.1967484951019287,
      "step": 1350
    },
    {
      "epoch": 0.11063681683710551,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3654380751274504e-05,
      "loss": 1.0309585332870483,
      "step": 1351
    },
    {
      "epoch": 0.11071870937362446,
      "grad_norm": 0.208984375,
      "learning_rate": 2.3653784396535835e-05,
      "loss": 0.666429340839386,
      "step": 1352
    },
    {
      "epoch": 0.11080060191014342,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3653187535275702e-05,
      "loss": 1.0589696168899536,
      "step": 1353
    },
    {
      "epoch": 0.11088249444666237,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3652590167520038e-05,
      "loss": 0.8405098915100098,
      "step": 1354
    },
    {
      "epoch": 0.11096438698318133,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3651992293294812e-05,
      "loss": 0.9821621179580688,
      "step": 1355
    },
    {
      "epoch": 0.11104627951970027,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3651393912626013e-05,
      "loss": 1.0876461267471313,
      "step": 1356
    },
    {
      "epoch": 0.11112817205621922,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3650795025539643e-05,
      "loss": 0.7555913329124451,
      "step": 1357
    },
    {
      "epoch": 0.11121006459273818,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3650195632061737e-05,
      "loss": 0.8301786184310913,
      "step": 1358
    },
    {
      "epoch": 0.11129195712925713,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3649595732218344e-05,
      "loss": 0.8340923190116882,
      "step": 1359
    },
    {
      "epoch": 0.11137384966577608,
      "grad_norm": 0.177734375,
      "learning_rate": 2.364899532603554e-05,
      "loss": 0.6152247786521912,
      "step": 1360
    },
    {
      "epoch": 0.11145574220229504,
      "grad_norm": 0.16796875,
      "learning_rate": 2.364839441353942e-05,
      "loss": 0.8651869893074036,
      "step": 1361
    },
    {
      "epoch": 0.11153763473881399,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3647792994756105e-05,
      "loss": 0.750312328338623,
      "step": 1362
    },
    {
      "epoch": 0.11161952727533295,
      "grad_norm": 0.171875,
      "learning_rate": 2.3647191069711728e-05,
      "loss": 0.6322492361068726,
      "step": 1363
    },
    {
      "epoch": 0.1117014198118519,
      "grad_norm": 0.23046875,
      "learning_rate": 2.364658863843246e-05,
      "loss": 0.7558035254478455,
      "step": 1364
    },
    {
      "epoch": 0.11178331234837086,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.3645985700944483e-05,
      "loss": 0.9625349044799805,
      "step": 1365
    },
    {
      "epoch": 0.11186520488488981,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3645382257273992e-05,
      "loss": 1.1611638069152832,
      "step": 1366
    },
    {
      "epoch": 0.11194709742140875,
      "grad_norm": 0.169921875,
      "learning_rate": 2.364477830744723e-05,
      "loss": 0.7579199075698853,
      "step": 1367
    },
    {
      "epoch": 0.1120289899579277,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3644173851490444e-05,
      "loss": 0.9915123581886292,
      "step": 1368
    },
    {
      "epoch": 0.11211088249444666,
      "grad_norm": 0.169921875,
      "learning_rate": 2.36435688894299e-05,
      "loss": 0.8808772563934326,
      "step": 1369
    },
    {
      "epoch": 0.11219277503096561,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3642963421291897e-05,
      "loss": 0.7593473792076111,
      "step": 1370
    },
    {
      "epoch": 0.11227466756748457,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.3642357447102748e-05,
      "loss": 0.8324772119522095,
      "step": 1371
    },
    {
      "epoch": 0.11235656010400352,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3641750966888795e-05,
      "loss": 0.93071049451828,
      "step": 1372
    },
    {
      "epoch": 0.11243845264052248,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.36411439806764e-05,
      "loss": 1.1701622009277344,
      "step": 1373
    },
    {
      "epoch": 0.11252034517704143,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3640536488491934e-05,
      "loss": 1.3812800645828247,
      "step": 1374
    },
    {
      "epoch": 0.11260223771356039,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3639928490361814e-05,
      "loss": 0.6663912534713745,
      "step": 1375
    },
    {
      "epoch": 0.11268413025007934,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3639319986312456e-05,
      "loss": 1.032552719116211,
      "step": 1376
    },
    {
      "epoch": 0.11276602278659828,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3638710976370313e-05,
      "loss": 0.7712052464485168,
      "step": 1377
    },
    {
      "epoch": 0.11284791532311723,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3638101460561855e-05,
      "loss": 1.2373815774917603,
      "step": 1378
    },
    {
      "epoch": 0.11292980785963619,
      "grad_norm": 0.171875,
      "learning_rate": 2.3637491438913575e-05,
      "loss": 1.1107985973358154,
      "step": 1379
    },
    {
      "epoch": 0.11301170039615514,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3636880911451987e-05,
      "loss": 0.795857310295105,
      "step": 1380
    },
    {
      "epoch": 0.1130935929326741,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3636269878203622e-05,
      "loss": 0.660552978515625,
      "step": 1381
    },
    {
      "epoch": 0.11317548546919305,
      "grad_norm": 0.212890625,
      "learning_rate": 2.3635658339195045e-05,
      "loss": 0.8467762470245361,
      "step": 1382
    },
    {
      "epoch": 0.113257378005712,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3635046294452833e-05,
      "loss": 0.7546990513801575,
      "step": 1383
    },
    {
      "epoch": 0.11333927054223096,
      "grad_norm": 0.134765625,
      "learning_rate": 2.3634433744003585e-05,
      "loss": 0.6038516759872437,
      "step": 1384
    },
    {
      "epoch": 0.11342116307874991,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.363382068787393e-05,
      "loss": 0.716826319694519,
      "step": 1385
    },
    {
      "epoch": 0.11350305561526887,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3633207126090512e-05,
      "loss": 0.6720520257949829,
      "step": 1386
    },
    {
      "epoch": 0.11358494815178781,
      "grad_norm": 0.18359375,
      "learning_rate": 2.363259305868e-05,
      "loss": 0.9037716388702393,
      "step": 1387
    },
    {
      "epoch": 0.11366684068830676,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3631978485669078e-05,
      "loss": 0.9661259651184082,
      "step": 1388
    },
    {
      "epoch": 0.11374873322482572,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3631363407084466e-05,
      "loss": 1.051241397857666,
      "step": 1389
    },
    {
      "epoch": 0.11383062576134467,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.363074782295289e-05,
      "loss": 1.0183197259902954,
      "step": 1390
    },
    {
      "epoch": 0.11391251829786363,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3630131733301116e-05,
      "loss": 0.7491849660873413,
      "step": 1391
    },
    {
      "epoch": 0.11399441083438258,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3629515138155914e-05,
      "loss": 0.8053170442581177,
      "step": 1392
    },
    {
      "epoch": 0.11407630337090154,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3628898037544083e-05,
      "loss": 0.7001284956932068,
      "step": 1393
    },
    {
      "epoch": 0.11415819590742049,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3628280431492448e-05,
      "loss": 0.6447217464447021,
      "step": 1394
    },
    {
      "epoch": 0.11424008844393944,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3627662320027855e-05,
      "loss": 0.7947399616241455,
      "step": 1395
    },
    {
      "epoch": 0.1143219809804584,
      "grad_norm": 0.255859375,
      "learning_rate": 2.3627043703177164e-05,
      "loss": 1.1202263832092285,
      "step": 1396
    },
    {
      "epoch": 0.11440387351697735,
      "grad_norm": 0.2109375,
      "learning_rate": 2.3626424580967264e-05,
      "loss": 0.8530346751213074,
      "step": 1397
    },
    {
      "epoch": 0.1144857660534963,
      "grad_norm": 0.185546875,
      "learning_rate": 2.362580495342507e-05,
      "loss": 0.8882307410240173,
      "step": 1398
    },
    {
      "epoch": 0.11456765859001525,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.362518482057751e-05,
      "loss": 0.9850342869758606,
      "step": 1399
    },
    {
      "epoch": 0.1146495511265342,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3624564182451533e-05,
      "loss": 1.0179409980773926,
      "step": 1400
    },
    {
      "epoch": 0.11473144366305316,
      "grad_norm": 0.203125,
      "learning_rate": 2.3623943039074118e-05,
      "loss": 1.0570334196090698,
      "step": 1401
    },
    {
      "epoch": 0.11481333619957211,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3623321390472267e-05,
      "loss": 0.7722122669219971,
      "step": 1402
    },
    {
      "epoch": 0.11489522873609107,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3622699236672993e-05,
      "loss": 1.1823748350143433,
      "step": 1403
    },
    {
      "epoch": 0.11497712127261002,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3622076577703344e-05,
      "loss": 0.8372278809547424,
      "step": 1404
    },
    {
      "epoch": 0.11505901380912897,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3621453413590375e-05,
      "loss": 0.8335072994232178,
      "step": 1405
    },
    {
      "epoch": 0.11514090634564793,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3620829744361175e-05,
      "loss": 1.0014448165893555,
      "step": 1406
    },
    {
      "epoch": 0.11522279888216688,
      "grad_norm": 0.2421875,
      "learning_rate": 2.362020557004285e-05,
      "loss": 0.9114251732826233,
      "step": 1407
    },
    {
      "epoch": 0.11530469141868582,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3619580890662536e-05,
      "loss": 0.8452286124229431,
      "step": 1408
    },
    {
      "epoch": 0.11538658395520478,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3618955706247376e-05,
      "loss": 0.7000858187675476,
      "step": 1409
    },
    {
      "epoch": 0.11546847649172373,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3618330016824547e-05,
      "loss": 1.0321189165115356,
      "step": 1410
    },
    {
      "epoch": 0.11555036902824269,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.361770382242124e-05,
      "loss": 0.6062124967575073,
      "step": 1411
    },
    {
      "epoch": 0.11563226156476164,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.361707712306468e-05,
      "loss": 0.6719691753387451,
      "step": 1412
    },
    {
      "epoch": 0.1157141541012806,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3616449918782097e-05,
      "loss": 1.252720832824707,
      "step": 1413
    },
    {
      "epoch": 0.11579604663779955,
      "grad_norm": 0.21484375,
      "learning_rate": 2.361582220960076e-05,
      "loss": 0.6980074644088745,
      "step": 1414
    },
    {
      "epoch": 0.1158779391743185,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3615193995547945e-05,
      "loss": 0.7396358847618103,
      "step": 1415
    },
    {
      "epoch": 0.11595983171083746,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.361456527665096e-05,
      "loss": 0.527680516242981,
      "step": 1416
    },
    {
      "epoch": 0.11604172424735641,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3613936052937133e-05,
      "loss": 1.0667195320129395,
      "step": 1417
    },
    {
      "epoch": 0.11612361678387537,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3613306324433805e-05,
      "loss": 0.7322046756744385,
      "step": 1418
    },
    {
      "epoch": 0.1162055093203943,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3612676091168358e-05,
      "loss": 0.7747830152511597,
      "step": 1419
    },
    {
      "epoch": 0.11628740185691326,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3612045353168175e-05,
      "loss": 0.8319998979568481,
      "step": 1420
    },
    {
      "epoch": 0.11636929439343222,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.361141411046068e-05,
      "loss": 0.6736820340156555,
      "step": 1421
    },
    {
      "epoch": 0.11645118692995117,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3610782363073302e-05,
      "loss": 0.9293571710586548,
      "step": 1422
    },
    {
      "epoch": 0.11653307946647012,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3610150111033496e-05,
      "loss": 0.8201375007629395,
      "step": 1423
    },
    {
      "epoch": 0.11661497200298908,
      "grad_norm": 0.1796875,
      "learning_rate": 2.360951735436875e-05,
      "loss": 1.0424195528030396,
      "step": 1424
    },
    {
      "epoch": 0.11669686453950803,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.3608884093106566e-05,
      "loss": 0.950284481048584,
      "step": 1425
    },
    {
      "epoch": 0.11677875707602699,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3608250327274465e-05,
      "loss": 0.8470015525817871,
      "step": 1426
    },
    {
      "epoch": 0.11686064961254594,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3607616056899995e-05,
      "loss": 0.7615162134170532,
      "step": 1427
    },
    {
      "epoch": 0.1169425421490649,
      "grad_norm": 0.154296875,
      "learning_rate": 2.360698128201072e-05,
      "loss": 0.6562010049819946,
      "step": 1428
    },
    {
      "epoch": 0.11702443468558384,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3606346002634232e-05,
      "loss": 0.917264461517334,
      "step": 1429
    },
    {
      "epoch": 0.11710632722210279,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.360571021879815e-05,
      "loss": 0.8433929681777954,
      "step": 1430
    },
    {
      "epoch": 0.11718821975862175,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3605073930530096e-05,
      "loss": 1.0980688333511353,
      "step": 1431
    },
    {
      "epoch": 0.1172701122951407,
      "grad_norm": 0.16015625,
      "learning_rate": 2.360443713785773e-05,
      "loss": 0.9113644361495972,
      "step": 1432
    },
    {
      "epoch": 0.11735200483165965,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.360379984080874e-05,
      "loss": 0.8511938452720642,
      "step": 1433
    },
    {
      "epoch": 0.11743389736817861,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3603162039410807e-05,
      "loss": 1.0827933549880981,
      "step": 1434
    },
    {
      "epoch": 0.11751578990469756,
      "grad_norm": 0.197265625,
      "learning_rate": 2.360252373369166e-05,
      "loss": 0.9591037034988403,
      "step": 1435
    },
    {
      "epoch": 0.11759768244121652,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3601884923679053e-05,
      "loss": 0.973788321018219,
      "step": 1436
    },
    {
      "epoch": 0.11767957497773547,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.360124560940074e-05,
      "loss": 0.8529676795005798,
      "step": 1437
    },
    {
      "epoch": 0.11776146751425443,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3600605790884508e-05,
      "loss": 1.0046101808547974,
      "step": 1438
    },
    {
      "epoch": 0.11784336005077337,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.359996546815817e-05,
      "loss": 1.1490534543991089,
      "step": 1439
    },
    {
      "epoch": 0.11792525258729232,
      "grad_norm": 0.12890625,
      "learning_rate": 2.3599324641249557e-05,
      "loss": 0.5618139505386353,
      "step": 1440
    },
    {
      "epoch": 0.11800714512381127,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.359868331018652e-05,
      "loss": 0.9995988607406616,
      "step": 1441
    },
    {
      "epoch": 0.11808903766033023,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3598041474996936e-05,
      "loss": 0.8002933263778687,
      "step": 1442
    },
    {
      "epoch": 0.11817093019684918,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3597399135708698e-05,
      "loss": 0.6839928030967712,
      "step": 1443
    },
    {
      "epoch": 0.11825282273336814,
      "grad_norm": 0.173828125,
      "learning_rate": 2.359675629234973e-05,
      "loss": 0.6120706796646118,
      "step": 1444
    },
    {
      "epoch": 0.11833471526988709,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.359611294494797e-05,
      "loss": 1.011182427406311,
      "step": 1445
    },
    {
      "epoch": 0.11841660780640605,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.359546909353138e-05,
      "loss": 0.8566934466362,
      "step": 1446
    },
    {
      "epoch": 0.118498500342925,
      "grad_norm": 0.20703125,
      "learning_rate": 2.3594824738127945e-05,
      "loss": 0.9829684495925903,
      "step": 1447
    },
    {
      "epoch": 0.11858039287944395,
      "grad_norm": 0.185546875,
      "learning_rate": 2.359417987876567e-05,
      "loss": 0.8409010171890259,
      "step": 1448
    },
    {
      "epoch": 0.11866228541596291,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3593534515472586e-05,
      "loss": 0.9676474332809448,
      "step": 1449
    },
    {
      "epoch": 0.11874417795248185,
      "grad_norm": 0.228515625,
      "learning_rate": 2.359288864827674e-05,
      "loss": 0.9093770384788513,
      "step": 1450
    },
    {
      "epoch": 0.1188260704890008,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.359224227720621e-05,
      "loss": 0.7728246450424194,
      "step": 1451
    },
    {
      "epoch": 0.11890796302551976,
      "grad_norm": 0.177734375,
      "learning_rate": 2.359159540228908e-05,
      "loss": 0.9202204942703247,
      "step": 1452
    },
    {
      "epoch": 0.11898985556203871,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3590948023553475e-05,
      "loss": 1.076067566871643,
      "step": 1453
    },
    {
      "epoch": 0.11907174809855767,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3590300141027527e-05,
      "loss": 0.8276640176773071,
      "step": 1454
    },
    {
      "epoch": 0.11915364063507662,
      "grad_norm": 0.15625,
      "learning_rate": 2.3589651754739402e-05,
      "loss": 0.7712468504905701,
      "step": 1455
    },
    {
      "epoch": 0.11923553317159558,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3589002864717273e-05,
      "loss": 0.5769593715667725,
      "step": 1456
    },
    {
      "epoch": 0.11931742570811453,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.358835347098935e-05,
      "loss": 0.4050445556640625,
      "step": 1457
    },
    {
      "epoch": 0.11939931824463348,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3587703573583857e-05,
      "loss": 0.7761590480804443,
      "step": 1458
    },
    {
      "epoch": 0.11948121078115244,
      "grad_norm": 0.16015625,
      "learning_rate": 2.358705317252904e-05,
      "loss": 0.8701906800270081,
      "step": 1459
    },
    {
      "epoch": 0.11956310331767138,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3586402267853167e-05,
      "loss": 1.0924251079559326,
      "step": 1460
    },
    {
      "epoch": 0.11964499585419033,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3585750859584532e-05,
      "loss": 1.1683921813964844,
      "step": 1461
    },
    {
      "epoch": 0.11972688839070929,
      "grad_norm": 0.220703125,
      "learning_rate": 2.3585098947751446e-05,
      "loss": 0.8024615049362183,
      "step": 1462
    },
    {
      "epoch": 0.11980878092722824,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.358444653238224e-05,
      "loss": 0.9534985423088074,
      "step": 1463
    },
    {
      "epoch": 0.1198906734637472,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3583793613505283e-05,
      "loss": 0.6281746029853821,
      "step": 1464
    },
    {
      "epoch": 0.11997256600026615,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3583140191148943e-05,
      "loss": 1.019468903541565,
      "step": 1465
    },
    {
      "epoch": 0.1200544585367851,
      "grad_norm": 0.1298828125,
      "learning_rate": 2.358248626534162e-05,
      "loss": 0.6304879188537598,
      "step": 1466
    },
    {
      "epoch": 0.12013635107330406,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.358183183611174e-05,
      "loss": 0.7297646999359131,
      "step": 1467
    },
    {
      "epoch": 0.12021824360982301,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3581176903487746e-05,
      "loss": 0.6875846982002258,
      "step": 1468
    },
    {
      "epoch": 0.12030013614634197,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3580521467498107e-05,
      "loss": 0.662101149559021,
      "step": 1469
    },
    {
      "epoch": 0.12038202868286092,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3579865528171306e-05,
      "loss": 0.9831692576408386,
      "step": 1470
    },
    {
      "epoch": 0.12046392121937986,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3579209085535854e-05,
      "loss": 0.901297390460968,
      "step": 1471
    },
    {
      "epoch": 0.12054581375589882,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3578552139620286e-05,
      "loss": 0.7877477407455444,
      "step": 1472
    },
    {
      "epoch": 0.12062770629241777,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3577894690453154e-05,
      "loss": 0.7524563074111938,
      "step": 1473
    },
    {
      "epoch": 0.12070959882893673,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.357723673806303e-05,
      "loss": 0.7991000413894653,
      "step": 1474
    },
    {
      "epoch": 0.12079149136545568,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3576578282478522e-05,
      "loss": 0.8455972671508789,
      "step": 1475
    },
    {
      "epoch": 0.12087338390197463,
      "grad_norm": 0.13671875,
      "learning_rate": 2.357591932372823e-05,
      "loss": 0.6629756689071655,
      "step": 1476
    },
    {
      "epoch": 0.12095527643849359,
      "grad_norm": 0.2314453125,
      "learning_rate": 2.3575259861840813e-05,
      "loss": 0.94914710521698,
      "step": 1477
    },
    {
      "epoch": 0.12103716897501254,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3574599896844926e-05,
      "loss": 1.237830400466919,
      "step": 1478
    },
    {
      "epoch": 0.1211190615115315,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3573939428769255e-05,
      "loss": 1.1077529191970825,
      "step": 1479
    },
    {
      "epoch": 0.12120095404805045,
      "grad_norm": 0.17578125,
      "learning_rate": 2.357327845764251e-05,
      "loss": 0.6586318016052246,
      "step": 1480
    },
    {
      "epoch": 0.12128284658456939,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.3572616983493417e-05,
      "loss": 1.1534115076065063,
      "step": 1481
    },
    {
      "epoch": 0.12136473912108835,
      "grad_norm": 0.15234375,
      "learning_rate": 2.357195500635072e-05,
      "loss": 0.8297834992408752,
      "step": 1482
    },
    {
      "epoch": 0.1214466316576073,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.35712925262432e-05,
      "loss": 0.5930123329162598,
      "step": 1483
    },
    {
      "epoch": 0.12152852419412626,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3570629543199648e-05,
      "loss": 0.9174181222915649,
      "step": 1484
    },
    {
      "epoch": 0.12161041673064521,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.356996605724888e-05,
      "loss": 0.677579402923584,
      "step": 1485
    },
    {
      "epoch": 0.12169230926716416,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3569302068419734e-05,
      "loss": 0.5644012689590454,
      "step": 1486
    },
    {
      "epoch": 0.12177420180368312,
      "grad_norm": 0.171875,
      "learning_rate": 2.3568637576741072e-05,
      "loss": 0.8930041790008545,
      "step": 1487
    },
    {
      "epoch": 0.12185609434020207,
      "grad_norm": 0.197265625,
      "learning_rate": 2.356797258224177e-05,
      "loss": 1.1511998176574707,
      "step": 1488
    },
    {
      "epoch": 0.12193798687672103,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3567307084950736e-05,
      "loss": 1.2148027420043945,
      "step": 1489
    },
    {
      "epoch": 0.12201987941323998,
      "grad_norm": 0.142578125,
      "learning_rate": 2.3566641084896894e-05,
      "loss": 0.7131853103637695,
      "step": 1490
    },
    {
      "epoch": 0.12210177194975892,
      "grad_norm": 0.490234375,
      "learning_rate": 2.3565974582109186e-05,
      "loss": 1.1556023359298706,
      "step": 1491
    },
    {
      "epoch": 0.12218366448627788,
      "grad_norm": 0.2294921875,
      "learning_rate": 2.3565307576616593e-05,
      "loss": 1.1626863479614258,
      "step": 1492
    },
    {
      "epoch": 0.12226555702279683,
      "grad_norm": 0.1376953125,
      "learning_rate": 2.3564640068448096e-05,
      "loss": 1.0818698406219482,
      "step": 1493
    },
    {
      "epoch": 0.12234744955931579,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.356397205763271e-05,
      "loss": 0.8868678212165833,
      "step": 1494
    },
    {
      "epoch": 0.12242934209583474,
      "grad_norm": 0.140625,
      "learning_rate": 2.356330354419947e-05,
      "loss": 1.000571608543396,
      "step": 1495
    },
    {
      "epoch": 0.1225112346323537,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3562634528177433e-05,
      "loss": 0.9135953783988953,
      "step": 1496
    },
    {
      "epoch": 0.12259312716887265,
      "grad_norm": 0.1279296875,
      "learning_rate": 2.3561965009595674e-05,
      "loss": 0.6179946064949036,
      "step": 1497
    },
    {
      "epoch": 0.1226750197053916,
      "grad_norm": 0.1796875,
      "learning_rate": 2.35612949884833e-05,
      "loss": 1.0819332599639893,
      "step": 1498
    },
    {
      "epoch": 0.12275691224191056,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3560624464869427e-05,
      "loss": 0.7386307120323181,
      "step": 1499
    },
    {
      "epoch": 0.12283880477842951,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3559953438783195e-05,
      "loss": 0.6335099935531616,
      "step": 1500
    },
    {
      "epoch": 0.12292069731494847,
      "grad_norm": 0.162109375,
      "learning_rate": 2.355928191025378e-05,
      "loss": 0.7543211579322815,
      "step": 1501
    },
    {
      "epoch": 0.1230025898514674,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.355860987931036e-05,
      "loss": 0.7102307081222534,
      "step": 1502
    },
    {
      "epoch": 0.12308448238798636,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3557937345982152e-05,
      "loss": 0.8057283163070679,
      "step": 1503
    },
    {
      "epoch": 0.12316637492450531,
      "grad_norm": 0.16796875,
      "learning_rate": 2.355726431029838e-05,
      "loss": 1.0413039922714233,
      "step": 1504
    },
    {
      "epoch": 0.12324826746102427,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3556590772288304e-05,
      "loss": 0.7117171287536621,
      "step": 1505
    },
    {
      "epoch": 0.12333015999754322,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3555916731981194e-05,
      "loss": 1.0694845914840698,
      "step": 1506
    },
    {
      "epoch": 0.12341205253406218,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3555242189406344e-05,
      "loss": 0.8402178883552551,
      "step": 1507
    },
    {
      "epoch": 0.12349394507058113,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.355456714459308e-05,
      "loss": 0.9071080684661865,
      "step": 1508
    },
    {
      "epoch": 0.12357583760710009,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3553891597570737e-05,
      "loss": 0.8198522329330444,
      "step": 1509
    },
    {
      "epoch": 0.12365773014361904,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3553215548368677e-05,
      "loss": 0.4995543956756592,
      "step": 1510
    },
    {
      "epoch": 0.123739622680138,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3552538997016284e-05,
      "loss": 0.8689622282981873,
      "step": 1511
    },
    {
      "epoch": 0.12382151521665694,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3551861943542967e-05,
      "loss": 1.2427992820739746,
      "step": 1512
    },
    {
      "epoch": 0.12390340775317589,
      "grad_norm": 0.169921875,
      "learning_rate": 2.355118438797815e-05,
      "loss": 0.8442955017089844,
      "step": 1513
    },
    {
      "epoch": 0.12398530028969484,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3550506330351287e-05,
      "loss": 0.8814522624015808,
      "step": 1514
    },
    {
      "epoch": 0.1240671928262138,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.354982777069184e-05,
      "loss": 0.9446765184402466,
      "step": 1515
    },
    {
      "epoch": 0.12414908536273275,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3549148709029314e-05,
      "loss": 0.8336957693099976,
      "step": 1516
    },
    {
      "epoch": 0.12423097789925171,
      "grad_norm": 0.203125,
      "learning_rate": 2.3548469145393214e-05,
      "loss": 1.1811503171920776,
      "step": 1517
    },
    {
      "epoch": 0.12431287043577066,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.354778907981308e-05,
      "loss": 1.0255887508392334,
      "step": 1518
    },
    {
      "epoch": 0.12439476297228962,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3547108512318475e-05,
      "loss": 0.6583116054534912,
      "step": 1519
    },
    {
      "epoch": 0.12447665550880857,
      "grad_norm": 0.181640625,
      "learning_rate": 2.354642744293897e-05,
      "loss": 1.0843958854675293,
      "step": 1520
    },
    {
      "epoch": 0.12455854804532752,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3545745871704175e-05,
      "loss": 0.955223798751831,
      "step": 1521
    },
    {
      "epoch": 0.12464044058184648,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3545063798643708e-05,
      "loss": 0.7975376844406128,
      "step": 1522
    },
    {
      "epoch": 0.12472233311836542,
      "grad_norm": 0.158203125,
      "learning_rate": 2.354438122378722e-05,
      "loss": 0.6243172883987427,
      "step": 1523
    },
    {
      "epoch": 0.12480422565488437,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3543698147164376e-05,
      "loss": 0.5798383951187134,
      "step": 1524
    },
    {
      "epoch": 0.12488611819140333,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3543014568804867e-05,
      "loss": 0.8330278396606445,
      "step": 1525
    },
    {
      "epoch": 0.12496801072792228,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.35423304887384e-05,
      "loss": 0.6851657629013062,
      "step": 1526
    },
    {
      "epoch": 0.12504990326444124,
      "grad_norm": 0.166015625,
      "learning_rate": 2.354164590699471e-05,
      "loss": 1.025659203529358,
      "step": 1527
    },
    {
      "epoch": 0.12513179580096018,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3540960823603553e-05,
      "loss": 0.8657687306404114,
      "step": 1528
    },
    {
      "epoch": 0.12521368833747915,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3540275238594708e-05,
      "loss": 0.7867289781570435,
      "step": 1529
    },
    {
      "epoch": 0.12529558087399809,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.353958915199797e-05,
      "loss": 1.0214639902114868,
      "step": 1530
    },
    {
      "epoch": 0.12537747341051705,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3538902563843153e-05,
      "loss": 1.1388967037200928,
      "step": 1531
    },
    {
      "epoch": 0.125459365947036,
      "grad_norm": 0.169921875,
      "learning_rate": 2.353821547416011e-05,
      "loss": 0.6637215614318848,
      "step": 1532
    },
    {
      "epoch": 0.12554125848355496,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3537527882978698e-05,
      "loss": 0.7284050583839417,
      "step": 1533
    },
    {
      "epoch": 0.1256231510200739,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.353683979032881e-05,
      "loss": 0.6624422073364258,
      "step": 1534
    },
    {
      "epoch": 0.12570504355659287,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3536151196240342e-05,
      "loss": 1.0671190023422241,
      "step": 1535
    },
    {
      "epoch": 0.1257869360931118,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.3535462100743234e-05,
      "loss": 0.7406049370765686,
      "step": 1536
    },
    {
      "epoch": 0.12586882862963078,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.353477250386743e-05,
      "loss": 0.8896205425262451,
      "step": 1537
    },
    {
      "epoch": 0.12595072116614972,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3534082405642902e-05,
      "loss": 0.6648122072219849,
      "step": 1538
    },
    {
      "epoch": 0.12603261370266866,
      "grad_norm": 0.189453125,
      "learning_rate": 2.353339180609965e-05,
      "loss": 1.1986068487167358,
      "step": 1539
    },
    {
      "epoch": 0.12611450623918763,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3532700705267685e-05,
      "loss": 0.8703520894050598,
      "step": 1540
    },
    {
      "epoch": 0.12619639877570657,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3532009103177052e-05,
      "loss": 0.8013082146644592,
      "step": 1541
    },
    {
      "epoch": 0.12627829131222554,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.35313169998578e-05,
      "loss": 0.6455479860305786,
      "step": 1542
    },
    {
      "epoch": 0.12636018384874448,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3530624395340025e-05,
      "loss": 1.1092280149459839,
      "step": 1543
    },
    {
      "epoch": 0.12644207638526345,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3529931289653817e-05,
      "loss": 0.8820127248764038,
      "step": 1544
    },
    {
      "epoch": 0.1265239689217824,
      "grad_norm": 0.15625,
      "learning_rate": 2.3529237682829308e-05,
      "loss": 0.7131547927856445,
      "step": 1545
    },
    {
      "epoch": 0.12660586145830136,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3528543574896645e-05,
      "loss": 0.7285799980163574,
      "step": 1546
    },
    {
      "epoch": 0.1266877539948203,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3527848965885998e-05,
      "loss": 0.8894945383071899,
      "step": 1547
    },
    {
      "epoch": 0.12676964653133924,
      "grad_norm": 0.142578125,
      "learning_rate": 2.352715385582755e-05,
      "loss": 0.6435328722000122,
      "step": 1548
    },
    {
      "epoch": 0.1268515390678582,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3526458244751523e-05,
      "loss": 0.759693443775177,
      "step": 1549
    },
    {
      "epoch": 0.12693343160437714,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3525762132688145e-05,
      "loss": 1.003343939781189,
      "step": 1550
    },
    {
      "epoch": 0.1270153241408961,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.352506551966767e-05,
      "loss": 0.7220286130905151,
      "step": 1551
    },
    {
      "epoch": 0.12709721667741505,
      "grad_norm": 0.232421875,
      "learning_rate": 2.3524368405720383e-05,
      "loss": 0.8319891691207886,
      "step": 1552
    },
    {
      "epoch": 0.12717910921393402,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.352367079087658e-05,
      "loss": 0.8849310874938965,
      "step": 1553
    },
    {
      "epoch": 0.12726100175045296,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3522972675166583e-05,
      "loss": 0.9899094700813293,
      "step": 1554
    },
    {
      "epoch": 0.12734289428697193,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3522274058620733e-05,
      "loss": 1.2100054025650024,
      "step": 1555
    },
    {
      "epoch": 0.12742478682349087,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3521574941269395e-05,
      "loss": 0.9192343354225159,
      "step": 1556
    },
    {
      "epoch": 0.12750667936000984,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3520875323142956e-05,
      "loss": 0.6621610522270203,
      "step": 1557
    },
    {
      "epoch": 0.12758857189652878,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.352017520427182e-05,
      "loss": 0.8334894776344299,
      "step": 1558
    },
    {
      "epoch": 0.12767046443304772,
      "grad_norm": 0.18359375,
      "learning_rate": 2.351947458468643e-05,
      "loss": 1.0577324628829956,
      "step": 1559
    },
    {
      "epoch": 0.1277523569695667,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3518773464417223e-05,
      "loss": 0.8690276145935059,
      "step": 1560
    },
    {
      "epoch": 0.12783424950608563,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3518071843494682e-05,
      "loss": 1.1270685195922852,
      "step": 1561
    },
    {
      "epoch": 0.1279161420426046,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3517369721949296e-05,
      "loss": 1.227945327758789,
      "step": 1562
    },
    {
      "epoch": 0.12799803457912354,
      "grad_norm": 0.17578125,
      "learning_rate": 2.351666709981159e-05,
      "loss": 0.7913717031478882,
      "step": 1563
    },
    {
      "epoch": 0.1280799271156425,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3515963977112097e-05,
      "loss": 0.8744975328445435,
      "step": 1564
    },
    {
      "epoch": 0.12816181965216145,
      "grad_norm": 0.212890625,
      "learning_rate": 2.3515260353881376e-05,
      "loss": 0.8258058428764343,
      "step": 1565
    },
    {
      "epoch": 0.12824371218868041,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3514556230150012e-05,
      "loss": 0.7909632325172424,
      "step": 1566
    },
    {
      "epoch": 0.12832560472519935,
      "grad_norm": 0.16796875,
      "learning_rate": 2.351385160594861e-05,
      "loss": 0.8148907423019409,
      "step": 1567
    },
    {
      "epoch": 0.12840749726171832,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.35131464813078e-05,
      "loss": 0.9934531450271606,
      "step": 1568
    },
    {
      "epoch": 0.12848938979823726,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3512440856258222e-05,
      "loss": 0.8845478892326355,
      "step": 1569
    },
    {
      "epoch": 0.1285712823347562,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3511734730830544e-05,
      "loss": 0.9996358752250671,
      "step": 1570
    },
    {
      "epoch": 0.12865317487127517,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3511028105055463e-05,
      "loss": 0.890006422996521,
      "step": 1571
    },
    {
      "epoch": 0.1287350674077941,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3510320978963694e-05,
      "loss": 0.8518478870391846,
      "step": 1572
    },
    {
      "epoch": 0.12881695994431308,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3509613352585963e-05,
      "loss": 1.0976216793060303,
      "step": 1573
    },
    {
      "epoch": 0.12889885248083202,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3508905225953037e-05,
      "loss": 0.6429983973503113,
      "step": 1574
    },
    {
      "epoch": 0.128980745017351,
      "grad_norm": 0.271484375,
      "learning_rate": 2.3508196599095682e-05,
      "loss": 0.8002221584320068,
      "step": 1575
    },
    {
      "epoch": 0.12906263755386993,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3507487472044705e-05,
      "loss": 0.7020823359489441,
      "step": 1576
    },
    {
      "epoch": 0.1291445300903889,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.350677784483093e-05,
      "loss": 0.9947074055671692,
      "step": 1577
    },
    {
      "epoch": 0.12922642262690784,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3506067717485192e-05,
      "loss": 0.9710113406181335,
      "step": 1578
    },
    {
      "epoch": 0.1293083151634268,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3505357090038365e-05,
      "loss": 0.39010822772979736,
      "step": 1579
    },
    {
      "epoch": 0.12939020769994575,
      "grad_norm": 0.193359375,
      "learning_rate": 2.350464596252133e-05,
      "loss": 1.1331063508987427,
      "step": 1580
    },
    {
      "epoch": 0.1294721002364647,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3503934334964996e-05,
      "loss": 1.0230588912963867,
      "step": 1581
    },
    {
      "epoch": 0.12955399277298366,
      "grad_norm": 0.1875,
      "learning_rate": 2.35032222074003e-05,
      "loss": 0.8720943331718445,
      "step": 1582
    },
    {
      "epoch": 0.1296358853095026,
      "grad_norm": 0.203125,
      "learning_rate": 2.3502509579858185e-05,
      "loss": 0.8469151258468628,
      "step": 1583
    },
    {
      "epoch": 0.12971777784602156,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3501796452369626e-05,
      "loss": 0.7596884369850159,
      "step": 1584
    },
    {
      "epoch": 0.1297996703825405,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3501082824965627e-05,
      "loss": 0.9370070695877075,
      "step": 1585
    },
    {
      "epoch": 0.12988156291905947,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.350036869767719e-05,
      "loss": 0.9339257478713989,
      "step": 1586
    },
    {
      "epoch": 0.1299634554555784,
      "grad_norm": 0.181640625,
      "learning_rate": 2.349965407053537e-05,
      "loss": 0.961476743221283,
      "step": 1587
    },
    {
      "epoch": 0.13004534799209738,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.349893894357122e-05,
      "loss": 0.7527668476104736,
      "step": 1588
    },
    {
      "epoch": 0.13012724052861632,
      "grad_norm": 0.203125,
      "learning_rate": 2.3498223316815822e-05,
      "loss": 1.0761133432388306,
      "step": 1589
    },
    {
      "epoch": 0.13020913306513526,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.349750719030028e-05,
      "loss": 1.0514628887176514,
      "step": 1590
    },
    {
      "epoch": 0.13029102560165423,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.349679056405572e-05,
      "loss": 0.8418586850166321,
      "step": 1591
    },
    {
      "epoch": 0.13037291813817317,
      "grad_norm": 0.177734375,
      "learning_rate": 2.349607343811329e-05,
      "loss": 1.0045831203460693,
      "step": 1592
    },
    {
      "epoch": 0.13045481067469214,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.349535581250416e-05,
      "loss": 0.7268948554992676,
      "step": 1593
    },
    {
      "epoch": 0.13053670321121108,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3494637687259526e-05,
      "loss": 1.1282360553741455,
      "step": 1594
    },
    {
      "epoch": 0.13061859574773005,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3493919062410588e-05,
      "loss": 0.6756092309951782,
      "step": 1595
    },
    {
      "epoch": 0.130700488284249,
      "grad_norm": 0.181640625,
      "learning_rate": 2.349319993798859e-05,
      "loss": 0.9167482852935791,
      "step": 1596
    },
    {
      "epoch": 0.13078238082076796,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3492480314024785e-05,
      "loss": 0.6549324989318848,
      "step": 1597
    },
    {
      "epoch": 0.1308642733572869,
      "grad_norm": 0.1875,
      "learning_rate": 2.349176019055045e-05,
      "loss": 1.1990950107574463,
      "step": 1598
    },
    {
      "epoch": 0.13094616589380587,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3491039567596892e-05,
      "loss": 1.0907965898513794,
      "step": 1599
    },
    {
      "epoch": 0.1310280584303248,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.349031844519542e-05,
      "loss": 0.7681134939193726,
      "step": 1600
    },
    {
      "epoch": 0.13110995096684375,
      "grad_norm": 0.150390625,
      "learning_rate": 2.348959682337739e-05,
      "loss": 0.991611123085022,
      "step": 1601
    },
    {
      "epoch": 0.13119184350336272,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3488874702174152e-05,
      "loss": 0.7560036182403564,
      "step": 1602
    },
    {
      "epoch": 0.13127373603988166,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3488152081617105e-05,
      "loss": 0.6427559852600098,
      "step": 1603
    },
    {
      "epoch": 0.13135562857640062,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3487428961737652e-05,
      "loss": 0.9197355508804321,
      "step": 1604
    },
    {
      "epoch": 0.13143752111291956,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3486705342567222e-05,
      "loss": 1.0656557083129883,
      "step": 1605
    },
    {
      "epoch": 0.13151941364943853,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3485981224137267e-05,
      "loss": 0.7695053219795227,
      "step": 1606
    },
    {
      "epoch": 0.13160130618595747,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3485256606479263e-05,
      "loss": 0.8690658211708069,
      "step": 1607
    },
    {
      "epoch": 0.13168319872247644,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3484531489624698e-05,
      "loss": 0.9840766787528992,
      "step": 1608
    },
    {
      "epoch": 0.13176509125899538,
      "grad_norm": 0.169921875,
      "learning_rate": 2.34838058736051e-05,
      "loss": 0.7369621992111206,
      "step": 1609
    },
    {
      "epoch": 0.13184698379551435,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3483079758451995e-05,
      "loss": 0.7512542009353638,
      "step": 1610
    },
    {
      "epoch": 0.1319288763320333,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.348235314419695e-05,
      "loss": 0.789833128452301,
      "step": 1611
    },
    {
      "epoch": 0.13201076886855223,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3481626030871546e-05,
      "loss": 0.7180898189544678,
      "step": 1612
    },
    {
      "epoch": 0.1320926614050712,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3480898418507387e-05,
      "loss": 0.640289306640625,
      "step": 1613
    },
    {
      "epoch": 0.13217455394159014,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.348017030713609e-05,
      "loss": 0.8316268920898438,
      "step": 1614
    },
    {
      "epoch": 0.1322564464781091,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3479441696789312e-05,
      "loss": 0.8022192120552063,
      "step": 1615
    },
    {
      "epoch": 0.13233833901462805,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3478712587498717e-05,
      "loss": 0.8319164514541626,
      "step": 1616
    },
    {
      "epoch": 0.13242023155114702,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3477982979296e-05,
      "loss": 0.8941062092781067,
      "step": 1617
    },
    {
      "epoch": 0.13250212408766596,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.347725287221286e-05,
      "loss": 0.8899974226951599,
      "step": 1618
    },
    {
      "epoch": 0.13258401662418492,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3476522266281044e-05,
      "loss": 0.8108752965927124,
      "step": 1619
    },
    {
      "epoch": 0.13266590916070387,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3475791161532302e-05,
      "loss": 1.1495907306671143,
      "step": 1620
    },
    {
      "epoch": 0.1327478016972228,
      "grad_norm": 0.18359375,
      "learning_rate": 2.347505955799841e-05,
      "loss": 0.7595223784446716,
      "step": 1621
    },
    {
      "epoch": 0.13282969423374177,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3474327455711166e-05,
      "loss": 1.1076464653015137,
      "step": 1622
    },
    {
      "epoch": 0.13291158677026071,
      "grad_norm": 0.310546875,
      "learning_rate": 2.3473594854702392e-05,
      "loss": 1.1217509508132935,
      "step": 1623
    },
    {
      "epoch": 0.13299347930677968,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3472861755003932e-05,
      "loss": 0.9420228004455566,
      "step": 1624
    },
    {
      "epoch": 0.13307537184329862,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3472128156647646e-05,
      "loss": 0.9235535860061646,
      "step": 1625
    },
    {
      "epoch": 0.1331572643798176,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3471394059665418e-05,
      "loss": 0.6970453262329102,
      "step": 1626
    },
    {
      "epoch": 0.13323915691633653,
      "grad_norm": 0.177734375,
      "learning_rate": 2.347065946408916e-05,
      "loss": 0.8882928490638733,
      "step": 1627
    },
    {
      "epoch": 0.1333210494528555,
      "grad_norm": 0.2353515625,
      "learning_rate": 2.3469924369950794e-05,
      "loss": 1.0362622737884521,
      "step": 1628
    },
    {
      "epoch": 0.13340294198937444,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3469188777282274e-05,
      "loss": 1.0224241018295288,
      "step": 1629
    },
    {
      "epoch": 0.1334848345258934,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3468452686115575e-05,
      "loss": 0.7838633060455322,
      "step": 1630
    },
    {
      "epoch": 0.13356672706241235,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3467716096482683e-05,
      "loss": 1.1424044370651245,
      "step": 1631
    },
    {
      "epoch": 0.1336486195989313,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3466979008415618e-05,
      "loss": 0.9850019812583923,
      "step": 1632
    },
    {
      "epoch": 0.13373051213545026,
      "grad_norm": 0.24609375,
      "learning_rate": 2.346624142194642e-05,
      "loss": 0.9179469347000122,
      "step": 1633
    },
    {
      "epoch": 0.1338124046719692,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3465503337107142e-05,
      "loss": 0.7566598057746887,
      "step": 1634
    },
    {
      "epoch": 0.13389429720848817,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3464764753929865e-05,
      "loss": 1.0667723417282104,
      "step": 1635
    },
    {
      "epoch": 0.1339761897450071,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3464025672446693e-05,
      "loss": 0.8108075857162476,
      "step": 1636
    },
    {
      "epoch": 0.13405808228152608,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3463286092689746e-05,
      "loss": 0.9261921644210815,
      "step": 1637
    },
    {
      "epoch": 0.13413997481804502,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3462546014691177e-05,
      "loss": 0.6809033751487732,
      "step": 1638
    },
    {
      "epoch": 0.13422186735456398,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3461805438483143e-05,
      "loss": 1.0155786275863647,
      "step": 1639
    },
    {
      "epoch": 0.13430375989108292,
      "grad_norm": 0.396484375,
      "learning_rate": 2.3461064364097837e-05,
      "loss": 0.8724837303161621,
      "step": 1640
    },
    {
      "epoch": 0.1343856524276019,
      "grad_norm": 0.2314453125,
      "learning_rate": 2.346032279156747e-05,
      "loss": 1.0166274309158325,
      "step": 1641
    },
    {
      "epoch": 0.13446754496412083,
      "grad_norm": 0.171875,
      "learning_rate": 2.3459580720924273e-05,
      "loss": 0.9664424061775208,
      "step": 1642
    },
    {
      "epoch": 0.13454943750063977,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3458838152200498e-05,
      "loss": 0.9100191593170166,
      "step": 1643
    },
    {
      "epoch": 0.13463133003715874,
      "grad_norm": 0.142578125,
      "learning_rate": 2.3458095085428427e-05,
      "loss": 0.658198356628418,
      "step": 1644
    },
    {
      "epoch": 0.13471322257367768,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3457351520640344e-05,
      "loss": 1.0236546993255615,
      "step": 1645
    },
    {
      "epoch": 0.13479511511019665,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3456607457868577e-05,
      "loss": 0.6518593430519104,
      "step": 1646
    },
    {
      "epoch": 0.1348770076467156,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3455862897145466e-05,
      "loss": 0.6210149526596069,
      "step": 1647
    },
    {
      "epoch": 0.13495890018323456,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3455117838503368e-05,
      "loss": 0.9304356575012207,
      "step": 1648
    },
    {
      "epoch": 0.1350407927197535,
      "grad_norm": 0.216796875,
      "learning_rate": 2.3454372281974666e-05,
      "loss": 0.9017465710639954,
      "step": 1649
    },
    {
      "epoch": 0.13512268525627247,
      "grad_norm": 0.201171875,
      "learning_rate": 2.345362622759177e-05,
      "loss": 0.961761474609375,
      "step": 1650
    },
    {
      "epoch": 0.1352045777927914,
      "grad_norm": 0.205078125,
      "learning_rate": 2.3452879675387107e-05,
      "loss": 0.9097169637680054,
      "step": 1651
    },
    {
      "epoch": 0.13528647032931035,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.345213262539312e-05,
      "loss": 0.7993609309196472,
      "step": 1652
    },
    {
      "epoch": 0.13536836286582932,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3451385077642282e-05,
      "loss": 0.933063268661499,
      "step": 1653
    },
    {
      "epoch": 0.13545025540234826,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.345063703216708e-05,
      "loss": 0.8823580741882324,
      "step": 1654
    },
    {
      "epoch": 0.13553214793886723,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3449888489000036e-05,
      "loss": 0.7844700813293457,
      "step": 1655
    },
    {
      "epoch": 0.13561404047538617,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3449139448173673e-05,
      "loss": 0.8455648422241211,
      "step": 1656
    },
    {
      "epoch": 0.13569593301190513,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3448389909720562e-05,
      "loss": 0.8499159812927246,
      "step": 1657
    },
    {
      "epoch": 0.13577782554842407,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.3447639873673267e-05,
      "loss": 1.277330994606018,
      "step": 1658
    },
    {
      "epoch": 0.13585971808494304,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3446889340064396e-05,
      "loss": 0.808635950088501,
      "step": 1659
    },
    {
      "epoch": 0.13594161062146198,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3446138308926566e-05,
      "loss": 1.0380194187164307,
      "step": 1660
    },
    {
      "epoch": 0.13602350315798095,
      "grad_norm": 0.177734375,
      "learning_rate": 2.344538678029243e-05,
      "loss": 1.3031929731369019,
      "step": 1661
    },
    {
      "epoch": 0.1361053956944999,
      "grad_norm": 0.15625,
      "learning_rate": 2.344463475419463e-05,
      "loss": 0.9188863039016724,
      "step": 1662
    },
    {
      "epoch": 0.13618728823101883,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3443882230665877e-05,
      "loss": 0.836012065410614,
      "step": 1663
    },
    {
      "epoch": 0.1362691807675378,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3443129209738868e-05,
      "loss": 1.0730888843536377,
      "step": 1664
    },
    {
      "epoch": 0.13635107330405674,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3442375691446333e-05,
      "loss": 0.7292597889900208,
      "step": 1665
    },
    {
      "epoch": 0.1364329658405757,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.344162167582102e-05,
      "loss": 0.8135591745376587,
      "step": 1666
    },
    {
      "epoch": 0.13651485837709465,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3440867162895706e-05,
      "loss": 0.8758120536804199,
      "step": 1667
    },
    {
      "epoch": 0.13659675091361362,
      "grad_norm": 0.193359375,
      "learning_rate": 2.344011215270318e-05,
      "loss": 1.052804946899414,
      "step": 1668
    },
    {
      "epoch": 0.13667864345013256,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3439356645276266e-05,
      "loss": 0.7884631156921387,
      "step": 1669
    },
    {
      "epoch": 0.13676053598665153,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3438600640647796e-05,
      "loss": 0.9983245134353638,
      "step": 1670
    },
    {
      "epoch": 0.13684242852317047,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3437844138850626e-05,
      "loss": 0.9895610213279724,
      "step": 1671
    },
    {
      "epoch": 0.13692432105968944,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3437087139917646e-05,
      "loss": 1.0172215700149536,
      "step": 1672
    },
    {
      "epoch": 0.13700621359620838,
      "grad_norm": 0.177734375,
      "learning_rate": 2.343632964388175e-05,
      "loss": 0.8625044226646423,
      "step": 1673
    },
    {
      "epoch": 0.13708810613272732,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3435571650775868e-05,
      "loss": 0.934339702129364,
      "step": 1674
    },
    {
      "epoch": 0.13716999866924628,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.343481316063294e-05,
      "loss": 0.6338703036308289,
      "step": 1675
    },
    {
      "epoch": 0.13725189120576523,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3434054173485935e-05,
      "loss": 0.9846989512443542,
      "step": 1676
    },
    {
      "epoch": 0.1373337837422842,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3433294689367846e-05,
      "loss": 0.834060788154602,
      "step": 1677
    },
    {
      "epoch": 0.13741567627880313,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3432534708311674e-05,
      "loss": 0.9962806701660156,
      "step": 1678
    },
    {
      "epoch": 0.1374975688153221,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3431774230350458e-05,
      "loss": 0.7456172108650208,
      "step": 1679
    },
    {
      "epoch": 0.13757946135184104,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3431013255517252e-05,
      "loss": 1.105972170829773,
      "step": 1680
    },
    {
      "epoch": 0.13766135388836,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.343025178384513e-05,
      "loss": 0.9536287188529968,
      "step": 1681
    },
    {
      "epoch": 0.13774324642487895,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3429489815367183e-05,
      "loss": 0.8721716403961182,
      "step": 1682
    },
    {
      "epoch": 0.13782513896139792,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3428727350116534e-05,
      "loss": 0.9927589893341064,
      "step": 1683
    },
    {
      "epoch": 0.13790703149791686,
      "grad_norm": 0.146484375,
      "learning_rate": 2.342796438812633e-05,
      "loss": 0.7557966709136963,
      "step": 1684
    },
    {
      "epoch": 0.1379889240344358,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.342720092942972e-05,
      "loss": 0.960281491279602,
      "step": 1685
    },
    {
      "epoch": 0.13807081657095477,
      "grad_norm": 0.23828125,
      "learning_rate": 2.3426436974059895e-05,
      "loss": 1.0987720489501953,
      "step": 1686
    },
    {
      "epoch": 0.1381527091074737,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3425672522050056e-05,
      "loss": 0.8334921598434448,
      "step": 1687
    },
    {
      "epoch": 0.13823460164399268,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3424907573433427e-05,
      "loss": 0.7242124676704407,
      "step": 1688
    },
    {
      "epoch": 0.13831649418051162,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.342414212824326e-05,
      "loss": 0.5259591341018677,
      "step": 1689
    },
    {
      "epoch": 0.13839838671703059,
      "grad_norm": 0.13671875,
      "learning_rate": 2.3423376186512827e-05,
      "loss": 0.6272844672203064,
      "step": 1690
    },
    {
      "epoch": 0.13848027925354953,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3422609748275414e-05,
      "loss": 0.8940479755401611,
      "step": 1691
    },
    {
      "epoch": 0.1385621717900685,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3421842813564336e-05,
      "loss": 0.943976104259491,
      "step": 1692
    },
    {
      "epoch": 0.13864406432658744,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3421075382412923e-05,
      "loss": 0.7209089398384094,
      "step": 1693
    },
    {
      "epoch": 0.13872595686310638,
      "grad_norm": 0.1875,
      "learning_rate": 2.3420307454854537e-05,
      "loss": 0.6116722822189331,
      "step": 1694
    },
    {
      "epoch": 0.13880784939962534,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3419539030922552e-05,
      "loss": 0.6617465615272522,
      "step": 1695
    },
    {
      "epoch": 0.13888974193614428,
      "grad_norm": 0.16015625,
      "learning_rate": 2.341877011065037e-05,
      "loss": 0.6561160683631897,
      "step": 1696
    },
    {
      "epoch": 0.13897163447266325,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3418000694071403e-05,
      "loss": 0.8326876759529114,
      "step": 1697
    },
    {
      "epoch": 0.1390535270091822,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3417230781219096e-05,
      "loss": 1.0738868713378906,
      "step": 1698
    },
    {
      "epoch": 0.13913541954570116,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3416460372126923e-05,
      "loss": 0.8097984194755554,
      "step": 1699
    },
    {
      "epoch": 0.1392173120822201,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3415689466828355e-05,
      "loss": 0.8695337772369385,
      "step": 1700
    },
    {
      "epoch": 0.13929920461873907,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3414918065356907e-05,
      "loss": 0.8281451463699341,
      "step": 1701
    },
    {
      "epoch": 0.139381097155258,
      "grad_norm": 0.1640625,
      "learning_rate": 2.341414616774611e-05,
      "loss": 0.8183345198631287,
      "step": 1702
    },
    {
      "epoch": 0.13946298969177698,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3413373774029503e-05,
      "loss": 0.7993346452713013,
      "step": 1703
    },
    {
      "epoch": 0.13954488222829592,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3412600884240665e-05,
      "loss": 0.8513830304145813,
      "step": 1704
    },
    {
      "epoch": 0.13962677476481486,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.341182749841319e-05,
      "loss": 0.8494141101837158,
      "step": 1705
    },
    {
      "epoch": 0.13970866730133383,
      "grad_norm": 0.18359375,
      "learning_rate": 2.341105361658069e-05,
      "loss": 0.7807350158691406,
      "step": 1706
    },
    {
      "epoch": 0.13979055983785277,
      "grad_norm": 0.212890625,
      "learning_rate": 2.34102792387768e-05,
      "loss": 1.30607271194458,
      "step": 1707
    },
    {
      "epoch": 0.13987245237437174,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3409504365035175e-05,
      "loss": 0.9800342917442322,
      "step": 1708
    },
    {
      "epoch": 0.13995434491089068,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3408728995389502e-05,
      "loss": 0.996315598487854,
      "step": 1709
    },
    {
      "epoch": 0.14003623744740964,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3407953129873476e-05,
      "loss": 0.7794680595397949,
      "step": 1710
    },
    {
      "epoch": 0.14011812998392859,
      "grad_norm": 0.205078125,
      "learning_rate": 2.3407176768520826e-05,
      "loss": 0.7499452829360962,
      "step": 1711
    },
    {
      "epoch": 0.14020002252044755,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3406399911365283e-05,
      "loss": 0.5939576625823975,
      "step": 1712
    },
    {
      "epoch": 0.1402819150569665,
      "grad_norm": 0.171875,
      "learning_rate": 2.3405622558440628e-05,
      "loss": 0.9416583776473999,
      "step": 1713
    },
    {
      "epoch": 0.14036380759348546,
      "grad_norm": 0.1875,
      "learning_rate": 2.3404844709780633e-05,
      "loss": 0.7452306747436523,
      "step": 1714
    },
    {
      "epoch": 0.1404457001300044,
      "grad_norm": 0.158203125,
      "learning_rate": 2.340406636541912e-05,
      "loss": 0.5363136529922485,
      "step": 1715
    },
    {
      "epoch": 0.14052759266652334,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.340328752538991e-05,
      "loss": 0.7687567472457886,
      "step": 1716
    },
    {
      "epoch": 0.1406094852030423,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3402508189726856e-05,
      "loss": 0.9629082679748535,
      "step": 1717
    },
    {
      "epoch": 0.14069137773956125,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3401728358463835e-05,
      "loss": 0.8481153845787048,
      "step": 1718
    },
    {
      "epoch": 0.14077327027608022,
      "grad_norm": 0.14453125,
      "learning_rate": 2.340094803163474e-05,
      "loss": 0.6960976123809814,
      "step": 1719
    },
    {
      "epoch": 0.14085516281259916,
      "grad_norm": 0.1875,
      "learning_rate": 2.3400167209273486e-05,
      "loss": 0.7860400676727295,
      "step": 1720
    },
    {
      "epoch": 0.14093705534911813,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3399385891414008e-05,
      "loss": 0.9449538588523865,
      "step": 1721
    },
    {
      "epoch": 0.14101894788563707,
      "grad_norm": 0.1484375,
      "learning_rate": 2.339860407809027e-05,
      "loss": 0.8932732939720154,
      "step": 1722
    },
    {
      "epoch": 0.14110084042215604,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.3397821769336254e-05,
      "loss": 0.7036584615707397,
      "step": 1723
    },
    {
      "epoch": 0.14118273295867498,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3397038965185956e-05,
      "loss": 0.7544406056404114,
      "step": 1724
    },
    {
      "epoch": 0.14126462549519392,
      "grad_norm": 0.177734375,
      "learning_rate": 2.339625566567341e-05,
      "loss": 0.8169337511062622,
      "step": 1725
    },
    {
      "epoch": 0.1413465180317129,
      "grad_norm": 0.142578125,
      "learning_rate": 2.339547187083265e-05,
      "loss": 0.8005601763725281,
      "step": 1726
    },
    {
      "epoch": 0.14142841056823183,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.3394687580697748e-05,
      "loss": 1.0061776638031006,
      "step": 1727
    },
    {
      "epoch": 0.1415103031047508,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.339390279530279e-05,
      "loss": 1.0360643863677979,
      "step": 1728
    },
    {
      "epoch": 0.14159219564126974,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3393117514681896e-05,
      "loss": 0.5064568519592285,
      "step": 1729
    },
    {
      "epoch": 0.1416740881777887,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.3392331738869182e-05,
      "loss": 1.1183488368988037,
      "step": 1730
    },
    {
      "epoch": 0.14175598071430764,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3391545467898816e-05,
      "loss": 0.9721965193748474,
      "step": 1731
    },
    {
      "epoch": 0.1418378732508266,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3390758701804962e-05,
      "loss": 0.9406471252441406,
      "step": 1732
    },
    {
      "epoch": 0.14191976578734555,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3389971440621815e-05,
      "loss": 0.7524545788764954,
      "step": 1733
    },
    {
      "epoch": 0.14200165832386452,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3389183684383604e-05,
      "loss": 0.8204085826873779,
      "step": 1734
    },
    {
      "epoch": 0.14208355086038346,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3388395433124562e-05,
      "loss": 0.5364327430725098,
      "step": 1735
    },
    {
      "epoch": 0.1421654433969024,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3387606686878945e-05,
      "loss": 0.8509069085121155,
      "step": 1736
    },
    {
      "epoch": 0.14224733593342137,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3386817445681043e-05,
      "loss": 0.49502792954444885,
      "step": 1737
    },
    {
      "epoch": 0.1423292284699403,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3386027709565155e-05,
      "loss": 0.9358277916908264,
      "step": 1738
    },
    {
      "epoch": 0.14241112100645928,
      "grad_norm": 0.15625,
      "learning_rate": 2.3385237478565605e-05,
      "loss": 0.7964696884155273,
      "step": 1739
    },
    {
      "epoch": 0.14249301354297822,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3384446752716742e-05,
      "loss": 0.9974815845489502,
      "step": 1740
    },
    {
      "epoch": 0.1425749060794972,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3383655532052936e-05,
      "loss": 0.7631921768188477,
      "step": 1741
    },
    {
      "epoch": 0.14265679861601613,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.3382863816608576e-05,
      "loss": 0.6960386037826538,
      "step": 1742
    },
    {
      "epoch": 0.1427386911525351,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3382071606418065e-05,
      "loss": 0.689273476600647,
      "step": 1743
    },
    {
      "epoch": 0.14282058368905404,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.338127890151585e-05,
      "loss": 0.9112082123756409,
      "step": 1744
    },
    {
      "epoch": 0.142902476225573,
      "grad_norm": 0.203125,
      "learning_rate": 2.3380485701936376e-05,
      "loss": 1.1433637142181396,
      "step": 1745
    },
    {
      "epoch": 0.14298436876209195,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3379692007714123e-05,
      "loss": 1.0371371507644653,
      "step": 1746
    },
    {
      "epoch": 0.1430662612986109,
      "grad_norm": 0.162109375,
      "learning_rate": 2.337889781888358e-05,
      "loss": 1.0123802423477173,
      "step": 1747
    },
    {
      "epoch": 0.14314815383512985,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3378103135479275e-05,
      "loss": 1.2243375778198242,
      "step": 1748
    },
    {
      "epoch": 0.1432300463716488,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3377307957535744e-05,
      "loss": 0.7974395751953125,
      "step": 1749
    },
    {
      "epoch": 0.14331193890816776,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.337651228508755e-05,
      "loss": 0.8220515251159668,
      "step": 1750
    },
    {
      "epoch": 0.1433938314446867,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3375716118169276e-05,
      "loss": 0.5945915579795837,
      "step": 1751
    },
    {
      "epoch": 0.14347572398120567,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3374919456815525e-05,
      "loss": 1.0424771308898926,
      "step": 1752
    },
    {
      "epoch": 0.1435576165177246,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3374122301060924e-05,
      "loss": 0.6379466652870178,
      "step": 1753
    },
    {
      "epoch": 0.14363950905424358,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.337332465094012e-05,
      "loss": 0.9607183933258057,
      "step": 1754
    },
    {
      "epoch": 0.14372140159076252,
      "grad_norm": 0.16796875,
      "learning_rate": 2.337252650648779e-05,
      "loss": 0.9077478647232056,
      "step": 1755
    },
    {
      "epoch": 0.14380329412728146,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3371727867738612e-05,
      "loss": 0.8585506677627563,
      "step": 1756
    },
    {
      "epoch": 0.14388518666380043,
      "grad_norm": 0.171875,
      "learning_rate": 2.337092873472731e-05,
      "loss": 1.0332348346710205,
      "step": 1757
    },
    {
      "epoch": 0.14396707920031937,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3370129107488604e-05,
      "loss": 0.8863798975944519,
      "step": 1758
    },
    {
      "epoch": 0.14404897173683834,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.336932898605726e-05,
      "loss": 0.9455370903015137,
      "step": 1759
    },
    {
      "epoch": 0.14413086427335728,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3368528370468055e-05,
      "loss": 0.8861120343208313,
      "step": 1760
    },
    {
      "epoch": 0.14421275680987625,
      "grad_norm": 0.22265625,
      "learning_rate": 2.336772726075578e-05,
      "loss": 1.1002259254455566,
      "step": 1761
    },
    {
      "epoch": 0.1442946493463952,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3366925656955257e-05,
      "loss": 0.558000922203064,
      "step": 1762
    },
    {
      "epoch": 0.14437654188291416,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.336612355910133e-05,
      "loss": 0.7646887898445129,
      "step": 1763
    },
    {
      "epoch": 0.1444584344194331,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.336532096722886e-05,
      "loss": 1.1245468854904175,
      "step": 1764
    },
    {
      "epoch": 0.14454032695595206,
      "grad_norm": 0.197265625,
      "learning_rate": 2.336451788137273e-05,
      "loss": 1.103338599205017,
      "step": 1765
    },
    {
      "epoch": 0.144622219492471,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3363714301567846e-05,
      "loss": 0.978534996509552,
      "step": 1766
    },
    {
      "epoch": 0.14470411202898995,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3362910227849136e-05,
      "loss": 0.890990138053894,
      "step": 1767
    },
    {
      "epoch": 0.1447860045655089,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3362105660251545e-05,
      "loss": 0.8109412789344788,
      "step": 1768
    },
    {
      "epoch": 0.14486789710202785,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3361300598810043e-05,
      "loss": 0.8064419031143188,
      "step": 1769
    },
    {
      "epoch": 0.14494978963854682,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3360495043559625e-05,
      "loss": 0.5979679822921753,
      "step": 1770
    },
    {
      "epoch": 0.14503168217506576,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.3359688994535302e-05,
      "loss": 0.6205583214759827,
      "step": 1771
    },
    {
      "epoch": 0.14511357471158473,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.335888245177211e-05,
      "loss": 0.8708716630935669,
      "step": 1772
    },
    {
      "epoch": 0.14519546724810367,
      "grad_norm": 0.203125,
      "learning_rate": 2.33580754153051e-05,
      "loss": 0.9297593832015991,
      "step": 1773
    },
    {
      "epoch": 0.14527735978462264,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3357267885169356e-05,
      "loss": 0.9373528361320496,
      "step": 1774
    },
    {
      "epoch": 0.14535925232114158,
      "grad_norm": 0.181640625,
      "learning_rate": 2.335645986139997e-05,
      "loss": 0.5473214983940125,
      "step": 1775
    },
    {
      "epoch": 0.14544114485766055,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3355651344032064e-05,
      "loss": 0.8224920034408569,
      "step": 1776
    },
    {
      "epoch": 0.1455230373941795,
      "grad_norm": 0.1640625,
      "learning_rate": 2.335484233310078e-05,
      "loss": 0.539886474609375,
      "step": 1777
    },
    {
      "epoch": 0.14560492993069843,
      "grad_norm": 0.177734375,
      "learning_rate": 2.335403282864128e-05,
      "loss": 0.6082117557525635,
      "step": 1778
    },
    {
      "epoch": 0.1456868224672174,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3353222830688754e-05,
      "loss": 0.975936770439148,
      "step": 1779
    },
    {
      "epoch": 0.14576871500373634,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.33524123392784e-05,
      "loss": 0.626113772392273,
      "step": 1780
    },
    {
      "epoch": 0.1458506075402553,
      "grad_norm": 0.228515625,
      "learning_rate": 2.3351601354445448e-05,
      "loss": 0.8247703909873962,
      "step": 1781
    },
    {
      "epoch": 0.14593250007677425,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.335078987622515e-05,
      "loss": 0.652069628238678,
      "step": 1782
    },
    {
      "epoch": 0.14601439261329321,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.334997790465277e-05,
      "loss": 0.8391812443733215,
      "step": 1783
    },
    {
      "epoch": 0.14609628514981216,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3349165439763607e-05,
      "loss": 0.991180419921875,
      "step": 1784
    },
    {
      "epoch": 0.14617817768633112,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3348352481592966e-05,
      "loss": 0.6333591938018799,
      "step": 1785
    },
    {
      "epoch": 0.14626007022285006,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3347539030176185e-05,
      "loss": 1.0944880247116089,
      "step": 1786
    },
    {
      "epoch": 0.146341962759369,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3346725085548626e-05,
      "loss": 0.6372475624084473,
      "step": 1787
    },
    {
      "epoch": 0.14642385529588797,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3345910647745658e-05,
      "loss": 0.8103689551353455,
      "step": 1788
    },
    {
      "epoch": 0.1465057478324069,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3345095716802686e-05,
      "loss": 0.9788298606872559,
      "step": 1789
    },
    {
      "epoch": 0.14658764036892588,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.334428029275512e-05,
      "loss": 0.9469529390335083,
      "step": 1790
    },
    {
      "epoch": 0.14666953290544482,
      "grad_norm": 0.1953125,
      "learning_rate": 2.3343464375638417e-05,
      "loss": 1.1397786140441895,
      "step": 1791
    },
    {
      "epoch": 0.1467514254419638,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.334264796548803e-05,
      "loss": 0.8082143068313599,
      "step": 1792
    },
    {
      "epoch": 0.14683331797848273,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3341831062339442e-05,
      "loss": 1.1135767698287964,
      "step": 1793
    },
    {
      "epoch": 0.1469152105150017,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3341013666228165e-05,
      "loss": 0.7053612470626831,
      "step": 1794
    },
    {
      "epoch": 0.14699710305152064,
      "grad_norm": 0.150390625,
      "learning_rate": 2.334019577718972e-05,
      "loss": 0.8920732736587524,
      "step": 1795
    },
    {
      "epoch": 0.1470789955880396,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3339377395259664e-05,
      "loss": 0.9755126237869263,
      "step": 1796
    },
    {
      "epoch": 0.14716088812455855,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3338558520473562e-05,
      "loss": 0.8654794692993164,
      "step": 1797
    },
    {
      "epoch": 0.1472427806610775,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.333773915286701e-05,
      "loss": 0.8878124952316284,
      "step": 1798
    },
    {
      "epoch": 0.14732467319759646,
      "grad_norm": 0.33984375,
      "learning_rate": 2.333691929247561e-05,
      "loss": 0.8451465964317322,
      "step": 1799
    },
    {
      "epoch": 0.1474065657341154,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.333609893933501e-05,
      "loss": 0.9521805644035339,
      "step": 1800
    },
    {
      "epoch": 0.14748845827063436,
      "grad_norm": 0.1796875,
      "learning_rate": 2.333527809348086e-05,
      "loss": 0.8027971386909485,
      "step": 1801
    },
    {
      "epoch": 0.1475703508071533,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3334456754948836e-05,
      "loss": 0.8663756251335144,
      "step": 1802
    },
    {
      "epoch": 0.14765224334367227,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3333634923774636e-05,
      "loss": 0.9972551465034485,
      "step": 1803
    },
    {
      "epoch": 0.14773413588019121,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3332812599993986e-05,
      "loss": 1.2096611261367798,
      "step": 1804
    },
    {
      "epoch": 0.14781602841671018,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3331989783642624e-05,
      "loss": 1.4040031433105469,
      "step": 1805
    },
    {
      "epoch": 0.14789792095322912,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.333116647475631e-05,
      "loss": 0.7512882351875305,
      "step": 1806
    },
    {
      "epoch": 0.1479798134897481,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3330342673370835e-05,
      "loss": 0.9009363055229187,
      "step": 1807
    },
    {
      "epoch": 0.14806170602626703,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3329518379521996e-05,
      "loss": 0.9386667013168335,
      "step": 1808
    },
    {
      "epoch": 0.14814359856278597,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3328693593245627e-05,
      "loss": 0.7612395286560059,
      "step": 1809
    },
    {
      "epoch": 0.14822549109930494,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.3327868314577577e-05,
      "loss": 0.6654129028320312,
      "step": 1810
    },
    {
      "epoch": 0.14830738363582388,
      "grad_norm": 0.234375,
      "learning_rate": 2.3327042543553714e-05,
      "loss": 0.7936034798622131,
      "step": 1811
    },
    {
      "epoch": 0.14838927617234285,
      "grad_norm": 0.21484375,
      "learning_rate": 2.332621628020993e-05,
      "loss": 1.1812770366668701,
      "step": 1812
    },
    {
      "epoch": 0.1484711687088618,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3325389524582136e-05,
      "loss": 0.7320436239242554,
      "step": 1813
    },
    {
      "epoch": 0.14855306124538076,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3324562276706266e-05,
      "loss": 0.9512939453125,
      "step": 1814
    },
    {
      "epoch": 0.1486349537818997,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3323734536618277e-05,
      "loss": 0.6637625098228455,
      "step": 1815
    },
    {
      "epoch": 0.14871684631841867,
      "grad_norm": 0.193359375,
      "learning_rate": 2.332290630435415e-05,
      "loss": 0.9432358741760254,
      "step": 1816
    },
    {
      "epoch": 0.1487987388549376,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3322077579949875e-05,
      "loss": 0.852164089679718,
      "step": 1817
    },
    {
      "epoch": 0.14888063139145657,
      "grad_norm": 0.154296875,
      "learning_rate": 2.332124836344148e-05,
      "loss": 0.9127488136291504,
      "step": 1818
    },
    {
      "epoch": 0.14896252392797552,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3320418654865e-05,
      "loss": 0.9190155267715454,
      "step": 1819
    },
    {
      "epoch": 0.14904441646449446,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.33195884542565e-05,
      "loss": 0.6365010738372803,
      "step": 1820
    },
    {
      "epoch": 0.14912630900101342,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3318757761652064e-05,
      "loss": 1.0945080518722534,
      "step": 1821
    },
    {
      "epoch": 0.14920820153753236,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3317926577087796e-05,
      "loss": 0.8778930902481079,
      "step": 1822
    },
    {
      "epoch": 0.14929009407405133,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3317094900599823e-05,
      "loss": 0.8118739724159241,
      "step": 1823
    },
    {
      "epoch": 0.14937198661057027,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.3316262732224297e-05,
      "loss": 0.9301413893699646,
      "step": 1824
    },
    {
      "epoch": 0.14945387914708924,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3315430071997386e-05,
      "loss": 0.9562491774559021,
      "step": 1825
    },
    {
      "epoch": 0.14953577168360818,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3314596919955275e-05,
      "loss": 0.9944878816604614,
      "step": 1826
    },
    {
      "epoch": 0.14961766422012715,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3313763276134183e-05,
      "loss": 0.6779910922050476,
      "step": 1827
    },
    {
      "epoch": 0.1496995567566461,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3312929140570336e-05,
      "loss": 0.6554465293884277,
      "step": 1828
    },
    {
      "epoch": 0.14978144929316503,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3312094513299998e-05,
      "loss": 0.9440432190895081,
      "step": 1829
    },
    {
      "epoch": 0.149863341829684,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3311259394359444e-05,
      "loss": 0.8022931814193726,
      "step": 1830
    },
    {
      "epoch": 0.14994523436620294,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3310423783784966e-05,
      "loss": 0.9430667161941528,
      "step": 1831
    },
    {
      "epoch": 0.1500271269027219,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3309587681612887e-05,
      "loss": 0.9594816565513611,
      "step": 1832
    },
    {
      "epoch": 0.15010901943924085,
      "grad_norm": 0.1796875,
      "learning_rate": 2.330875108787955e-05,
      "loss": 0.988732099533081,
      "step": 1833
    },
    {
      "epoch": 0.15019091197575982,
      "grad_norm": 0.146484375,
      "learning_rate": 2.3307914002621305e-05,
      "loss": 0.845299243927002,
      "step": 1834
    },
    {
      "epoch": 0.15027280451227876,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3307076425874546e-05,
      "loss": 0.7803511023521423,
      "step": 1835
    },
    {
      "epoch": 0.15035469704879773,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.330623835767568e-05,
      "loss": 0.8433563113212585,
      "step": 1836
    },
    {
      "epoch": 0.15043658958531667,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3305399798061124e-05,
      "loss": 1.142472505569458,
      "step": 1837
    },
    {
      "epoch": 0.15051848212183563,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3304560747067333e-05,
      "loss": 0.8218322992324829,
      "step": 1838
    },
    {
      "epoch": 0.15060037465835457,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3303721204730766e-05,
      "loss": 0.9977097511291504,
      "step": 1839
    },
    {
      "epoch": 0.15068226719487351,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3302881171087924e-05,
      "loss": 0.8673079013824463,
      "step": 1840
    },
    {
      "epoch": 0.15076415973139248,
      "grad_norm": 0.1640625,
      "learning_rate": 2.330204064617531e-05,
      "loss": 0.9375929832458496,
      "step": 1841
    },
    {
      "epoch": 0.15084605226791142,
      "grad_norm": 0.16796875,
      "learning_rate": 2.330119963002946e-05,
      "loss": 0.613758385181427,
      "step": 1842
    },
    {
      "epoch": 0.1509279448044304,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3300358122686932e-05,
      "loss": 0.8373093008995056,
      "step": 1843
    },
    {
      "epoch": 0.15100983734094933,
      "grad_norm": 0.5390625,
      "learning_rate": 2.3299516124184293e-05,
      "loss": 0.7188383936882019,
      "step": 1844
    },
    {
      "epoch": 0.1510917298774683,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3298673634558145e-05,
      "loss": 0.8333261609077454,
      "step": 1845
    },
    {
      "epoch": 0.15117362241398724,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3297830653845103e-05,
      "loss": 0.7658025026321411,
      "step": 1846
    },
    {
      "epoch": 0.1512555149505062,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3296987182081813e-05,
      "loss": 0.8517516851425171,
      "step": 1847
    },
    {
      "epoch": 0.15133740748702515,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.3296143219304927e-05,
      "loss": 0.8477591872215271,
      "step": 1848
    },
    {
      "epoch": 0.15141930002354412,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3295298765551134e-05,
      "loss": 0.8994050025939941,
      "step": 1849
    },
    {
      "epoch": 0.15150119256006306,
      "grad_norm": 0.171875,
      "learning_rate": 2.3294453820857136e-05,
      "loss": 0.7842777371406555,
      "step": 1850
    },
    {
      "epoch": 0.151583085096582,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3293608385259653e-05,
      "loss": 0.8934636116027832,
      "step": 1851
    },
    {
      "epoch": 0.15166497763310097,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3292762458795438e-05,
      "loss": 0.7866275310516357,
      "step": 1852
    },
    {
      "epoch": 0.1517468701696199,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.329191604150125e-05,
      "loss": 0.9011574983596802,
      "step": 1853
    },
    {
      "epoch": 0.15182876270613888,
      "grad_norm": 0.15625,
      "learning_rate": 2.3291069133413886e-05,
      "loss": 0.8706033825874329,
      "step": 1854
    },
    {
      "epoch": 0.15191065524265782,
      "grad_norm": 0.21484375,
      "learning_rate": 2.329022173457015e-05,
      "loss": 1.3024297952651978,
      "step": 1855
    },
    {
      "epoch": 0.15199254777917678,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3289373845006882e-05,
      "loss": 0.9255737066268921,
      "step": 1856
    },
    {
      "epoch": 0.15207444031569572,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3288525464760928e-05,
      "loss": 0.7151395082473755,
      "step": 1857
    },
    {
      "epoch": 0.1521563328522147,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.328767659386916e-05,
      "loss": 0.8222572803497314,
      "step": 1858
    },
    {
      "epoch": 0.15223822538873363,
      "grad_norm": 0.173828125,
      "learning_rate": 2.328682723236848e-05,
      "loss": 0.912538468837738,
      "step": 1859
    },
    {
      "epoch": 0.15232011792525257,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.32859773802958e-05,
      "loss": 1.010779857635498,
      "step": 1860
    },
    {
      "epoch": 0.15240201046177154,
      "grad_norm": 0.162109375,
      "learning_rate": 2.328512703768806e-05,
      "loss": 0.8114819526672363,
      "step": 1861
    },
    {
      "epoch": 0.15248390299829048,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.328427620458222e-05,
      "loss": 0.6525090932846069,
      "step": 1862
    },
    {
      "epoch": 0.15256579553480945,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3283424881015256e-05,
      "loss": 1.1830549240112305,
      "step": 1863
    },
    {
      "epoch": 0.1526476880713284,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.328257306702417e-05,
      "loss": 0.8467474579811096,
      "step": 1864
    },
    {
      "epoch": 0.15272958060784736,
      "grad_norm": 0.212890625,
      "learning_rate": 2.3281720762646e-05,
      "loss": 1.3070322275161743,
      "step": 1865
    },
    {
      "epoch": 0.1528114731443663,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3280867967917775e-05,
      "loss": 0.8158149123191833,
      "step": 1866
    },
    {
      "epoch": 0.15289336568088527,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3280014682876562e-05,
      "loss": 1.0151132345199585,
      "step": 1867
    },
    {
      "epoch": 0.1529752582174042,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3279160907559458e-05,
      "loss": 0.8197196125984192,
      "step": 1868
    },
    {
      "epoch": 0.15305715075392318,
      "grad_norm": 0.134765625,
      "learning_rate": 2.327830664200356e-05,
      "loss": 0.6891666650772095,
      "step": 1869
    },
    {
      "epoch": 0.15313904329044212,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3277451886246007e-05,
      "loss": 0.8846073746681213,
      "step": 1870
    },
    {
      "epoch": 0.15322093582696106,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3276596640323944e-05,
      "loss": 0.8270590305328369,
      "step": 1871
    },
    {
      "epoch": 0.15330282836348003,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3275740904274547e-05,
      "loss": 0.9666038155555725,
      "step": 1872
    },
    {
      "epoch": 0.15338472089999897,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3274884678135007e-05,
      "loss": 0.5371145009994507,
      "step": 1873
    },
    {
      "epoch": 0.15346661343651793,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3274027961942546e-05,
      "loss": 0.48053568601608276,
      "step": 1874
    },
    {
      "epoch": 0.15354850597303688,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.327317075573439e-05,
      "loss": 0.8310407400131226,
      "step": 1875
    },
    {
      "epoch": 0.15363039850955584,
      "grad_norm": 0.185546875,
      "learning_rate": 2.3272313059547805e-05,
      "loss": 0.9808354377746582,
      "step": 1876
    },
    {
      "epoch": 0.15371229104607478,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.327145487342007e-05,
      "loss": 0.8298897743225098,
      "step": 1877
    },
    {
      "epoch": 0.15379418358259375,
      "grad_norm": 0.15625,
      "learning_rate": 2.327059619738848e-05,
      "loss": 0.7051181793212891,
      "step": 1878
    },
    {
      "epoch": 0.1538760761191127,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.3269737031490357e-05,
      "loss": 0.9141343832015991,
      "step": 1879
    },
    {
      "epoch": 0.15395796865563166,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3268877375763047e-05,
      "loss": 0.7681975364685059,
      "step": 1880
    },
    {
      "epoch": 0.1540398611921506,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3268017230243916e-05,
      "loss": 0.9187538027763367,
      "step": 1881
    },
    {
      "epoch": 0.15412175372866954,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3267156594970344e-05,
      "loss": 0.8955500721931458,
      "step": 1882
    },
    {
      "epoch": 0.1542036462651885,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.3266295469979742e-05,
      "loss": 0.7968420386314392,
      "step": 1883
    },
    {
      "epoch": 0.15428553880170745,
      "grad_norm": 0.13671875,
      "learning_rate": 2.3265433855309536e-05,
      "loss": 0.7414544820785522,
      "step": 1884
    },
    {
      "epoch": 0.15436743133822642,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3264571750997173e-05,
      "loss": 0.8185932636260986,
      "step": 1885
    },
    {
      "epoch": 0.15444932387474536,
      "grad_norm": 0.16015625,
      "learning_rate": 2.326370915708013e-05,
      "loss": 1.0999760627746582,
      "step": 1886
    },
    {
      "epoch": 0.15453121641126433,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3262846073595894e-05,
      "loss": 1.0285556316375732,
      "step": 1887
    },
    {
      "epoch": 0.15461310894778327,
      "grad_norm": 0.205078125,
      "learning_rate": 2.3261982500581977e-05,
      "loss": 0.9064048528671265,
      "step": 1888
    },
    {
      "epoch": 0.15469500148430224,
      "grad_norm": 0.154296875,
      "learning_rate": 2.326111843807592e-05,
      "loss": 0.8146162629127502,
      "step": 1889
    },
    {
      "epoch": 0.15477689402082118,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.326025388611527e-05,
      "loss": 0.4510672390460968,
      "step": 1890
    },
    {
      "epoch": 0.15485878655734012,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.325938884473761e-05,
      "loss": 0.7402224540710449,
      "step": 1891
    },
    {
      "epoch": 0.15494067909385908,
      "grad_norm": 0.16015625,
      "learning_rate": 2.325852331398054e-05,
      "loss": 0.5941665768623352,
      "step": 1892
    },
    {
      "epoch": 0.15502257163037803,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3257657293881674e-05,
      "loss": 0.8247563242912292,
      "step": 1893
    },
    {
      "epoch": 0.155104464166897,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.3256790784478653e-05,
      "loss": 0.8122414350509644,
      "step": 1894
    },
    {
      "epoch": 0.15518635670341593,
      "grad_norm": 0.1875,
      "learning_rate": 2.3255923785809147e-05,
      "loss": 1.1455626487731934,
      "step": 1895
    },
    {
      "epoch": 0.1552682492399349,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3255056297910825e-05,
      "loss": 0.6622958183288574,
      "step": 1896
    },
    {
      "epoch": 0.15535014177645384,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3254188320821405e-05,
      "loss": 0.7829557061195374,
      "step": 1897
    },
    {
      "epoch": 0.1554320343129728,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.325331985457861e-05,
      "loss": 0.9498754143714905,
      "step": 1898
    },
    {
      "epoch": 0.15551392684949175,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3252450899220183e-05,
      "loss": 1.2340171337127686,
      "step": 1899
    },
    {
      "epoch": 0.15559581938601072,
      "grad_norm": 0.146484375,
      "learning_rate": 2.3251581454783896e-05,
      "loss": 0.7974053621292114,
      "step": 1900
    },
    {
      "epoch": 0.15567771192252966,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3250711521307533e-05,
      "loss": 1.127620816230774,
      "step": 1901
    },
    {
      "epoch": 0.1557596044590486,
      "grad_norm": 0.171875,
      "learning_rate": 2.3249841098828912e-05,
      "loss": 0.443057119846344,
      "step": 1902
    },
    {
      "epoch": 0.15584149699556757,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.324897018738586e-05,
      "loss": 0.9318500757217407,
      "step": 1903
    },
    {
      "epoch": 0.1559233895320865,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3248098787016238e-05,
      "loss": 0.7802523374557495,
      "step": 1904
    },
    {
      "epoch": 0.15600528206860548,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.324722689775791e-05,
      "loss": 0.7112146615982056,
      "step": 1905
    },
    {
      "epoch": 0.15608717460512442,
      "grad_norm": 0.1875,
      "learning_rate": 2.324635451964878e-05,
      "loss": 0.6115627884864807,
      "step": 1906
    },
    {
      "epoch": 0.1561690671416434,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3245481652726758e-05,
      "loss": 0.9857809543609619,
      "step": 1907
    },
    {
      "epoch": 0.15625095967816233,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3244608297029792e-05,
      "loss": 0.6501709818840027,
      "step": 1908
    },
    {
      "epoch": 0.1563328522146813,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3243734452595832e-05,
      "loss": 0.507906973361969,
      "step": 1909
    },
    {
      "epoch": 0.15641474475120024,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3242860119462865e-05,
      "loss": 0.9455816149711609,
      "step": 1910
    },
    {
      "epoch": 0.1564966372877192,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.324198529766889e-05,
      "loss": 0.889655590057373,
      "step": 1911
    },
    {
      "epoch": 0.15657852982423814,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3241109987251935e-05,
      "loss": 0.8881437182426453,
      "step": 1912
    },
    {
      "epoch": 0.15666042236075708,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3240234188250038e-05,
      "loss": 0.5286319255828857,
      "step": 1913
    },
    {
      "epoch": 0.15674231489727605,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3239357900701272e-05,
      "loss": 0.9125378727912903,
      "step": 1914
    },
    {
      "epoch": 0.156824207433795,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3238481124643716e-05,
      "loss": 0.974899411201477,
      "step": 1915
    },
    {
      "epoch": 0.15690609997031396,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3237603860115484e-05,
      "loss": 0.9207587242126465,
      "step": 1916
    },
    {
      "epoch": 0.1569879925068329,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.32367261071547e-05,
      "loss": 0.6555114984512329,
      "step": 1917
    },
    {
      "epoch": 0.15706988504335187,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3235847865799525e-05,
      "loss": 0.8852350115776062,
      "step": 1918
    },
    {
      "epoch": 0.1571517775798708,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.323496913608812e-05,
      "loss": 0.6822836399078369,
      "step": 1919
    },
    {
      "epoch": 0.15723367011638978,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.3234089918058688e-05,
      "loss": 0.6815710067749023,
      "step": 1920
    },
    {
      "epoch": 0.15731556265290872,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.323321021174943e-05,
      "loss": 0.7130799293518066,
      "step": 1921
    },
    {
      "epoch": 0.15739745518942766,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3232330017198597e-05,
      "loss": 0.7802451848983765,
      "step": 1922
    },
    {
      "epoch": 0.15747934772594663,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3231449334444437e-05,
      "loss": 1.2767488956451416,
      "step": 1923
    },
    {
      "epoch": 0.15756124026246557,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3230568163525233e-05,
      "loss": 1.0317450761795044,
      "step": 1924
    },
    {
      "epoch": 0.15764313279898454,
      "grad_norm": 0.22265625,
      "learning_rate": 2.3229686504479275e-05,
      "loss": 1.0305235385894775,
      "step": 1925
    },
    {
      "epoch": 0.15772502533550348,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3228804357344896e-05,
      "loss": 0.6864871382713318,
      "step": 1926
    },
    {
      "epoch": 0.15780691787202245,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.322792172216043e-05,
      "loss": 0.9575198888778687,
      "step": 1927
    },
    {
      "epoch": 0.15788881040854139,
      "grad_norm": 0.171875,
      "learning_rate": 2.3227038598964238e-05,
      "loss": 1.0892506837844849,
      "step": 1928
    },
    {
      "epoch": 0.15797070294506035,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3226154987794707e-05,
      "loss": 1.0408154726028442,
      "step": 1929
    },
    {
      "epoch": 0.1580525954815793,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.322527088869025e-05,
      "loss": 0.9348576068878174,
      "step": 1930
    },
    {
      "epoch": 0.15813448801809826,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3224386301689282e-05,
      "loss": 1.0929203033447266,
      "step": 1931
    },
    {
      "epoch": 0.1582163805546172,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3223501226830255e-05,
      "loss": 0.9947420358657837,
      "step": 1932
    },
    {
      "epoch": 0.15829827309113614,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.322261566415164e-05,
      "loss": 0.7739285826683044,
      "step": 1933
    },
    {
      "epoch": 0.1583801656276551,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3221729613691926e-05,
      "loss": 0.5848917365074158,
      "step": 1934
    },
    {
      "epoch": 0.15846205816417405,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3220843075489623e-05,
      "loss": 0.6993266344070435,
      "step": 1935
    },
    {
      "epoch": 0.15854395070069302,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3219956049583267e-05,
      "loss": 0.5696324706077576,
      "step": 1936
    },
    {
      "epoch": 0.15862584323721196,
      "grad_norm": 0.171875,
      "learning_rate": 2.3219068536011406e-05,
      "loss": 0.8747787475585938,
      "step": 1937
    },
    {
      "epoch": 0.15870773577373093,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.321818053481262e-05,
      "loss": 0.93418288230896,
      "step": 1938
    },
    {
      "epoch": 0.15878962831024987,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.3217292046025504e-05,
      "loss": 1.0705628395080566,
      "step": 1939
    },
    {
      "epoch": 0.15887152084676884,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3216403069688675e-05,
      "loss": 1.0094441175460815,
      "step": 1940
    },
    {
      "epoch": 0.15895341338328778,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.321551360584077e-05,
      "loss": 1.2365726232528687,
      "step": 1941
    },
    {
      "epoch": 0.15903530591980675,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3214623654520454e-05,
      "loss": 0.779843270778656,
      "step": 1942
    },
    {
      "epoch": 0.1591171984563257,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.32137332157664e-05,
      "loss": 0.9291587471961975,
      "step": 1943
    },
    {
      "epoch": 0.15919909099284463,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.321284228961732e-05,
      "loss": 0.8547040820121765,
      "step": 1944
    },
    {
      "epoch": 0.1592809835293636,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3211950876111924e-05,
      "loss": 1.061223030090332,
      "step": 1945
    },
    {
      "epoch": 0.15936287606588254,
      "grad_norm": 0.2109375,
      "learning_rate": 2.321105897528897e-05,
      "loss": 0.9094790816307068,
      "step": 1946
    },
    {
      "epoch": 0.1594447686024015,
      "grad_norm": 0.1484375,
      "learning_rate": 2.3210166587187217e-05,
      "loss": 0.6813600063323975,
      "step": 1947
    },
    {
      "epoch": 0.15952666113892044,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3209273711845453e-05,
      "loss": 0.809687077999115,
      "step": 1948
    },
    {
      "epoch": 0.1596085536754394,
      "grad_norm": 0.203125,
      "learning_rate": 2.3208380349302484e-05,
      "loss": 0.8648678064346313,
      "step": 1949
    },
    {
      "epoch": 0.15969044621195835,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3207486499597142e-05,
      "loss": 1.147385835647583,
      "step": 1950
    },
    {
      "epoch": 0.15977233874847732,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3206592162768278e-05,
      "loss": 0.9705239534378052,
      "step": 1951
    },
    {
      "epoch": 0.15985423128499626,
      "grad_norm": 0.166015625,
      "learning_rate": 2.320569733885476e-05,
      "loss": 1.0364453792572021,
      "step": 1952
    },
    {
      "epoch": 0.15993612382151523,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3204802027895486e-05,
      "loss": 0.8680760264396667,
      "step": 1953
    },
    {
      "epoch": 0.16001801635803417,
      "grad_norm": 0.15625,
      "learning_rate": 2.3203906229929363e-05,
      "loss": 0.6885971426963806,
      "step": 1954
    },
    {
      "epoch": 0.1600999088945531,
      "grad_norm": 0.181640625,
      "learning_rate": 2.320300994499533e-05,
      "loss": 0.9312105774879456,
      "step": 1955
    },
    {
      "epoch": 0.16018180143107208,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3202113173132347e-05,
      "loss": 0.6682813167572021,
      "step": 1956
    },
    {
      "epoch": 0.16026369396759102,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3201215914379383e-05,
      "loss": 0.8751347661018372,
      "step": 1957
    },
    {
      "epoch": 0.16034558650411,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3200318168775443e-05,
      "loss": 1.0036970376968384,
      "step": 1958
    },
    {
      "epoch": 0.16042747904062893,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3199419936359543e-05,
      "loss": 0.7621043920516968,
      "step": 1959
    },
    {
      "epoch": 0.1605093715771479,
      "grad_norm": 0.19140625,
      "learning_rate": 2.319852121717073e-05,
      "loss": 0.9809600710868835,
      "step": 1960
    },
    {
      "epoch": 0.16059126411366684,
      "grad_norm": 0.154296875,
      "learning_rate": 2.3197622011248055e-05,
      "loss": 1.0563632249832153,
      "step": 1961
    },
    {
      "epoch": 0.1606731566501858,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3196722318630614e-05,
      "loss": 0.7657145857810974,
      "step": 1962
    },
    {
      "epoch": 0.16075504918670475,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3195822139357502e-05,
      "loss": 0.603901207447052,
      "step": 1963
    },
    {
      "epoch": 0.1608369417232237,
      "grad_norm": 0.142578125,
      "learning_rate": 2.3194921473467848e-05,
      "loss": 0.6035488843917847,
      "step": 1964
    },
    {
      "epoch": 0.16091883425974265,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3194020321000798e-05,
      "loss": 0.8689208030700684,
      "step": 1965
    },
    {
      "epoch": 0.1610007267962616,
      "grad_norm": 0.181640625,
      "learning_rate": 2.3193118681995516e-05,
      "loss": 1.0088626146316528,
      "step": 1966
    },
    {
      "epoch": 0.16108261933278056,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3192216556491202e-05,
      "loss": 0.8131465315818787,
      "step": 1967
    },
    {
      "epoch": 0.1611645118692995,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3191313944527054e-05,
      "loss": 0.5557928681373596,
      "step": 1968
    },
    {
      "epoch": 0.16124640440581847,
      "grad_norm": 0.16015625,
      "learning_rate": 2.319041084614231e-05,
      "loss": 1.294734239578247,
      "step": 1969
    },
    {
      "epoch": 0.1613282969423374,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3189507261376222e-05,
      "loss": 0.7189170122146606,
      "step": 1970
    },
    {
      "epoch": 0.16141018947885638,
      "grad_norm": 0.1875,
      "learning_rate": 2.3188603190268064e-05,
      "loss": 1.066721796989441,
      "step": 1971
    },
    {
      "epoch": 0.16149208201537532,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3187698632857125e-05,
      "loss": 1.1146469116210938,
      "step": 1972
    },
    {
      "epoch": 0.1615739745518943,
      "grad_norm": 0.1796875,
      "learning_rate": 2.318679358918273e-05,
      "loss": 1.247765302658081,
      "step": 1973
    },
    {
      "epoch": 0.16165586708841323,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.31858880592842e-05,
      "loss": 0.6710013747215271,
      "step": 1974
    },
    {
      "epoch": 0.16173775962493217,
      "grad_norm": 0.212890625,
      "learning_rate": 2.3184982043200913e-05,
      "loss": 0.9186426997184753,
      "step": 1975
    },
    {
      "epoch": 0.16181965216145114,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3184075540972234e-05,
      "loss": 0.8672297596931458,
      "step": 1976
    },
    {
      "epoch": 0.16190154469797008,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3183168552637574e-05,
      "loss": 0.8545089364051819,
      "step": 1977
    },
    {
      "epoch": 0.16198343723448905,
      "grad_norm": 0.15625,
      "learning_rate": 2.3182261078236344e-05,
      "loss": 0.7210851907730103,
      "step": 1978
    },
    {
      "epoch": 0.162065329771008,
      "grad_norm": 0.16796875,
      "learning_rate": 2.318135311780799e-05,
      "loss": 0.9063572287559509,
      "step": 1979
    },
    {
      "epoch": 0.16214722230752696,
      "grad_norm": 0.1796875,
      "learning_rate": 2.318044467139198e-05,
      "loss": 1.060003399848938,
      "step": 1980
    },
    {
      "epoch": 0.1622291148440459,
      "grad_norm": 0.171875,
      "learning_rate": 2.3179535739027796e-05,
      "loss": 0.8971522450447083,
      "step": 1981
    },
    {
      "epoch": 0.16231100738056486,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.317862632075494e-05,
      "loss": 0.6887643337249756,
      "step": 1982
    },
    {
      "epoch": 0.1623928999170838,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.317771641661295e-05,
      "loss": 1.041860580444336,
      "step": 1983
    },
    {
      "epoch": 0.16247479245360277,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3176806026641356e-05,
      "loss": 0.9931774139404297,
      "step": 1984
    },
    {
      "epoch": 0.1625566849901217,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3175895150879745e-05,
      "loss": 0.6979286670684814,
      "step": 1985
    },
    {
      "epoch": 0.16263857752664065,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.31749837893677e-05,
      "loss": 0.9371703863143921,
      "step": 1986
    },
    {
      "epoch": 0.16272047006315962,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3174071942144833e-05,
      "loss": 0.6812039613723755,
      "step": 1987
    },
    {
      "epoch": 0.16280236259967856,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.317315960925078e-05,
      "loss": 1.0110540390014648,
      "step": 1988
    },
    {
      "epoch": 0.16288425513619753,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3172246790725186e-05,
      "loss": 0.8606452345848083,
      "step": 1989
    },
    {
      "epoch": 0.16296614767271647,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.3171333486607733e-05,
      "loss": 0.5210959911346436,
      "step": 1990
    },
    {
      "epoch": 0.16304804020923544,
      "grad_norm": 0.1875,
      "learning_rate": 2.3170419696938116e-05,
      "loss": 0.7098227739334106,
      "step": 1991
    },
    {
      "epoch": 0.16312993274575438,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.316950542175605e-05,
      "loss": 1.1202596426010132,
      "step": 1992
    },
    {
      "epoch": 0.16321182528227335,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3168590661101277e-05,
      "loss": 0.7595298290252686,
      "step": 1993
    },
    {
      "epoch": 0.1632937178187923,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3167675415013554e-05,
      "loss": 0.8345801830291748,
      "step": 1994
    },
    {
      "epoch": 0.16337561035531123,
      "grad_norm": 0.140625,
      "learning_rate": 2.316675968353266e-05,
      "loss": 0.6969337463378906,
      "step": 1995
    },
    {
      "epoch": 0.1634575028918302,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3165843466698395e-05,
      "loss": 0.8890387415885925,
      "step": 1996
    },
    {
      "epoch": 0.16353939542834914,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3164926764550587e-05,
      "loss": 0.8984854817390442,
      "step": 1997
    },
    {
      "epoch": 0.1636212879648681,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.3164009577129075e-05,
      "loss": 1.393692135810852,
      "step": 1998
    },
    {
      "epoch": 0.16370318050138705,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3163091904473728e-05,
      "loss": 0.7264331579208374,
      "step": 1999
    },
    {
      "epoch": 0.16378507303790601,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.316217374662443e-05,
      "loss": 1.0736693143844604,
      "step": 2000
    },
    {
      "epoch": 0.16386696557442496,
      "grad_norm": 0.146484375,
      "learning_rate": 2.3161255103621086e-05,
      "loss": 0.9192960262298584,
      "step": 2001
    },
    {
      "epoch": 0.16394885811094392,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3160335975503622e-05,
      "loss": 0.9903260469436646,
      "step": 2002
    },
    {
      "epoch": 0.16403075064746286,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.3159416362311998e-05,
      "loss": 0.739921510219574,
      "step": 2003
    },
    {
      "epoch": 0.16411264318398183,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.315849626408617e-05,
      "loss": 1.1444851160049438,
      "step": 2004
    },
    {
      "epoch": 0.16419453572050077,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.315757568086614e-05,
      "loss": 0.6804274320602417,
      "step": 2005
    },
    {
      "epoch": 0.1642764282570197,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3156654612691913e-05,
      "loss": 0.7933639287948608,
      "step": 2006
    },
    {
      "epoch": 0.16435832079353868,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3155733059603526e-05,
      "loss": 0.7657653093338013,
      "step": 2007
    },
    {
      "epoch": 0.16444021333005762,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3154811021641035e-05,
      "loss": 0.8075305223464966,
      "step": 2008
    },
    {
      "epoch": 0.1645221058665766,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3153888498844512e-05,
      "loss": 0.8470852375030518,
      "step": 2009
    },
    {
      "epoch": 0.16460399840309553,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.315296549125406e-05,
      "loss": 0.584110677242279,
      "step": 2010
    },
    {
      "epoch": 0.1646858909396145,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3152041998909787e-05,
      "loss": 1.0222033262252808,
      "step": 2011
    },
    {
      "epoch": 0.16476778347613344,
      "grad_norm": 0.185546875,
      "learning_rate": 2.315111802185184e-05,
      "loss": 0.8398823142051697,
      "step": 2012
    },
    {
      "epoch": 0.1648496760126524,
      "grad_norm": 0.173828125,
      "learning_rate": 2.315019356012037e-05,
      "loss": 0.7560561895370483,
      "step": 2013
    },
    {
      "epoch": 0.16493156854917135,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.314926861375557e-05,
      "loss": 0.7072266340255737,
      "step": 2014
    },
    {
      "epoch": 0.16501346108569032,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3148343182797636e-05,
      "loss": 0.8676473498344421,
      "step": 2015
    },
    {
      "epoch": 0.16509535362220926,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3147417267286784e-05,
      "loss": 0.8015260696411133,
      "step": 2016
    },
    {
      "epoch": 0.1651772461587282,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.314649086726327e-05,
      "loss": 1.1094400882720947,
      "step": 2017
    },
    {
      "epoch": 0.16525913869524717,
      "grad_norm": 0.220703125,
      "learning_rate": 2.314556398276735e-05,
      "loss": 1.2217539548873901,
      "step": 2018
    },
    {
      "epoch": 0.1653410312317661,
      "grad_norm": 0.138671875,
      "learning_rate": 2.314463661383932e-05,
      "loss": 0.8388075828552246,
      "step": 2019
    },
    {
      "epoch": 0.16542292376828507,
      "grad_norm": 0.15625,
      "learning_rate": 2.3143708760519474e-05,
      "loss": 1.0679703950881958,
      "step": 2020
    },
    {
      "epoch": 0.16550481630480401,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.314278042284815e-05,
      "loss": 0.9242013692855835,
      "step": 2021
    },
    {
      "epoch": 0.16558670884132298,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3141851600865697e-05,
      "loss": 0.9729780554771423,
      "step": 2022
    },
    {
      "epoch": 0.16566860137784192,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.3140922294612483e-05,
      "loss": 0.9754358530044556,
      "step": 2023
    },
    {
      "epoch": 0.1657504939143609,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.31399925041289e-05,
      "loss": 0.9507209658622742,
      "step": 2024
    },
    {
      "epoch": 0.16583238645087983,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3139062229455363e-05,
      "loss": 0.9255710244178772,
      "step": 2025
    },
    {
      "epoch": 0.16591427898739877,
      "grad_norm": 0.142578125,
      "learning_rate": 2.31381314706323e-05,
      "loss": 0.7302979230880737,
      "step": 2026
    },
    {
      "epoch": 0.16599617152391774,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3137200227700172e-05,
      "loss": 0.9877808094024658,
      "step": 2027
    },
    {
      "epoch": 0.16607806406043668,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.3136268500699444e-05,
      "loss": 0.9927208423614502,
      "step": 2028
    },
    {
      "epoch": 0.16615995659695565,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3135336289670625e-05,
      "loss": 0.6989736557006836,
      "step": 2029
    },
    {
      "epoch": 0.1662418491334746,
      "grad_norm": 0.171875,
      "learning_rate": 2.3134403594654227e-05,
      "loss": 1.0436620712280273,
      "step": 2030
    },
    {
      "epoch": 0.16632374166999356,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3133470415690793e-05,
      "loss": 0.9086163640022278,
      "step": 2031
    },
    {
      "epoch": 0.1664056342065125,
      "grad_norm": 0.203125,
      "learning_rate": 2.3132536752820875e-05,
      "loss": 0.8396445512771606,
      "step": 2032
    },
    {
      "epoch": 0.16648752674303147,
      "grad_norm": 0.150390625,
      "learning_rate": 2.313160260608506e-05,
      "loss": 0.8068231344223022,
      "step": 2033
    },
    {
      "epoch": 0.1665694192795504,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.3130667975523947e-05,
      "loss": 0.8173387050628662,
      "step": 2034
    },
    {
      "epoch": 0.16665131181606938,
      "grad_norm": 0.173828125,
      "learning_rate": 2.312973286117816e-05,
      "loss": 0.9024599194526672,
      "step": 2035
    },
    {
      "epoch": 0.16673320435258832,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3128797263088345e-05,
      "loss": 0.817664623260498,
      "step": 2036
    },
    {
      "epoch": 0.16681509688910726,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3127861181295164e-05,
      "loss": 0.694021463394165,
      "step": 2037
    },
    {
      "epoch": 0.16689698942562622,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3126924615839302e-05,
      "loss": 0.749485433101654,
      "step": 2038
    },
    {
      "epoch": 0.16697888196214516,
      "grad_norm": 0.166015625,
      "learning_rate": 2.3125987566761468e-05,
      "loss": 0.7049168348312378,
      "step": 2039
    },
    {
      "epoch": 0.16706077449866413,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.312505003410239e-05,
      "loss": 0.8267319202423096,
      "step": 2040
    },
    {
      "epoch": 0.16714266703518307,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3124112017902814e-05,
      "loss": 0.8198041915893555,
      "step": 2041
    },
    {
      "epoch": 0.16722455957170204,
      "grad_norm": 0.171875,
      "learning_rate": 2.312317351820352e-05,
      "loss": 0.8141278028488159,
      "step": 2042
    },
    {
      "epoch": 0.16730645210822098,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3122234535045288e-05,
      "loss": 0.8274849057197571,
      "step": 2043
    },
    {
      "epoch": 0.16738834464473995,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3121295068468934e-05,
      "loss": 0.9098456501960754,
      "step": 2044
    },
    {
      "epoch": 0.1674702371812589,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.3120355118515288e-05,
      "loss": 0.9314104318618774,
      "step": 2045
    },
    {
      "epoch": 0.16755212971777786,
      "grad_norm": 0.18359375,
      "learning_rate": 2.311941468522521e-05,
      "loss": 1.144854187965393,
      "step": 2046
    },
    {
      "epoch": 0.1676340222542968,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3118473768639575e-05,
      "loss": 1.1058868169784546,
      "step": 2047
    },
    {
      "epoch": 0.16771591479081574,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3117532368799275e-05,
      "loss": 1.0394481420516968,
      "step": 2048
    },
    {
      "epoch": 0.1677978073273347,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.3116590485745224e-05,
      "loss": 0.7576707005500793,
      "step": 2049
    },
    {
      "epoch": 0.16787969986385365,
      "grad_norm": 0.1875,
      "learning_rate": 2.311564811951837e-05,
      "loss": 0.9654994606971741,
      "step": 2050
    },
    {
      "epoch": 0.16796159240037262,
      "grad_norm": 0.212890625,
      "learning_rate": 2.3114705270159666e-05,
      "loss": 1.1635290384292603,
      "step": 2051
    },
    {
      "epoch": 0.16804348493689156,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.3113761937710093e-05,
      "loss": 0.7197125554084778,
      "step": 2052
    },
    {
      "epoch": 0.16812537747341053,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.3112818122210648e-05,
      "loss": 0.8843101859092712,
      "step": 2053
    },
    {
      "epoch": 0.16820727000992947,
      "grad_norm": 0.14453125,
      "learning_rate": 2.311187382370236e-05,
      "loss": 0.8230028748512268,
      "step": 2054
    },
    {
      "epoch": 0.16828916254644843,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3110929042226268e-05,
      "loss": 1.0551960468292236,
      "step": 2055
    },
    {
      "epoch": 0.16837105508296737,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3109983777823438e-05,
      "loss": 0.9145382642745972,
      "step": 2056
    },
    {
      "epoch": 0.16845294761948634,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3109038030534954e-05,
      "loss": 0.8974514007568359,
      "step": 2057
    },
    {
      "epoch": 0.16853484015600528,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3108091800401922e-05,
      "loss": 0.6485768556594849,
      "step": 2058
    },
    {
      "epoch": 0.16861673269252422,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.3107145087465473e-05,
      "loss": 0.7728451490402222,
      "step": 2059
    },
    {
      "epoch": 0.1686986252290432,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3106197891766747e-05,
      "loss": 0.5592920780181885,
      "step": 2060
    },
    {
      "epoch": 0.16878051776556213,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3105250213346922e-05,
      "loss": 0.9258127212524414,
      "step": 2061
    },
    {
      "epoch": 0.1688624103020811,
      "grad_norm": 0.16015625,
      "learning_rate": 2.310430205224718e-05,
      "loss": 0.948998749256134,
      "step": 2062
    },
    {
      "epoch": 0.16894430283860004,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3103353408508738e-05,
      "loss": 0.9782514572143555,
      "step": 2063
    },
    {
      "epoch": 0.169026195375119,
      "grad_norm": 0.17578125,
      "learning_rate": 2.3102404282172823e-05,
      "loss": 0.9074375629425049,
      "step": 2064
    },
    {
      "epoch": 0.16910808791163795,
      "grad_norm": 0.203125,
      "learning_rate": 2.310145467328069e-05,
      "loss": 0.8254364132881165,
      "step": 2065
    },
    {
      "epoch": 0.16918998044815692,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3100504581873614e-05,
      "loss": 0.8852524757385254,
      "step": 2066
    },
    {
      "epoch": 0.16927187298467586,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3099554007992893e-05,
      "loss": 0.6029716730117798,
      "step": 2067
    },
    {
      "epoch": 0.1693537655211948,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.3098602951679834e-05,
      "loss": 0.631441593170166,
      "step": 2068
    },
    {
      "epoch": 0.16943565805771377,
      "grad_norm": 0.171875,
      "learning_rate": 2.309765141297578e-05,
      "loss": 0.7891640067100525,
      "step": 2069
    },
    {
      "epoch": 0.1695175505942327,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3096699391922086e-05,
      "loss": 0.8181204795837402,
      "step": 2070
    },
    {
      "epoch": 0.16959944313075168,
      "grad_norm": 0.16796875,
      "learning_rate": 2.309574688856014e-05,
      "loss": 0.7776691913604736,
      "step": 2071
    },
    {
      "epoch": 0.16968133566727062,
      "grad_norm": 0.15625,
      "learning_rate": 2.3094793902931327e-05,
      "loss": 1.044589877128601,
      "step": 2072
    },
    {
      "epoch": 0.16976322820378958,
      "grad_norm": 0.16796875,
      "learning_rate": 2.309384043507708e-05,
      "loss": 0.8124565482139587,
      "step": 2073
    },
    {
      "epoch": 0.16984512074030853,
      "grad_norm": 0.1796875,
      "learning_rate": 2.309288648503883e-05,
      "loss": 0.5685734748840332,
      "step": 2074
    },
    {
      "epoch": 0.1699270132768275,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.309193205285805e-05,
      "loss": 0.5505271553993225,
      "step": 2075
    },
    {
      "epoch": 0.17000890581334643,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3090977138576213e-05,
      "loss": 0.7584421634674072,
      "step": 2076
    },
    {
      "epoch": 0.1700907983498654,
      "grad_norm": 0.138671875,
      "learning_rate": 2.3090021742234838e-05,
      "loss": 0.8737994432449341,
      "step": 2077
    },
    {
      "epoch": 0.17017269088638434,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.3089065863875435e-05,
      "loss": 0.8054501414299011,
      "step": 2078
    },
    {
      "epoch": 0.17025458342290328,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3088109503539557e-05,
      "loss": 0.7641791105270386,
      "step": 2079
    },
    {
      "epoch": 0.17033647595942225,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3087152661268773e-05,
      "loss": 0.9865874648094177,
      "step": 2080
    },
    {
      "epoch": 0.1704183684959412,
      "grad_norm": 0.171875,
      "learning_rate": 2.3086195337104672e-05,
      "loss": 0.6361075043678284,
      "step": 2081
    },
    {
      "epoch": 0.17050026103246016,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3085237531088857e-05,
      "loss": 0.8034785985946655,
      "step": 2082
    },
    {
      "epoch": 0.1705821535689791,
      "grad_norm": 0.1640625,
      "learning_rate": 2.3084279243262964e-05,
      "loss": 1.0324164628982544,
      "step": 2083
    },
    {
      "epoch": 0.17066404610549807,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.308332047366864e-05,
      "loss": 0.8297885060310364,
      "step": 2084
    },
    {
      "epoch": 0.170745938642017,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3082361222347563e-05,
      "loss": 0.7850766181945801,
      "step": 2085
    },
    {
      "epoch": 0.17082783117853598,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.308140148934142e-05,
      "loss": 0.7645384669303894,
      "step": 2086
    },
    {
      "epoch": 0.17090972371505492,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.308044127469193e-05,
      "loss": 1.186807632446289,
      "step": 2087
    },
    {
      "epoch": 0.17099161625157389,
      "grad_norm": 0.15234375,
      "learning_rate": 2.3079480578440823e-05,
      "loss": 0.5767953395843506,
      "step": 2088
    },
    {
      "epoch": 0.17107350878809283,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.3078519400629858e-05,
      "loss": 0.839798092842102,
      "step": 2089
    },
    {
      "epoch": 0.17115540132461177,
      "grad_norm": 0.181640625,
      "learning_rate": 2.307755774130081e-05,
      "loss": 0.8485851287841797,
      "step": 2090
    },
    {
      "epoch": 0.17123729386113073,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.3076595600495475e-05,
      "loss": 0.7233185172080994,
      "step": 2091
    },
    {
      "epoch": 0.17131918639764968,
      "grad_norm": 0.21484375,
      "learning_rate": 2.307563297825568e-05,
      "loss": 1.4081238508224487,
      "step": 2092
    },
    {
      "epoch": 0.17140107893416864,
      "grad_norm": 0.24609375,
      "learning_rate": 2.3074669874623253e-05,
      "loss": 0.9178886413574219,
      "step": 2093
    },
    {
      "epoch": 0.17148297147068758,
      "grad_norm": 0.1279296875,
      "learning_rate": 2.3073706289640062e-05,
      "loss": 0.9092115163803101,
      "step": 2094
    },
    {
      "epoch": 0.17156486400720655,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.3072742223347987e-05,
      "loss": 1.0433051586151123,
      "step": 2095
    },
    {
      "epoch": 0.1716467565437255,
      "grad_norm": 0.173828125,
      "learning_rate": 2.307177767578893e-05,
      "loss": 0.6784330606460571,
      "step": 2096
    },
    {
      "epoch": 0.17172864908024446,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.3070812647004815e-05,
      "loss": 1.0354896783828735,
      "step": 2097
    },
    {
      "epoch": 0.1718105416167634,
      "grad_norm": 0.1875,
      "learning_rate": 2.3069847137037584e-05,
      "loss": 1.0092010498046875,
      "step": 2098
    },
    {
      "epoch": 0.17189243415328234,
      "grad_norm": 0.173828125,
      "learning_rate": 2.3068881145929198e-05,
      "loss": 1.04713773727417,
      "step": 2099
    },
    {
      "epoch": 0.1719743266898013,
      "grad_norm": 0.18359375,
      "learning_rate": 2.3067914673721656e-05,
      "loss": 0.7055670619010925,
      "step": 2100
    },
    {
      "epoch": 0.17205621922632025,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.306694772045695e-05,
      "loss": 1.066530466079712,
      "step": 2101
    },
    {
      "epoch": 0.17213811176283922,
      "grad_norm": 0.17578125,
      "learning_rate": 2.306598028617712e-05,
      "loss": 1.0950257778167725,
      "step": 2102
    },
    {
      "epoch": 0.17222000429935816,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3065012370924204e-05,
      "loss": 0.7926771640777588,
      "step": 2103
    },
    {
      "epoch": 0.17230189683587713,
      "grad_norm": 0.14453125,
      "learning_rate": 2.3064043974740283e-05,
      "loss": 0.814128041267395,
      "step": 2104
    },
    {
      "epoch": 0.17238378937239607,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3063075097667437e-05,
      "loss": 0.864211916923523,
      "step": 2105
    },
    {
      "epoch": 0.17246568190891504,
      "grad_norm": 0.16015625,
      "learning_rate": 2.3062105739747786e-05,
      "loss": 0.8817232251167297,
      "step": 2106
    },
    {
      "epoch": 0.17254757444543398,
      "grad_norm": 0.14453125,
      "learning_rate": 2.306113590102345e-05,
      "loss": 0.5957571268081665,
      "step": 2107
    },
    {
      "epoch": 0.17262946698195294,
      "grad_norm": 0.171875,
      "learning_rate": 2.30601655815366e-05,
      "loss": 0.9646713137626648,
      "step": 2108
    },
    {
      "epoch": 0.17271135951847189,
      "grad_norm": 0.169921875,
      "learning_rate": 2.3059194781329396e-05,
      "loss": 0.5499109029769897,
      "step": 2109
    },
    {
      "epoch": 0.17279325205499083,
      "grad_norm": 0.1875,
      "learning_rate": 2.3058223500444038e-05,
      "loss": 1.030017614364624,
      "step": 2110
    },
    {
      "epoch": 0.1728751445915098,
      "grad_norm": 0.17578125,
      "learning_rate": 2.305725173892274e-05,
      "loss": 0.7581807971000671,
      "step": 2111
    },
    {
      "epoch": 0.17295703712802873,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3056279496807738e-05,
      "loss": 0.8523330688476562,
      "step": 2112
    },
    {
      "epoch": 0.1730389296645477,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3055306774141296e-05,
      "loss": 0.8295431137084961,
      "step": 2113
    },
    {
      "epoch": 0.17312082220106664,
      "grad_norm": 0.158203125,
      "learning_rate": 2.3054333570965686e-05,
      "loss": 0.5596280694007874,
      "step": 2114
    },
    {
      "epoch": 0.1732027147375856,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.305335988732321e-05,
      "loss": 0.8161397576332092,
      "step": 2115
    },
    {
      "epoch": 0.17328460727410455,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3052385723256187e-05,
      "loss": 0.608106791973114,
      "step": 2116
    },
    {
      "epoch": 0.17336649981062352,
      "grad_norm": 0.16015625,
      "learning_rate": 2.305141107880696e-05,
      "loss": 0.7428133487701416,
      "step": 2117
    },
    {
      "epoch": 0.17344839234714246,
      "grad_norm": 0.169921875,
      "learning_rate": 2.305043595401789e-05,
      "loss": 0.6883682608604431,
      "step": 2118
    },
    {
      "epoch": 0.17353028488366143,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3049460348931357e-05,
      "loss": 1.2211458683013916,
      "step": 2119
    },
    {
      "epoch": 0.17361217742018037,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.3048484263589773e-05,
      "loss": 0.8047317862510681,
      "step": 2120
    },
    {
      "epoch": 0.1736940699566993,
      "grad_norm": 0.208984375,
      "learning_rate": 2.3047507698035557e-05,
      "loss": 1.1279809474945068,
      "step": 2121
    },
    {
      "epoch": 0.17377596249321828,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.3046530652311154e-05,
      "loss": 0.9892963767051697,
      "step": 2122
    },
    {
      "epoch": 0.17385785502973722,
      "grad_norm": 0.169921875,
      "learning_rate": 2.304555312645903e-05,
      "loss": 1.1074419021606445,
      "step": 2123
    },
    {
      "epoch": 0.1739397475662562,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3044575120521677e-05,
      "loss": 1.0455310344696045,
      "step": 2124
    },
    {
      "epoch": 0.17402164010277513,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.3043596634541593e-05,
      "loss": 0.7842694520950317,
      "step": 2125
    },
    {
      "epoch": 0.1741035326392941,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.304261766856132e-05,
      "loss": 0.927800714969635,
      "step": 2126
    },
    {
      "epoch": 0.17418542517581304,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.3041638222623403e-05,
      "loss": 0.8272109031677246,
      "step": 2127
    },
    {
      "epoch": 0.174267317712332,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.3040658296770407e-05,
      "loss": 0.5830202698707581,
      "step": 2128
    },
    {
      "epoch": 0.17434921024885094,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.303967789104493e-05,
      "loss": 1.0474522113800049,
      "step": 2129
    },
    {
      "epoch": 0.17443110278536988,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3038697005489585e-05,
      "loss": 1.3157117366790771,
      "step": 2130
    },
    {
      "epoch": 0.17451299532188885,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3037715640147e-05,
      "loss": 0.7627755403518677,
      "step": 2131
    },
    {
      "epoch": 0.1745948878584078,
      "grad_norm": 0.189453125,
      "learning_rate": 2.3036733795059835e-05,
      "loss": 0.7738596796989441,
      "step": 2132
    },
    {
      "epoch": 0.17467678039492676,
      "grad_norm": 0.20703125,
      "learning_rate": 2.303575147027076e-05,
      "loss": 0.9290691614151001,
      "step": 2133
    },
    {
      "epoch": 0.1747586729314457,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.3034768665822473e-05,
      "loss": 0.4888533651828766,
      "step": 2134
    },
    {
      "epoch": 0.17484056546796467,
      "grad_norm": 0.19140625,
      "learning_rate": 2.3033785381757694e-05,
      "loss": 1.016878604888916,
      "step": 2135
    },
    {
      "epoch": 0.1749224580044836,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.3032801618119153e-05,
      "loss": 0.8429866433143616,
      "step": 2136
    },
    {
      "epoch": 0.17500435054100258,
      "grad_norm": 0.216796875,
      "learning_rate": 2.3031817374949615e-05,
      "loss": 0.8701666593551636,
      "step": 2137
    },
    {
      "epoch": 0.17508624307752152,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.303083265229186e-05,
      "loss": 1.0822858810424805,
      "step": 2138
    },
    {
      "epoch": 0.1751681356140405,
      "grad_norm": 0.166015625,
      "learning_rate": 2.302984745018868e-05,
      "loss": 0.6886203289031982,
      "step": 2139
    },
    {
      "epoch": 0.17525002815055943,
      "grad_norm": 0.1875,
      "learning_rate": 2.3028861768682907e-05,
      "loss": 0.6356196403503418,
      "step": 2140
    },
    {
      "epoch": 0.17533192068707837,
      "grad_norm": 0.162109375,
      "learning_rate": 2.3027875607817374e-05,
      "loss": 0.5120158195495605,
      "step": 2141
    },
    {
      "epoch": 0.17541381322359734,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.3026888967634947e-05,
      "loss": 0.7832841873168945,
      "step": 2142
    },
    {
      "epoch": 0.17549570576011628,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.302590184817851e-05,
      "loss": 0.9659411311149597,
      "step": 2143
    },
    {
      "epoch": 0.17557759829663525,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.302491424949096e-05,
      "loss": 0.6503680944442749,
      "step": 2144
    },
    {
      "epoch": 0.17565949083315419,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3023926171615237e-05,
      "loss": 0.8655272722244263,
      "step": 2145
    },
    {
      "epoch": 0.17574138336967315,
      "grad_norm": 0.16796875,
      "learning_rate": 2.3022937614594276e-05,
      "loss": 0.7896318435668945,
      "step": 2146
    },
    {
      "epoch": 0.1758232759061921,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.3021948578471043e-05,
      "loss": 0.9102089405059814,
      "step": 2147
    },
    {
      "epoch": 0.17590516844271106,
      "grad_norm": 0.1796875,
      "learning_rate": 2.3020959063288533e-05,
      "loss": 1.209578514099121,
      "step": 2148
    },
    {
      "epoch": 0.17598706097923,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.301996906908975e-05,
      "loss": 0.948302686214447,
      "step": 2149
    },
    {
      "epoch": 0.17606895351574897,
      "grad_norm": 0.16015625,
      "learning_rate": 2.301897859591772e-05,
      "loss": 0.7943192720413208,
      "step": 2150
    },
    {
      "epoch": 0.1761508460522679,
      "grad_norm": 0.201171875,
      "learning_rate": 2.3017987643815496e-05,
      "loss": 0.7528172731399536,
      "step": 2151
    },
    {
      "epoch": 0.17623273858878685,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.3016996212826148e-05,
      "loss": 0.8487266302108765,
      "step": 2152
    },
    {
      "epoch": 0.17631463112530582,
      "grad_norm": 0.177734375,
      "learning_rate": 2.3016004302992774e-05,
      "loss": 0.9192425608634949,
      "step": 2153
    },
    {
      "epoch": 0.17639652366182476,
      "grad_norm": 0.1640625,
      "learning_rate": 2.301501191435848e-05,
      "loss": 0.6583120226860046,
      "step": 2154
    },
    {
      "epoch": 0.17647841619834373,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.30140190469664e-05,
      "loss": 0.8925849199295044,
      "step": 2155
    },
    {
      "epoch": 0.17656030873486267,
      "grad_norm": 0.171875,
      "learning_rate": 2.301302570085969e-05,
      "loss": 0.9125053286552429,
      "step": 2156
    },
    {
      "epoch": 0.17664220127138164,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.3012031876081523e-05,
      "loss": 0.9876391291618347,
      "step": 2157
    },
    {
      "epoch": 0.17672409380790058,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.301103757267509e-05,
      "loss": 1.21542227268219,
      "step": 2158
    },
    {
      "epoch": 0.17680598634441955,
      "grad_norm": 0.169921875,
      "learning_rate": 2.301004279068362e-05,
      "loss": 0.8203423619270325,
      "step": 2159
    },
    {
      "epoch": 0.1768878788809385,
      "grad_norm": 0.1875,
      "learning_rate": 2.300904753015034e-05,
      "loss": 0.9300917387008667,
      "step": 2160
    },
    {
      "epoch": 0.17696977141745743,
      "grad_norm": 0.189453125,
      "learning_rate": 2.300805179111851e-05,
      "loss": 1.0471628904342651,
      "step": 2161
    },
    {
      "epoch": 0.1770516639539764,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.3007055573631416e-05,
      "loss": 0.694509744644165,
      "step": 2162
    },
    {
      "epoch": 0.17713355649049534,
      "grad_norm": 0.19140625,
      "learning_rate": 2.300605887773235e-05,
      "loss": 0.8566156625747681,
      "step": 2163
    },
    {
      "epoch": 0.1772154490270143,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.300506170346463e-05,
      "loss": 1.1923434734344482,
      "step": 2164
    },
    {
      "epoch": 0.17729734156353325,
      "grad_norm": 0.15625,
      "learning_rate": 2.3004064050871607e-05,
      "loss": 0.7378466725349426,
      "step": 2165
    },
    {
      "epoch": 0.1773792341000522,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.3003065919996637e-05,
      "loss": 0.9856939911842346,
      "step": 2166
    },
    {
      "epoch": 0.17746112663657115,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3002067310883102e-05,
      "loss": 0.8166617751121521,
      "step": 2167
    },
    {
      "epoch": 0.17754301917309012,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3001068223574406e-05,
      "loss": 1.0910999774932861,
      "step": 2168
    },
    {
      "epoch": 0.17762491170960906,
      "grad_norm": 0.171875,
      "learning_rate": 2.300006865811398e-05,
      "loss": 0.8753263354301453,
      "step": 2169
    },
    {
      "epoch": 0.17770680424612803,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.299906861454526e-05,
      "loss": 0.8544874787330627,
      "step": 2170
    },
    {
      "epoch": 0.17778869678264697,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2998068092911717e-05,
      "loss": 0.8821544647216797,
      "step": 2171
    },
    {
      "epoch": 0.1778705893191659,
      "grad_norm": 0.19140625,
      "learning_rate": 2.299706709325684e-05,
      "loss": 0.6823170185089111,
      "step": 2172
    },
    {
      "epoch": 0.17795248185568488,
      "grad_norm": 0.17578125,
      "learning_rate": 2.2996065615624128e-05,
      "loss": 0.9480848908424377,
      "step": 2173
    },
    {
      "epoch": 0.17803437439220382,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.299506366005712e-05,
      "loss": 0.7389025688171387,
      "step": 2174
    },
    {
      "epoch": 0.1781162669287228,
      "grad_norm": 0.119140625,
      "learning_rate": 2.299406122659936e-05,
      "loss": 0.6551054120063782,
      "step": 2175
    },
    {
      "epoch": 0.17819815946524173,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.299305831529441e-05,
      "loss": 0.9082637429237366,
      "step": 2176
    },
    {
      "epoch": 0.1782800520017607,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2992054926185876e-05,
      "loss": 0.8177830576896667,
      "step": 2177
    },
    {
      "epoch": 0.17836194453827964,
      "grad_norm": 0.173828125,
      "learning_rate": 2.299105105931736e-05,
      "loss": 1.126342535018921,
      "step": 2178
    },
    {
      "epoch": 0.1784438370747986,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.2990046714732497e-05,
      "loss": 0.787146806716919,
      "step": 2179
    },
    {
      "epoch": 0.17852572961131755,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2989041892474942e-05,
      "loss": 0.5844379663467407,
      "step": 2180
    },
    {
      "epoch": 0.17860762214783651,
      "grad_norm": 0.181640625,
      "learning_rate": 2.2988036592588362e-05,
      "loss": 0.7549855709075928,
      "step": 2181
    },
    {
      "epoch": 0.17868951468435545,
      "grad_norm": 0.146484375,
      "learning_rate": 2.2987030815116455e-05,
      "loss": 0.8013399243354797,
      "step": 2182
    },
    {
      "epoch": 0.1787714072208744,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2986024560102937e-05,
      "loss": 0.8951451778411865,
      "step": 2183
    },
    {
      "epoch": 0.17885329975739336,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2985017827591543e-05,
      "loss": 0.7120135426521301,
      "step": 2184
    },
    {
      "epoch": 0.1789351922939123,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2984010617626033e-05,
      "loss": 0.6407305002212524,
      "step": 2185
    },
    {
      "epoch": 0.17901708483043127,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.298300293025018e-05,
      "loss": 0.7624015212059021,
      "step": 2186
    },
    {
      "epoch": 0.1790989773669502,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2981994765507783e-05,
      "loss": 0.6639080047607422,
      "step": 2187
    },
    {
      "epoch": 0.17918086990346918,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2980986123442662e-05,
      "loss": 0.7758611440658569,
      "step": 2188
    },
    {
      "epoch": 0.17926276243998812,
      "grad_norm": 0.19921875,
      "learning_rate": 2.2979977004098658e-05,
      "loss": 0.8872136473655701,
      "step": 2189
    },
    {
      "epoch": 0.1793446549765071,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2978967407519627e-05,
      "loss": 0.7779150009155273,
      "step": 2190
    },
    {
      "epoch": 0.17942654751302603,
      "grad_norm": 0.2294921875,
      "learning_rate": 2.2977957333749455e-05,
      "loss": 1.2085351943969727,
      "step": 2191
    },
    {
      "epoch": 0.179508440049545,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2976946782832044e-05,
      "loss": 0.6555917859077454,
      "step": 2192
    },
    {
      "epoch": 0.17959033258606394,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.2975935754811317e-05,
      "loss": 0.42303186655044556,
      "step": 2193
    },
    {
      "epoch": 0.17967222512258288,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2974924249731213e-05,
      "loss": 0.8524506688117981,
      "step": 2194
    },
    {
      "epoch": 0.17975411765910185,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.29739122676357e-05,
      "loss": 0.8667759895324707,
      "step": 2195
    },
    {
      "epoch": 0.1798360101956208,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.297289980856876e-05,
      "loss": 0.5838576555252075,
      "step": 2196
    },
    {
      "epoch": 0.17991790273213976,
      "grad_norm": 0.169921875,
      "learning_rate": 2.29718868725744e-05,
      "loss": 0.7316719889640808,
      "step": 2197
    },
    {
      "epoch": 0.1799997952686587,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.2970873459696646e-05,
      "loss": 0.9028429388999939,
      "step": 2198
    },
    {
      "epoch": 0.18008168780517766,
      "grad_norm": 0.220703125,
      "learning_rate": 2.2969859569979546e-05,
      "loss": 0.9240031242370605,
      "step": 2199
    },
    {
      "epoch": 0.1801635803416966,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.296884520346717e-05,
      "loss": 0.8532719612121582,
      "step": 2200
    },
    {
      "epoch": 0.18024547287821557,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.29678303602036e-05,
      "loss": 0.8681788444519043,
      "step": 2201
    },
    {
      "epoch": 0.18032736541473451,
      "grad_norm": 0.1484375,
      "learning_rate": 2.2966815040232953e-05,
      "loss": 0.6748262047767639,
      "step": 2202
    },
    {
      "epoch": 0.18040925795125345,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2965799243599354e-05,
      "loss": 0.7944550514221191,
      "step": 2203
    },
    {
      "epoch": 0.18049115048777242,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2964782970346956e-05,
      "loss": 0.8010057210922241,
      "step": 2204
    },
    {
      "epoch": 0.18057304302429136,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.296376622051993e-05,
      "loss": 0.5805017352104187,
      "step": 2205
    },
    {
      "epoch": 0.18065493556081033,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.296274899416247e-05,
      "loss": 0.8926481008529663,
      "step": 2206
    },
    {
      "epoch": 0.18073682809732927,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.296173129131878e-05,
      "loss": 0.8309744596481323,
      "step": 2207
    },
    {
      "epoch": 0.18081872063384824,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.29607131120331e-05,
      "loss": 0.669938862323761,
      "step": 2208
    },
    {
      "epoch": 0.18090061317036718,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.295969445634969e-05,
      "loss": 1.0254864692687988,
      "step": 2209
    },
    {
      "epoch": 0.18098250570688615,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.2958675324312814e-05,
      "loss": 0.7999016046524048,
      "step": 2210
    },
    {
      "epoch": 0.1810643982434051,
      "grad_norm": 0.244140625,
      "learning_rate": 2.2957655715966778e-05,
      "loss": 1.1470184326171875,
      "step": 2211
    },
    {
      "epoch": 0.18114629077992406,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.295663563135589e-05,
      "loss": 0.7634882926940918,
      "step": 2212
    },
    {
      "epoch": 0.181228183316443,
      "grad_norm": 0.1875,
      "learning_rate": 2.2955615070524494e-05,
      "loss": 0.7822745442390442,
      "step": 2213
    },
    {
      "epoch": 0.18131007585296194,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2954594033516937e-05,
      "loss": 0.6191521883010864,
      "step": 2214
    },
    {
      "epoch": 0.1813919683894809,
      "grad_norm": 0.25390625,
      "learning_rate": 2.2953572520377612e-05,
      "loss": 0.9916181564331055,
      "step": 2215
    },
    {
      "epoch": 0.18147386092599985,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2952550531150906e-05,
      "loss": 0.9677515029907227,
      "step": 2216
    },
    {
      "epoch": 0.18155575346251882,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2951528065881245e-05,
      "loss": 0.91322922706604,
      "step": 2217
    },
    {
      "epoch": 0.18163764599903776,
      "grad_norm": 0.1640625,
      "learning_rate": 2.295050512461307e-05,
      "loss": 0.7749010324478149,
      "step": 2218
    },
    {
      "epoch": 0.18171953853555672,
      "grad_norm": 0.232421875,
      "learning_rate": 2.294948170739084e-05,
      "loss": 0.6981696486473083,
      "step": 2219
    },
    {
      "epoch": 0.18180143107207566,
      "grad_norm": 0.1875,
      "learning_rate": 2.2948457814259035e-05,
      "loss": 0.7711498141288757,
      "step": 2220
    },
    {
      "epoch": 0.18188332360859463,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.294743344526216e-05,
      "loss": 1.117043137550354,
      "step": 2221
    },
    {
      "epoch": 0.18196521614511357,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.294640860044474e-05,
      "loss": 0.9432753324508667,
      "step": 2222
    },
    {
      "epoch": 0.18204710868163254,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.294538327985132e-05,
      "loss": 0.634803056716919,
      "step": 2223
    },
    {
      "epoch": 0.18212900121815148,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.2944357483526456e-05,
      "loss": 0.6533849835395813,
      "step": 2224
    },
    {
      "epoch": 0.18221089375467042,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2943331211514744e-05,
      "loss": 0.7691883444786072,
      "step": 2225
    },
    {
      "epoch": 0.1822927862911894,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2942304463860785e-05,
      "loss": 1.0226714611053467,
      "step": 2226
    },
    {
      "epoch": 0.18237467882770833,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.2941277240609203e-05,
      "loss": 0.8920809626579285,
      "step": 2227
    },
    {
      "epoch": 0.1824565713642273,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.294024954180465e-05,
      "loss": 0.744800329208374,
      "step": 2228
    },
    {
      "epoch": 0.18253846390074624,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2939221367491793e-05,
      "loss": 0.8038849830627441,
      "step": 2229
    },
    {
      "epoch": 0.1826203564372652,
      "grad_norm": 0.154296875,
      "learning_rate": 2.293819271771532e-05,
      "loss": 0.8700615167617798,
      "step": 2230
    },
    {
      "epoch": 0.18270224897378415,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2937163592519936e-05,
      "loss": 0.7021470069885254,
      "step": 2231
    },
    {
      "epoch": 0.18278414151030312,
      "grad_norm": 0.18359375,
      "learning_rate": 2.293613399195038e-05,
      "loss": 1.1559795141220093,
      "step": 2232
    },
    {
      "epoch": 0.18286603404682206,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.29351039160514e-05,
      "loss": 0.7895053625106812,
      "step": 2233
    },
    {
      "epoch": 0.182947926583341,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2934073364867764e-05,
      "loss": 0.825248658657074,
      "step": 2234
    },
    {
      "epoch": 0.18302981911985997,
      "grad_norm": 0.150390625,
      "learning_rate": 2.2933042338444263e-05,
      "loss": 1.064957857131958,
      "step": 2235
    },
    {
      "epoch": 0.1831117116563789,
      "grad_norm": 0.1953125,
      "learning_rate": 2.293201083682571e-05,
      "loss": 1.0424407720565796,
      "step": 2236
    },
    {
      "epoch": 0.18319360419289787,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2930978860056944e-05,
      "loss": 0.9745092391967773,
      "step": 2237
    },
    {
      "epoch": 0.18327549672941681,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2929946408182817e-05,
      "loss": 0.7348380088806152,
      "step": 2238
    },
    {
      "epoch": 0.18335738926593578,
      "grad_norm": 0.185546875,
      "learning_rate": 2.29289134812482e-05,
      "loss": 0.8879343867301941,
      "step": 2239
    },
    {
      "epoch": 0.18343928180245472,
      "grad_norm": 0.142578125,
      "learning_rate": 2.2927880079297988e-05,
      "loss": 0.7733120322227478,
      "step": 2240
    },
    {
      "epoch": 0.1835211743389737,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.29268462023771e-05,
      "loss": 0.7583624720573425,
      "step": 2241
    },
    {
      "epoch": 0.18360306687549263,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2925811850530475e-05,
      "loss": 0.9134156703948975,
      "step": 2242
    },
    {
      "epoch": 0.1836849594120116,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.292477702380306e-05,
      "loss": 0.7790586948394775,
      "step": 2243
    },
    {
      "epoch": 0.18376685194853054,
      "grad_norm": 0.177734375,
      "learning_rate": 2.292374172223985e-05,
      "loss": 0.8823917508125305,
      "step": 2244
    },
    {
      "epoch": 0.18384874448504948,
      "grad_norm": 0.19921875,
      "learning_rate": 2.2922705945885826e-05,
      "loss": 0.747046947479248,
      "step": 2245
    },
    {
      "epoch": 0.18393063702156845,
      "grad_norm": 0.18359375,
      "learning_rate": 2.2921669694786015e-05,
      "loss": 0.8483587503433228,
      "step": 2246
    },
    {
      "epoch": 0.1840125295580874,
      "grad_norm": 0.203125,
      "learning_rate": 2.2920632968985457e-05,
      "loss": 0.7791857123374939,
      "step": 2247
    },
    {
      "epoch": 0.18409442209460636,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.291959576852921e-05,
      "loss": 0.819450855255127,
      "step": 2248
    },
    {
      "epoch": 0.1841763146311253,
      "grad_norm": 0.138671875,
      "learning_rate": 2.2918558093462357e-05,
      "loss": 0.5116016268730164,
      "step": 2249
    },
    {
      "epoch": 0.18425820716764427,
      "grad_norm": 0.19140625,
      "learning_rate": 2.291751994383e-05,
      "loss": 0.9610041379928589,
      "step": 2250
    },
    {
      "epoch": 0.1843400997041632,
      "grad_norm": 0.171875,
      "learning_rate": 2.291648131967726e-05,
      "loss": 0.8992875814437866,
      "step": 2251
    },
    {
      "epoch": 0.18442199224068218,
      "grad_norm": 0.197265625,
      "learning_rate": 2.291544222104928e-05,
      "loss": 0.8930961489677429,
      "step": 2252
    },
    {
      "epoch": 0.18450388477720112,
      "grad_norm": 0.212890625,
      "learning_rate": 2.2914402647991224e-05,
      "loss": 1.131935715675354,
      "step": 2253
    },
    {
      "epoch": 0.18458577731372008,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.2913362600548273e-05,
      "loss": 0.9759795665740967,
      "step": 2254
    },
    {
      "epoch": 0.18466766985023902,
      "grad_norm": 0.181640625,
      "learning_rate": 2.291232207876564e-05,
      "loss": 1.1040501594543457,
      "step": 2255
    },
    {
      "epoch": 0.18474956238675797,
      "grad_norm": 0.1953125,
      "learning_rate": 2.291128108268854e-05,
      "loss": 0.8537464141845703,
      "step": 2256
    },
    {
      "epoch": 0.18483145492327693,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.2910239612362225e-05,
      "loss": 0.8829268217086792,
      "step": 2257
    },
    {
      "epoch": 0.18491334745979587,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2909197667831963e-05,
      "loss": 0.8605987429618835,
      "step": 2258
    },
    {
      "epoch": 0.18499523999631484,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2908155249143038e-05,
      "loss": 1.1137299537658691,
      "step": 2259
    },
    {
      "epoch": 0.18507713253283378,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.2907112356340756e-05,
      "loss": 1.041411280632019,
      "step": 2260
    },
    {
      "epoch": 0.18515902506935275,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.2906068989470448e-05,
      "loss": 0.9275497794151306,
      "step": 2261
    },
    {
      "epoch": 0.1852409176058717,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2905025148577467e-05,
      "loss": 0.9737173318862915,
      "step": 2262
    },
    {
      "epoch": 0.18532281014239066,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.2903980833707172e-05,
      "loss": 1.1345387697219849,
      "step": 2263
    },
    {
      "epoch": 0.1854047026789096,
      "grad_norm": 0.1376953125,
      "learning_rate": 2.2902936044904963e-05,
      "loss": 0.5863505601882935,
      "step": 2264
    },
    {
      "epoch": 0.18548659521542854,
      "grad_norm": 0.140625,
      "learning_rate": 2.2901890782216252e-05,
      "loss": 0.8046703338623047,
      "step": 2265
    },
    {
      "epoch": 0.1855684877519475,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.290084504568646e-05,
      "loss": 0.929973840713501,
      "step": 2266
    },
    {
      "epoch": 0.18565038028846645,
      "grad_norm": 0.220703125,
      "learning_rate": 2.289979883536105e-05,
      "loss": 1.036879062652588,
      "step": 2267
    },
    {
      "epoch": 0.18573227282498542,
      "grad_norm": 0.14453125,
      "learning_rate": 2.2898752151285485e-05,
      "loss": 0.45054754614830017,
      "step": 2268
    },
    {
      "epoch": 0.18581416536150436,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2897704993505266e-05,
      "loss": 0.8195974230766296,
      "step": 2269
    },
    {
      "epoch": 0.18589605789802333,
      "grad_norm": 0.1484375,
      "learning_rate": 2.28966573620659e-05,
      "loss": 0.941970705986023,
      "step": 2270
    },
    {
      "epoch": 0.18597795043454227,
      "grad_norm": 0.19140625,
      "learning_rate": 2.2895609257012926e-05,
      "loss": 0.7935249209403992,
      "step": 2271
    },
    {
      "epoch": 0.18605984297106123,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.28945606783919e-05,
      "loss": 0.9533289074897766,
      "step": 2272
    },
    {
      "epoch": 0.18614173550758017,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2893511626248396e-05,
      "loss": 0.830804169178009,
      "step": 2273
    },
    {
      "epoch": 0.18622362804409914,
      "grad_norm": 0.296875,
      "learning_rate": 2.2892462100628006e-05,
      "loss": 0.9977461099624634,
      "step": 2274
    },
    {
      "epoch": 0.18630552058061808,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.2891412101576356e-05,
      "loss": 0.6901416182518005,
      "step": 2275
    },
    {
      "epoch": 0.18638741311713702,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.289036162913907e-05,
      "loss": 1.090146541595459,
      "step": 2276
    },
    {
      "epoch": 0.186469305653656,
      "grad_norm": 0.142578125,
      "learning_rate": 2.288931068336182e-05,
      "loss": 0.8273465633392334,
      "step": 2277
    },
    {
      "epoch": 0.18655119819017493,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.288825926429028e-05,
      "loss": 0.8297091126441956,
      "step": 2278
    },
    {
      "epoch": 0.1866330907266939,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.288720737197014e-05,
      "loss": 0.6991950273513794,
      "step": 2279
    },
    {
      "epoch": 0.18671498326321284,
      "grad_norm": 0.15625,
      "learning_rate": 2.2886155006447125e-05,
      "loss": 0.8491441607475281,
      "step": 2280
    },
    {
      "epoch": 0.1867968757997318,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.2885102167766985e-05,
      "loss": 0.8676788806915283,
      "step": 2281
    },
    {
      "epoch": 0.18687876833625075,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2884048855975465e-05,
      "loss": 0.8516773581504822,
      "step": 2282
    },
    {
      "epoch": 0.18696066087276972,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.2882995071118357e-05,
      "loss": 1.1552330255508423,
      "step": 2283
    },
    {
      "epoch": 0.18704255340928866,
      "grad_norm": 0.173828125,
      "learning_rate": 2.288194081324146e-05,
      "loss": 0.8443632125854492,
      "step": 2284
    },
    {
      "epoch": 0.18712444594580763,
      "grad_norm": 0.181640625,
      "learning_rate": 2.2880886082390595e-05,
      "loss": 0.8985230922698975,
      "step": 2285
    },
    {
      "epoch": 0.18720633848232657,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2879830878611606e-05,
      "loss": 0.9845249652862549,
      "step": 2286
    },
    {
      "epoch": 0.1872882310188455,
      "grad_norm": 0.17578125,
      "learning_rate": 2.2878775201950354e-05,
      "loss": 0.6851732134819031,
      "step": 2287
    },
    {
      "epoch": 0.18737012355536448,
      "grad_norm": 0.15234375,
      "learning_rate": 2.287771905245273e-05,
      "loss": 0.8991727828979492,
      "step": 2288
    },
    {
      "epoch": 0.18745201609188342,
      "grad_norm": 0.19921875,
      "learning_rate": 2.287666243016463e-05,
      "loss": 0.8521350622177124,
      "step": 2289
    },
    {
      "epoch": 0.18753390862840238,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.2875605335131986e-05,
      "loss": 0.8241181969642639,
      "step": 2290
    },
    {
      "epoch": 0.18761580116492133,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.287454776740074e-05,
      "loss": 1.0478544235229492,
      "step": 2291
    },
    {
      "epoch": 0.1876976937014403,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2873489727016856e-05,
      "loss": 0.9343612194061279,
      "step": 2292
    },
    {
      "epoch": 0.18777958623795923,
      "grad_norm": 0.2314453125,
      "learning_rate": 2.287243121402633e-05,
      "loss": 0.8143721222877502,
      "step": 2293
    },
    {
      "epoch": 0.1878614787744782,
      "grad_norm": 0.177734375,
      "learning_rate": 2.287137222847516e-05,
      "loss": 0.730394721031189,
      "step": 2294
    },
    {
      "epoch": 0.18794337131099714,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2870312770409374e-05,
      "loss": 0.9779607653617859,
      "step": 2295
    },
    {
      "epoch": 0.18802526384751608,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.2869252839875024e-05,
      "loss": 0.9047526121139526,
      "step": 2296
    },
    {
      "epoch": 0.18810715638403505,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2868192436918182e-05,
      "loss": 0.6098130941390991,
      "step": 2297
    },
    {
      "epoch": 0.188189048920554,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.2867131561584933e-05,
      "loss": 1.2453804016113281,
      "step": 2298
    },
    {
      "epoch": 0.18827094145707296,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.2866070213921385e-05,
      "loss": 1.0133761167526245,
      "step": 2299
    },
    {
      "epoch": 0.1883528339935919,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2865008393973676e-05,
      "loss": 0.9602872133255005,
      "step": 2300
    },
    {
      "epoch": 0.18843472653011087,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2863946101787948e-05,
      "loss": 0.8568207621574402,
      "step": 2301
    },
    {
      "epoch": 0.1885166190666298,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.2862883337410378e-05,
      "loss": 0.7341924905776978,
      "step": 2302
    },
    {
      "epoch": 0.18859851160314878,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.286182010088716e-05,
      "loss": 0.8664823770523071,
      "step": 2303
    },
    {
      "epoch": 0.18868040413966772,
      "grad_norm": 0.146484375,
      "learning_rate": 2.28607563922645e-05,
      "loss": 0.6416254043579102,
      "step": 2304
    },
    {
      "epoch": 0.18876229667618669,
      "grad_norm": 0.15234375,
      "learning_rate": 2.285969221158864e-05,
      "loss": 0.9313453435897827,
      "step": 2305
    },
    {
      "epoch": 0.18884418921270563,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.2858627558905822e-05,
      "loss": 1.1781384944915771,
      "step": 2306
    },
    {
      "epoch": 0.18892608174922457,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2857562434262332e-05,
      "loss": 0.8042875528335571,
      "step": 2307
    },
    {
      "epoch": 0.18900797428574354,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2856496837704458e-05,
      "loss": 0.8714444041252136,
      "step": 2308
    },
    {
      "epoch": 0.18908986682226248,
      "grad_norm": 0.16796875,
      "learning_rate": 2.285543076927852e-05,
      "loss": 0.8938601613044739,
      "step": 2309
    },
    {
      "epoch": 0.18917175935878144,
      "grad_norm": 0.197265625,
      "learning_rate": 2.2854364229030845e-05,
      "loss": 1.0221461057662964,
      "step": 2310
    },
    {
      "epoch": 0.18925365189530038,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.28532972170078e-05,
      "loss": 0.9000333547592163,
      "step": 2311
    },
    {
      "epoch": 0.18933554443181935,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2852229733255755e-05,
      "loss": 0.7862361669540405,
      "step": 2312
    },
    {
      "epoch": 0.1894174369683383,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2851161777821105e-05,
      "loss": 0.8476905822753906,
      "step": 2313
    },
    {
      "epoch": 0.18949932950485726,
      "grad_norm": 0.203125,
      "learning_rate": 2.2850093350750276e-05,
      "loss": 0.9489936232566833,
      "step": 2314
    },
    {
      "epoch": 0.1895812220413762,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2849024452089697e-05,
      "loss": 0.9766147136688232,
      "step": 2315
    },
    {
      "epoch": 0.18966311457789517,
      "grad_norm": 0.16796875,
      "learning_rate": 2.284795508188584e-05,
      "loss": 0.9581578969955444,
      "step": 2316
    },
    {
      "epoch": 0.1897450071144141,
      "grad_norm": 0.201171875,
      "learning_rate": 2.284688524018517e-05,
      "loss": 0.9424863457679749,
      "step": 2317
    },
    {
      "epoch": 0.18982689965093305,
      "grad_norm": 0.177734375,
      "learning_rate": 2.284581492703419e-05,
      "loss": 0.9911003112792969,
      "step": 2318
    },
    {
      "epoch": 0.18990879218745202,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2844744142479424e-05,
      "loss": 0.9128723740577698,
      "step": 2319
    },
    {
      "epoch": 0.18999068472397096,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.2843672886567413e-05,
      "loss": 0.8795434236526489,
      "step": 2320
    },
    {
      "epoch": 0.19007257726048993,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.284260115934472e-05,
      "loss": 0.9829942584037781,
      "step": 2321
    },
    {
      "epoch": 0.19015446979700887,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2841528960857917e-05,
      "loss": 0.7064555883407593,
      "step": 2322
    },
    {
      "epoch": 0.19023636233352784,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.2840456291153612e-05,
      "loss": 1.0536891222000122,
      "step": 2323
    },
    {
      "epoch": 0.19031825487004678,
      "grad_norm": 0.17578125,
      "learning_rate": 2.283938315027843e-05,
      "loss": 1.0481301546096802,
      "step": 2324
    },
    {
      "epoch": 0.19040014740656575,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.2838309538279015e-05,
      "loss": 0.758525013923645,
      "step": 2325
    },
    {
      "epoch": 0.19048203994308469,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2837235455202026e-05,
      "loss": 0.871162474155426,
      "step": 2326
    },
    {
      "epoch": 0.19056393247960365,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2836160901094148e-05,
      "loss": 0.9626050591468811,
      "step": 2327
    },
    {
      "epoch": 0.1906458250161226,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2835085876002084e-05,
      "loss": 0.7131202816963196,
      "step": 2328
    },
    {
      "epoch": 0.19072771755264153,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2834010379972566e-05,
      "loss": 0.6760917901992798,
      "step": 2329
    },
    {
      "epoch": 0.1908096100891605,
      "grad_norm": 0.181640625,
      "learning_rate": 2.2832934413052333e-05,
      "loss": 0.9866670966148376,
      "step": 2330
    },
    {
      "epoch": 0.19089150262567944,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2831857975288155e-05,
      "loss": 1.0584834814071655,
      "step": 2331
    },
    {
      "epoch": 0.1909733951621984,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2830781066726816e-05,
      "loss": 1.0270410776138306,
      "step": 2332
    },
    {
      "epoch": 0.19105528769871735,
      "grad_norm": 0.1796875,
      "learning_rate": 2.282970368741512e-05,
      "loss": 1.1342506408691406,
      "step": 2333
    },
    {
      "epoch": 0.19113718023523632,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2828625837399898e-05,
      "loss": 0.8717266321182251,
      "step": 2334
    },
    {
      "epoch": 0.19121907277175526,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2827547516727997e-05,
      "loss": 0.6416990160942078,
      "step": 2335
    },
    {
      "epoch": 0.19130096530827423,
      "grad_norm": 0.19921875,
      "learning_rate": 2.2826468725446285e-05,
      "loss": 0.7129364013671875,
      "step": 2336
    },
    {
      "epoch": 0.19138285784479317,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2825389463601654e-05,
      "loss": 0.745063066482544,
      "step": 2337
    },
    {
      "epoch": 0.1914647503813121,
      "grad_norm": 0.1484375,
      "learning_rate": 2.2824309731241008e-05,
      "loss": 0.6407421231269836,
      "step": 2338
    },
    {
      "epoch": 0.19154664291783108,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.282322952841128e-05,
      "loss": 0.6431829333305359,
      "step": 2339
    },
    {
      "epoch": 0.19162853545435002,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.282214885515942e-05,
      "loss": 0.6795263886451721,
      "step": 2340
    },
    {
      "epoch": 0.191710427990869,
      "grad_norm": 0.169921875,
      "learning_rate": 2.28210677115324e-05,
      "loss": 0.9546639919281006,
      "step": 2341
    },
    {
      "epoch": 0.19179232052738793,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.2819986097577203e-05,
      "loss": 0.6644640564918518,
      "step": 2342
    },
    {
      "epoch": 0.1918742130639069,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2818904013340847e-05,
      "loss": 0.7293444275856018,
      "step": 2343
    },
    {
      "epoch": 0.19195610560042584,
      "grad_norm": 0.193359375,
      "learning_rate": 2.2817821458870366e-05,
      "loss": 1.1363742351531982,
      "step": 2344
    },
    {
      "epoch": 0.1920379981369448,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.2816738434212808e-05,
      "loss": 0.7027411460876465,
      "step": 2345
    },
    {
      "epoch": 0.19211989067346374,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2815654939415243e-05,
      "loss": 0.8657201528549194,
      "step": 2346
    },
    {
      "epoch": 0.1922017832099827,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.2814570974524773e-05,
      "loss": 0.7708097696304321,
      "step": 2347
    },
    {
      "epoch": 0.19228367574650165,
      "grad_norm": 0.205078125,
      "learning_rate": 2.2813486539588505e-05,
      "loss": 1.0157450437545776,
      "step": 2348
    },
    {
      "epoch": 0.1923655682830206,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2812401634653572e-05,
      "loss": 1.0896971225738525,
      "step": 2349
    },
    {
      "epoch": 0.19244746081953956,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.2811316259767136e-05,
      "loss": 0.7974168658256531,
      "step": 2350
    },
    {
      "epoch": 0.1925293533560585,
      "grad_norm": 0.16015625,
      "learning_rate": 2.281023041497636e-05,
      "loss": 1.0025204420089722,
      "step": 2351
    },
    {
      "epoch": 0.19261124589257747,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2809144100328452e-05,
      "loss": 1.0021412372589111,
      "step": 2352
    },
    {
      "epoch": 0.1926931384290964,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.280805731587062e-05,
      "loss": 0.6389228105545044,
      "step": 2353
    },
    {
      "epoch": 0.19277503096561538,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2806970061650102e-05,
      "loss": 1.2039275169372559,
      "step": 2354
    },
    {
      "epoch": 0.19285692350213432,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2805882337714152e-05,
      "loss": 0.743495523929596,
      "step": 2355
    },
    {
      "epoch": 0.1929388160386533,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.280479414411005e-05,
      "loss": 0.9987788200378418,
      "step": 2356
    },
    {
      "epoch": 0.19302070857517223,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2803705480885094e-05,
      "loss": 0.7725076079368591,
      "step": 2357
    },
    {
      "epoch": 0.1931026011116912,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2802616348086602e-05,
      "loss": 0.9377445578575134,
      "step": 2358
    },
    {
      "epoch": 0.19318449364821014,
      "grad_norm": 0.1953125,
      "learning_rate": 2.280152674576191e-05,
      "loss": 0.9206058382987976,
      "step": 2359
    },
    {
      "epoch": 0.19326638618472908,
      "grad_norm": 0.28515625,
      "learning_rate": 2.2800436673958378e-05,
      "loss": 0.8345106244087219,
      "step": 2360
    },
    {
      "epoch": 0.19334827872124805,
      "grad_norm": 0.140625,
      "learning_rate": 2.2799346132723384e-05,
      "loss": 0.7060814499855042,
      "step": 2361
    },
    {
      "epoch": 0.193430171257767,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2798255122104324e-05,
      "loss": 0.9959428310394287,
      "step": 2362
    },
    {
      "epoch": 0.19351206379428595,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2797163642148628e-05,
      "loss": 0.9122245907783508,
      "step": 2363
    },
    {
      "epoch": 0.1935939563308049,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2796071692903726e-05,
      "loss": 0.9186090230941772,
      "step": 2364
    },
    {
      "epoch": 0.19367584886732386,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2794979274417086e-05,
      "loss": 0.7945417761802673,
      "step": 2365
    },
    {
      "epoch": 0.1937577414038428,
      "grad_norm": 0.171875,
      "learning_rate": 2.279388638673618e-05,
      "loss": 0.7534151673316956,
      "step": 2366
    },
    {
      "epoch": 0.19383963394036177,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2792793029908517e-05,
      "loss": 0.8852794170379639,
      "step": 2367
    },
    {
      "epoch": 0.1939215264768807,
      "grad_norm": 0.171875,
      "learning_rate": 2.279169920398162e-05,
      "loss": 0.87395840883255,
      "step": 2368
    },
    {
      "epoch": 0.19400341901339965,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2790604909003025e-05,
      "loss": 1.0682847499847412,
      "step": 2369
    },
    {
      "epoch": 0.19408531154991862,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2789510145020297e-05,
      "loss": 0.6849828362464905,
      "step": 2370
    },
    {
      "epoch": 0.19416720408643756,
      "grad_norm": 0.19140625,
      "learning_rate": 2.2788414912081017e-05,
      "loss": 0.8124204874038696,
      "step": 2371
    },
    {
      "epoch": 0.19424909662295653,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2787319210232795e-05,
      "loss": 0.7958555221557617,
      "step": 2372
    },
    {
      "epoch": 0.19433098915947547,
      "grad_norm": 0.19921875,
      "learning_rate": 2.278622303952325e-05,
      "loss": 1.0223562717437744,
      "step": 2373
    },
    {
      "epoch": 0.19441288169599444,
      "grad_norm": 0.208984375,
      "learning_rate": 2.2785126400000028e-05,
      "loss": 0.8524122834205627,
      "step": 2374
    },
    {
      "epoch": 0.19449477423251338,
      "grad_norm": 0.150390625,
      "learning_rate": 2.2784029291710785e-05,
      "loss": 0.9349551200866699,
      "step": 2375
    },
    {
      "epoch": 0.19457666676903235,
      "grad_norm": 0.181640625,
      "learning_rate": 2.278293171470322e-05,
      "loss": 1.0573779344558716,
      "step": 2376
    },
    {
      "epoch": 0.1946585593055513,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.2781833669025026e-05,
      "loss": 1.039783000946045,
      "step": 2377
    },
    {
      "epoch": 0.19474045184207026,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2780735154723937e-05,
      "loss": 1.1323715448379517,
      "step": 2378
    },
    {
      "epoch": 0.1948223443785892,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2779636171847694e-05,
      "loss": 0.8134557008743286,
      "step": 2379
    },
    {
      "epoch": 0.19490423691510814,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.2778536720444067e-05,
      "loss": 0.6291362643241882,
      "step": 2380
    },
    {
      "epoch": 0.1949861294516271,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.277743680056084e-05,
      "loss": 0.7305190563201904,
      "step": 2381
    },
    {
      "epoch": 0.19506802198814605,
      "grad_norm": 0.146484375,
      "learning_rate": 2.277633641224582e-05,
      "loss": 0.9288169145584106,
      "step": 2382
    },
    {
      "epoch": 0.195149914524665,
      "grad_norm": 0.19921875,
      "learning_rate": 2.2775235555546834e-05,
      "loss": 0.918010950088501,
      "step": 2383
    },
    {
      "epoch": 0.19523180706118395,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2774134230511728e-05,
      "loss": 0.7995038032531738,
      "step": 2384
    },
    {
      "epoch": 0.19531369959770292,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.277303243718838e-05,
      "loss": 0.9856702089309692,
      "step": 2385
    },
    {
      "epoch": 0.19539559213422186,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.2771930175624665e-05,
      "loss": 1.0170800685882568,
      "step": 2386
    },
    {
      "epoch": 0.19547748467074083,
      "grad_norm": 0.19140625,
      "learning_rate": 2.2770827445868504e-05,
      "loss": 0.7401501536369324,
      "step": 2387
    },
    {
      "epoch": 0.19555937720725977,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2769724247967817e-05,
      "loss": 0.7248157858848572,
      "step": 2388
    },
    {
      "epoch": 0.19564126974377874,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2768620581970557e-05,
      "loss": 0.8135899901390076,
      "step": 2389
    },
    {
      "epoch": 0.19572316228029768,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.2767516447924692e-05,
      "loss": 0.591747522354126,
      "step": 2390
    },
    {
      "epoch": 0.19580505481681662,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2766411845878217e-05,
      "loss": 0.9034044146537781,
      "step": 2391
    },
    {
      "epoch": 0.1958869473533356,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.276530677587914e-05,
      "loss": 0.6724435091018677,
      "step": 2392
    },
    {
      "epoch": 0.19596883988985453,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.2764201237975493e-05,
      "loss": 0.8551095724105835,
      "step": 2393
    },
    {
      "epoch": 0.1960507324263735,
      "grad_norm": 0.171875,
      "learning_rate": 2.276309523221532e-05,
      "loss": 0.6165861487388611,
      "step": 2394
    },
    {
      "epoch": 0.19613262496289244,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2761988758646703e-05,
      "loss": 1.0519481897354126,
      "step": 2395
    },
    {
      "epoch": 0.1962145174994114,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.276088181731773e-05,
      "loss": 0.8137823939323425,
      "step": 2396
    },
    {
      "epoch": 0.19629641003593035,
      "grad_norm": 0.16796875,
      "learning_rate": 2.275977440827651e-05,
      "loss": 1.119165301322937,
      "step": 2397
    },
    {
      "epoch": 0.19637830257244931,
      "grad_norm": 0.150390625,
      "learning_rate": 2.2758666531571176e-05,
      "loss": 0.9124765396118164,
      "step": 2398
    },
    {
      "epoch": 0.19646019510896826,
      "grad_norm": 0.193359375,
      "learning_rate": 2.2757558187249883e-05,
      "loss": 0.7939457893371582,
      "step": 2399
    },
    {
      "epoch": 0.1965420876454872,
      "grad_norm": 0.169921875,
      "learning_rate": 2.275644937536081e-05,
      "loss": 0.5007310509681702,
      "step": 2400
    },
    {
      "epoch": 0.19662398018200616,
      "grad_norm": 0.14453125,
      "learning_rate": 2.275534009595214e-05,
      "loss": 0.7983091473579407,
      "step": 2401
    },
    {
      "epoch": 0.1967058727185251,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.275423034907209e-05,
      "loss": 0.6883283257484436,
      "step": 2402
    },
    {
      "epoch": 0.19678776525504407,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.2753120134768897e-05,
      "loss": 0.8722920417785645,
      "step": 2403
    },
    {
      "epoch": 0.196869657791563,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2752009453090815e-05,
      "loss": 0.9087334871292114,
      "step": 2404
    },
    {
      "epoch": 0.19695155032808198,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.275089830408612e-05,
      "loss": 0.8070912957191467,
      "step": 2405
    },
    {
      "epoch": 0.19703344286460092,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2749786687803098e-05,
      "loss": 0.6768667697906494,
      "step": 2406
    },
    {
      "epoch": 0.1971153354011199,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.2748674604290076e-05,
      "loss": 0.9066121578216553,
      "step": 2407
    },
    {
      "epoch": 0.19719722793763883,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2747562053595388e-05,
      "loss": 0.799535870552063,
      "step": 2408
    },
    {
      "epoch": 0.1972791204741578,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2746449035767383e-05,
      "loss": 0.7664024233818054,
      "step": 2409
    },
    {
      "epoch": 0.19736101301067674,
      "grad_norm": 0.203125,
      "learning_rate": 2.2745335550854443e-05,
      "loss": 0.8135346174240112,
      "step": 2410
    },
    {
      "epoch": 0.19744290554719568,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.2744221598904965e-05,
      "loss": 1.1084506511688232,
      "step": 2411
    },
    {
      "epoch": 0.19752479808371465,
      "grad_norm": 0.154296875,
      "learning_rate": 2.274310717996736e-05,
      "loss": 0.977137565612793,
      "step": 2412
    },
    {
      "epoch": 0.1976066906202336,
      "grad_norm": 0.14453125,
      "learning_rate": 2.274199229409007e-05,
      "loss": 0.7611242532730103,
      "step": 2413
    },
    {
      "epoch": 0.19768858315675256,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.2740876941321554e-05,
      "loss": 0.8063002824783325,
      "step": 2414
    },
    {
      "epoch": 0.1977704756932715,
      "grad_norm": 0.21484375,
      "learning_rate": 2.2739761121710288e-05,
      "loss": 0.8747677803039551,
      "step": 2415
    },
    {
      "epoch": 0.19785236822979047,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.273864483530477e-05,
      "loss": 0.883658230304718,
      "step": 2416
    },
    {
      "epoch": 0.1979342607663094,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.273752808215352e-05,
      "loss": 0.707085907459259,
      "step": 2417
    },
    {
      "epoch": 0.19801615330282837,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.273641086230507e-05,
      "loss": 1.0529742240905762,
      "step": 2418
    },
    {
      "epoch": 0.19809804583934731,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2735293175807987e-05,
      "loss": 0.8960705995559692,
      "step": 2419
    },
    {
      "epoch": 0.19817993837586628,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2734175022710845e-05,
      "loss": 0.992230236530304,
      "step": 2420
    },
    {
      "epoch": 0.19826183091238522,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2733056403062245e-05,
      "loss": 0.6943228840827942,
      "step": 2421
    },
    {
      "epoch": 0.19834372344890416,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.273193731691081e-05,
      "loss": 0.829267144203186,
      "step": 2422
    },
    {
      "epoch": 0.19842561598542313,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2730817764305176e-05,
      "loss": 1.0084446668624878,
      "step": 2423
    },
    {
      "epoch": 0.19850750852194207,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2729697745294006e-05,
      "loss": 0.667164146900177,
      "step": 2424
    },
    {
      "epoch": 0.19858940105846104,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.272857725992598e-05,
      "loss": 0.8183192014694214,
      "step": 2425
    },
    {
      "epoch": 0.19867129359497998,
      "grad_norm": 0.158203125,
      "learning_rate": 2.27274563082498e-05,
      "loss": 0.924652099609375,
      "step": 2426
    },
    {
      "epoch": 0.19875318613149895,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.2726334890314182e-05,
      "loss": 0.9087849259376526,
      "step": 2427
    },
    {
      "epoch": 0.1988350786680179,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2725213006167873e-05,
      "loss": 1.170980453491211,
      "step": 2428
    },
    {
      "epoch": 0.19891697120453686,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.272409065585963e-05,
      "loss": 0.8187077045440674,
      "step": 2429
    },
    {
      "epoch": 0.1989988637410558,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2722967839438242e-05,
      "loss": 0.9408527612686157,
      "step": 2430
    },
    {
      "epoch": 0.19908075627757477,
      "grad_norm": 0.404296875,
      "learning_rate": 2.2721844556952505e-05,
      "loss": 0.669274091720581,
      "step": 2431
    },
    {
      "epoch": 0.1991626488140937,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2720720808451246e-05,
      "loss": 0.7295891046524048,
      "step": 2432
    },
    {
      "epoch": 0.19924454135061265,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2719596593983302e-05,
      "loss": 0.9046094417572021,
      "step": 2433
    },
    {
      "epoch": 0.19932643388713162,
      "grad_norm": 0.1640625,
      "learning_rate": 2.271847191359754e-05,
      "loss": 0.8711844682693481,
      "step": 2434
    },
    {
      "epoch": 0.19940832642365056,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2717346767342843e-05,
      "loss": 0.6914234757423401,
      "step": 2435
    },
    {
      "epoch": 0.19949021896016952,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.2716221155268114e-05,
      "loss": 1.1778360605239868,
      "step": 2436
    },
    {
      "epoch": 0.19957211149668846,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.271509507742228e-05,
      "loss": 0.7482850551605225,
      "step": 2437
    },
    {
      "epoch": 0.19965400403320743,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2713968533854274e-05,
      "loss": 0.8816819787025452,
      "step": 2438
    },
    {
      "epoch": 0.19973589656972637,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.2712841524613074e-05,
      "loss": 0.7633107900619507,
      "step": 2439
    },
    {
      "epoch": 0.19981778910624534,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2711714049747658e-05,
      "loss": 0.8391861915588379,
      "step": 2440
    },
    {
      "epoch": 0.19989968164276428,
      "grad_norm": 0.142578125,
      "learning_rate": 2.271058610930703e-05,
      "loss": 0.6692081093788147,
      "step": 2441
    },
    {
      "epoch": 0.19998157417928322,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2709457703340217e-05,
      "loss": 0.9222955107688904,
      "step": 2442
    },
    {
      "epoch": 0.2000634667158022,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.270832883189626e-05,
      "loss": 0.5949663519859314,
      "step": 2443
    },
    {
      "epoch": 0.20014535925232113,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.270719949502423e-05,
      "loss": 1.123129963874817,
      "step": 2444
    },
    {
      "epoch": 0.2002272517888401,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.270606969277321e-05,
      "loss": 1.356101632118225,
      "step": 2445
    },
    {
      "epoch": 0.20030914432535904,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2704939425192305e-05,
      "loss": 0.9444726705551147,
      "step": 2446
    },
    {
      "epoch": 0.200391036861878,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2703808692330643e-05,
      "loss": 0.8924837708473206,
      "step": 2447
    },
    {
      "epoch": 0.20047292939839695,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.270267749423737e-05,
      "loss": 0.8812747597694397,
      "step": 2448
    },
    {
      "epoch": 0.20055482193491592,
      "grad_norm": 0.14453125,
      "learning_rate": 2.270154583096165e-05,
      "loss": 0.64066481590271,
      "step": 2449
    },
    {
      "epoch": 0.20063671447143486,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.2700413702552673e-05,
      "loss": 0.7396572232246399,
      "step": 2450
    },
    {
      "epoch": 0.20071860700795383,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.269928110905965e-05,
      "loss": 0.847211480140686,
      "step": 2451
    },
    {
      "epoch": 0.20080049954447277,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.2698148050531794e-05,
      "loss": 0.7291325330734253,
      "step": 2452
    },
    {
      "epoch": 0.2008823920809917,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2697014527018365e-05,
      "loss": 0.7133866548538208,
      "step": 2453
    },
    {
      "epoch": 0.20096428461751067,
      "grad_norm": 0.162109375,
      "learning_rate": 2.269588053856863e-05,
      "loss": 0.984965980052948,
      "step": 2454
    },
    {
      "epoch": 0.20104617715402961,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.269474608523187e-05,
      "loss": 0.935857355594635,
      "step": 2455
    },
    {
      "epoch": 0.20112806969054858,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2693611167057397e-05,
      "loss": 0.7755582928657532,
      "step": 2456
    },
    {
      "epoch": 0.20120996222706752,
      "grad_norm": 0.177734375,
      "learning_rate": 2.269247578409454e-05,
      "loss": 0.7424017190933228,
      "step": 2457
    },
    {
      "epoch": 0.2012918547635865,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2691339936392644e-05,
      "loss": 0.7800498604774475,
      "step": 2458
    },
    {
      "epoch": 0.20137374730010543,
      "grad_norm": 0.16796875,
      "learning_rate": 2.269020362400108e-05,
      "loss": 0.6788142919540405,
      "step": 2459
    },
    {
      "epoch": 0.2014556398366244,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2689066846969243e-05,
      "loss": 0.755730390548706,
      "step": 2460
    },
    {
      "epoch": 0.20153753237314334,
      "grad_norm": 0.20703125,
      "learning_rate": 2.268792960534653e-05,
      "loss": 0.8113637566566467,
      "step": 2461
    },
    {
      "epoch": 0.2016194249096623,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.2686791899182377e-05,
      "loss": 1.0466136932373047,
      "step": 2462
    },
    {
      "epoch": 0.20170131744618125,
      "grad_norm": 0.15234375,
      "learning_rate": 2.268565372852623e-05,
      "loss": 0.8877158761024475,
      "step": 2463
    },
    {
      "epoch": 0.2017832099827002,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.2684515093427565e-05,
      "loss": 0.8467962145805359,
      "step": 2464
    },
    {
      "epoch": 0.20186510251921916,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.268337599393587e-05,
      "loss": 0.807767927646637,
      "step": 2465
    },
    {
      "epoch": 0.2019469950557381,
      "grad_norm": 0.2109375,
      "learning_rate": 2.268223643010065e-05,
      "loss": 0.7990671992301941,
      "step": 2466
    },
    {
      "epoch": 0.20202888759225707,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2681096401971433e-05,
      "loss": 0.8404882550239563,
      "step": 2467
    },
    {
      "epoch": 0.202110780128776,
      "grad_norm": 0.1640625,
      "learning_rate": 2.267995590959778e-05,
      "loss": 0.9492491483688354,
      "step": 2468
    },
    {
      "epoch": 0.20219267266529498,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2678814953029253e-05,
      "loss": 0.8310052156448364,
      "step": 2469
    },
    {
      "epoch": 0.20227456520181392,
      "grad_norm": 0.2109375,
      "learning_rate": 2.2677673532315448e-05,
      "loss": 0.7433747053146362,
      "step": 2470
    },
    {
      "epoch": 0.20235645773833288,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2676531647505968e-05,
      "loss": 0.8398047089576721,
      "step": 2471
    },
    {
      "epoch": 0.20243835027485182,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2675389298650454e-05,
      "loss": 0.6458292603492737,
      "step": 2472
    },
    {
      "epoch": 0.20252024281137077,
      "grad_norm": 0.208984375,
      "learning_rate": 2.2674246485798552e-05,
      "loss": 1.076598882675171,
      "step": 2473
    },
    {
      "epoch": 0.20260213534788973,
      "grad_norm": 0.267578125,
      "learning_rate": 2.267310320899993e-05,
      "loss": 0.8773399591445923,
      "step": 2474
    },
    {
      "epoch": 0.20268402788440867,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.267195946830429e-05,
      "loss": 0.9287348389625549,
      "step": 2475
    },
    {
      "epoch": 0.20276592042092764,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2670815263761334e-05,
      "loss": 0.9702558517456055,
      "step": 2476
    },
    {
      "epoch": 0.20284781295744658,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2669670595420793e-05,
      "loss": 0.705794632434845,
      "step": 2477
    },
    {
      "epoch": 0.20292970549396555,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.2668525463332426e-05,
      "loss": 1.1086801290512085,
      "step": 2478
    },
    {
      "epoch": 0.2030115980304845,
      "grad_norm": 0.1875,
      "learning_rate": 2.2667379867546003e-05,
      "loss": 0.732354998588562,
      "step": 2479
    },
    {
      "epoch": 0.20309349056700346,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.266623380811131e-05,
      "loss": 0.9052433371543884,
      "step": 2480
    },
    {
      "epoch": 0.2031753831035224,
      "grad_norm": 0.1484375,
      "learning_rate": 2.2665087285078166e-05,
      "loss": 0.6810624599456787,
      "step": 2481
    },
    {
      "epoch": 0.20325727564004137,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2663940298496407e-05,
      "loss": 1.0754530429840088,
      "step": 2482
    },
    {
      "epoch": 0.2033391681765603,
      "grad_norm": 0.173828125,
      "learning_rate": 2.266279284841588e-05,
      "loss": 0.7184799909591675,
      "step": 2483
    },
    {
      "epoch": 0.20342106071307925,
      "grad_norm": 0.181640625,
      "learning_rate": 2.2661644934886456e-05,
      "loss": 0.6900075674057007,
      "step": 2484
    },
    {
      "epoch": 0.20350295324959822,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2660496557958038e-05,
      "loss": 0.9767782092094421,
      "step": 2485
    },
    {
      "epoch": 0.20358484578611716,
      "grad_norm": 0.171875,
      "learning_rate": 2.2659347717680523e-05,
      "loss": 1.0323313474655151,
      "step": 2486
    },
    {
      "epoch": 0.20366673832263613,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2658198414103858e-05,
      "loss": 1.0238064527511597,
      "step": 2487
    },
    {
      "epoch": 0.20374863085915507,
      "grad_norm": 0.158203125,
      "learning_rate": 2.2657048647277992e-05,
      "loss": 0.9866200685501099,
      "step": 2488
    },
    {
      "epoch": 0.20383052339567403,
      "grad_norm": 0.33203125,
      "learning_rate": 2.2655898417252898e-05,
      "loss": 0.993322491645813,
      "step": 2489
    },
    {
      "epoch": 0.20391241593219298,
      "grad_norm": 0.1796875,
      "learning_rate": 2.265474772407857e-05,
      "loss": 0.8741251826286316,
      "step": 2490
    },
    {
      "epoch": 0.20399430846871194,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2653596567805025e-05,
      "loss": 0.7923253178596497,
      "step": 2491
    },
    {
      "epoch": 0.20407620100523088,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.2652444948482287e-05,
      "loss": 1.2638981342315674,
      "step": 2492
    },
    {
      "epoch": 0.20415809354174985,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.2651292866160424e-05,
      "loss": 0.9013453125953674,
      "step": 2493
    },
    {
      "epoch": 0.2042399860782688,
      "grad_norm": 0.171875,
      "learning_rate": 2.2650140320889502e-05,
      "loss": 0.840977132320404,
      "step": 2494
    },
    {
      "epoch": 0.20432187861478773,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.2648987312719618e-05,
      "loss": 0.810945987701416,
      "step": 2495
    },
    {
      "epoch": 0.2044037711513067,
      "grad_norm": 0.208984375,
      "learning_rate": 2.2647833841700886e-05,
      "loss": 0.9737492799758911,
      "step": 2496
    },
    {
      "epoch": 0.20448566368782564,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.264667990788344e-05,
      "loss": 0.8026618361473083,
      "step": 2497
    },
    {
      "epoch": 0.2045675562243446,
      "grad_norm": 0.185546875,
      "learning_rate": 2.264552551131743e-05,
      "loss": 0.8607867956161499,
      "step": 2498
    },
    {
      "epoch": 0.20464944876086355,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.264437065205304e-05,
      "loss": 0.7500647902488708,
      "step": 2499
    },
    {
      "epoch": 0.20473134129738252,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.264321533014046e-05,
      "loss": 0.6665752530097961,
      "step": 2500
    },
    {
      "epoch": 0.20481323383390146,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2642059545629903e-05,
      "loss": 1.1682158708572388,
      "step": 2501
    },
    {
      "epoch": 0.20489512637042043,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.2640903298571606e-05,
      "loss": 0.8135837912559509,
      "step": 2502
    },
    {
      "epoch": 0.20497701890693937,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.2639746589015827e-05,
      "loss": 0.5252673029899597,
      "step": 2503
    },
    {
      "epoch": 0.2050589114434583,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.263858941701284e-05,
      "loss": 0.734936535358429,
      "step": 2504
    },
    {
      "epoch": 0.20514080397997728,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.263743178261294e-05,
      "loss": 0.7587753534317017,
      "step": 2505
    },
    {
      "epoch": 0.20522269651649622,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.2636273685866445e-05,
      "loss": 0.9108928442001343,
      "step": 2506
    },
    {
      "epoch": 0.20530458905301519,
      "grad_norm": 0.185546875,
      "learning_rate": 2.2635115126823682e-05,
      "loss": 0.8398345708847046,
      "step": 2507
    },
    {
      "epoch": 0.20538648158953413,
      "grad_norm": 0.208984375,
      "learning_rate": 2.263395610553502e-05,
      "loss": 0.6165756583213806,
      "step": 2508
    },
    {
      "epoch": 0.2054683741260531,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.263279662205082e-05,
      "loss": 1.147895336151123,
      "step": 2509
    },
    {
      "epoch": 0.20555026666257203,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2631636676421488e-05,
      "loss": 0.7744183540344238,
      "step": 2510
    },
    {
      "epoch": 0.205632159199091,
      "grad_norm": 0.1640625,
      "learning_rate": 2.263047626869744e-05,
      "loss": 0.6623430252075195,
      "step": 2511
    },
    {
      "epoch": 0.20571405173560994,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2629315398929108e-05,
      "loss": 0.7683284282684326,
      "step": 2512
    },
    {
      "epoch": 0.2057959442721289,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.262815406716695e-05,
      "loss": 1.0426021814346313,
      "step": 2513
    },
    {
      "epoch": 0.20587783680864785,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.262699227346144e-05,
      "loss": 1.0416150093078613,
      "step": 2514
    },
    {
      "epoch": 0.2059597293451668,
      "grad_norm": 0.146484375,
      "learning_rate": 2.262583001786308e-05,
      "loss": 0.6190228462219238,
      "step": 2515
    },
    {
      "epoch": 0.20604162188168576,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.262466730042238e-05,
      "loss": 0.8024833798408508,
      "step": 2516
    },
    {
      "epoch": 0.2061235144182047,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.262350412118988e-05,
      "loss": 0.9323862195014954,
      "step": 2517
    },
    {
      "epoch": 0.20620540695472367,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.262234048021614e-05,
      "loss": 0.9894610047340393,
      "step": 2518
    },
    {
      "epoch": 0.2062872994912426,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.2621176377551726e-05,
      "loss": 0.9465169906616211,
      "step": 2519
    },
    {
      "epoch": 0.20636919202776158,
      "grad_norm": 0.2470703125,
      "learning_rate": 2.2620011813247246e-05,
      "loss": 0.8565041422843933,
      "step": 2520
    },
    {
      "epoch": 0.20645108456428052,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.261884678735331e-05,
      "loss": 0.7899689674377441,
      "step": 2521
    },
    {
      "epoch": 0.2065329771007995,
      "grad_norm": 0.140625,
      "learning_rate": 2.2617681299920557e-05,
      "loss": 0.6901161670684814,
      "step": 2522
    },
    {
      "epoch": 0.20661486963731843,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2616515350999642e-05,
      "loss": 1.0179699659347534,
      "step": 2523
    },
    {
      "epoch": 0.2066967621738374,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.2615348940641246e-05,
      "loss": 1.3159328699111938,
      "step": 2524
    },
    {
      "epoch": 0.20677865471035634,
      "grad_norm": 0.18359375,
      "learning_rate": 2.2614182068896063e-05,
      "loss": 0.6894040703773499,
      "step": 2525
    },
    {
      "epoch": 0.20686054724687528,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.261301473581481e-05,
      "loss": 0.9576308131217957,
      "step": 2526
    },
    {
      "epoch": 0.20694243978339424,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2611846941448225e-05,
      "loss": 0.7722223997116089,
      "step": 2527
    },
    {
      "epoch": 0.20702433231991318,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2610678685847067e-05,
      "loss": 1.1833792924880981,
      "step": 2528
    },
    {
      "epoch": 0.20710622485643215,
      "grad_norm": 0.18359375,
      "learning_rate": 2.260950996906211e-05,
      "loss": 0.7832878232002258,
      "step": 2529
    },
    {
      "epoch": 0.2071881173929511,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2608340791144153e-05,
      "loss": 1.098849892616272,
      "step": 2530
    },
    {
      "epoch": 0.20727000992947006,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2607171152144007e-05,
      "loss": 0.6746349334716797,
      "step": 2531
    },
    {
      "epoch": 0.207351902465989,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2606001052112522e-05,
      "loss": 1.3599038124084473,
      "step": 2532
    },
    {
      "epoch": 0.20743379500250797,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2604830491100544e-05,
      "loss": 1.0439919233322144,
      "step": 2533
    },
    {
      "epoch": 0.2075156875390269,
      "grad_norm": 0.166015625,
      "learning_rate": 2.260365946915896e-05,
      "loss": 0.6895688772201538,
      "step": 2534
    },
    {
      "epoch": 0.20759758007554585,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.2602487986338656e-05,
      "loss": 0.6307955384254456,
      "step": 2535
    },
    {
      "epoch": 0.20767947261206482,
      "grad_norm": 0.15234375,
      "learning_rate": 2.260131604269056e-05,
      "loss": 0.7039069533348083,
      "step": 2536
    },
    {
      "epoch": 0.20776136514858376,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2600143638265602e-05,
      "loss": 0.7854093909263611,
      "step": 2537
    },
    {
      "epoch": 0.20784325768510273,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.2598970773114746e-05,
      "loss": 0.6994537115097046,
      "step": 2538
    },
    {
      "epoch": 0.20792515022162167,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.2597797447288966e-05,
      "loss": 0.9157559871673584,
      "step": 2539
    },
    {
      "epoch": 0.20800704275814064,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.2596623660839255e-05,
      "loss": 0.6570084095001221,
      "step": 2540
    },
    {
      "epoch": 0.20808893529465958,
      "grad_norm": 0.13671875,
      "learning_rate": 2.259544941381664e-05,
      "loss": 1.0261468887329102,
      "step": 2541
    },
    {
      "epoch": 0.20817082783117855,
      "grad_norm": 0.171875,
      "learning_rate": 2.2594274706272155e-05,
      "loss": 0.6482420563697815,
      "step": 2542
    },
    {
      "epoch": 0.20825272036769749,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2593099538256854e-05,
      "loss": 0.8905519843101501,
      "step": 2543
    },
    {
      "epoch": 0.20833461290421645,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2591923909821815e-05,
      "loss": 0.8548463582992554,
      "step": 2544
    },
    {
      "epoch": 0.2084165054407354,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2590747821018142e-05,
      "loss": 0.8943751454353333,
      "step": 2545
    },
    {
      "epoch": 0.20849839797725433,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2589571271896948e-05,
      "loss": 1.1034404039382935,
      "step": 2546
    },
    {
      "epoch": 0.2085802905137733,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.2588394262509374e-05,
      "loss": 0.9230262041091919,
      "step": 2547
    },
    {
      "epoch": 0.20866218305029224,
      "grad_norm": 0.171875,
      "learning_rate": 2.258721679290657e-05,
      "loss": 0.9269582033157349,
      "step": 2548
    },
    {
      "epoch": 0.2087440755868112,
      "grad_norm": 0.171875,
      "learning_rate": 2.258603886313972e-05,
      "loss": 0.9056403636932373,
      "step": 2549
    },
    {
      "epoch": 0.20882596812333015,
      "grad_norm": 0.14453125,
      "learning_rate": 2.2584860473260017e-05,
      "loss": 0.775119423866272,
      "step": 2550
    },
    {
      "epoch": 0.20890786065984912,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2583681623318688e-05,
      "loss": 0.5513975024223328,
      "step": 2551
    },
    {
      "epoch": 0.20898975319636806,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2582502313366962e-05,
      "loss": 0.8215306997299194,
      "step": 2552
    },
    {
      "epoch": 0.20907164573288703,
      "grad_norm": 0.173828125,
      "learning_rate": 2.25813225434561e-05,
      "loss": 0.9845231175422668,
      "step": 2553
    },
    {
      "epoch": 0.20915353826940597,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2580142313637377e-05,
      "loss": 0.986456036567688,
      "step": 2554
    },
    {
      "epoch": 0.20923543080592494,
      "grad_norm": 0.1484375,
      "learning_rate": 2.2578961623962096e-05,
      "loss": 0.7501486539840698,
      "step": 2555
    },
    {
      "epoch": 0.20931732334244388,
      "grad_norm": 0.171875,
      "learning_rate": 2.257778047448157e-05,
      "loss": 1.0405055284500122,
      "step": 2556
    },
    {
      "epoch": 0.20939921587896282,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2576598865247135e-05,
      "loss": 1.2263362407684326,
      "step": 2557
    },
    {
      "epoch": 0.2094811084154818,
      "grad_norm": 0.146484375,
      "learning_rate": 2.2575416796310156e-05,
      "loss": 0.7137336730957031,
      "step": 2558
    },
    {
      "epoch": 0.20956300095200073,
      "grad_norm": 0.38671875,
      "learning_rate": 2.2574234267722003e-05,
      "loss": 1.294555425643921,
      "step": 2559
    },
    {
      "epoch": 0.2096448934885197,
      "grad_norm": 0.189453125,
      "learning_rate": 2.2573051279534078e-05,
      "loss": 0.9039717316627502,
      "step": 2560
    },
    {
      "epoch": 0.20972678602503864,
      "grad_norm": 0.197265625,
      "learning_rate": 2.2571867831797795e-05,
      "loss": 0.8533040881156921,
      "step": 2561
    },
    {
      "epoch": 0.2098086785615576,
      "grad_norm": 0.189453125,
      "learning_rate": 2.2570683924564598e-05,
      "loss": 0.8020855188369751,
      "step": 2562
    },
    {
      "epoch": 0.20989057109807654,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2569499557885937e-05,
      "loss": 1.0323879718780518,
      "step": 2563
    },
    {
      "epoch": 0.2099724636345955,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2568314731813292e-05,
      "loss": 0.9836834669113159,
      "step": 2564
    },
    {
      "epoch": 0.21005435617111445,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2567129446398163e-05,
      "loss": 0.9234398603439331,
      "step": 2565
    },
    {
      "epoch": 0.21013624870763342,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.2565943701692066e-05,
      "loss": 1.093058705329895,
      "step": 2566
    },
    {
      "epoch": 0.21021814124415236,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.2564757497746535e-05,
      "loss": 0.7526820302009583,
      "step": 2567
    },
    {
      "epoch": 0.2103000337806713,
      "grad_norm": 0.181640625,
      "learning_rate": 2.2563570834613133e-05,
      "loss": 0.7566789984703064,
      "step": 2568
    },
    {
      "epoch": 0.21038192631719027,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2562383712343432e-05,
      "loss": 0.9776017665863037,
      "step": 2569
    },
    {
      "epoch": 0.2104638188537092,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.2561196130989033e-05,
      "loss": 0.7597828507423401,
      "step": 2570
    },
    {
      "epoch": 0.21054571139022818,
      "grad_norm": 0.17578125,
      "learning_rate": 2.2560008090601554e-05,
      "loss": 0.7205164432525635,
      "step": 2571
    },
    {
      "epoch": 0.21062760392674712,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2558819591232626e-05,
      "loss": 1.1665676832199097,
      "step": 2572
    },
    {
      "epoch": 0.2107094964632661,
      "grad_norm": 0.154296875,
      "learning_rate": 2.2557630632933912e-05,
      "loss": 0.7656850814819336,
      "step": 2573
    },
    {
      "epoch": 0.21079138899978503,
      "grad_norm": 0.17578125,
      "learning_rate": 2.2556441215757085e-05,
      "loss": 0.7946710586547852,
      "step": 2574
    },
    {
      "epoch": 0.210873281536304,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2555251339753846e-05,
      "loss": 0.8632849454879761,
      "step": 2575
    },
    {
      "epoch": 0.21095517407282294,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2554061004975907e-05,
      "loss": 1.0044519901275635,
      "step": 2576
    },
    {
      "epoch": 0.21103706660934188,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2552870211475013e-05,
      "loss": 0.8329063653945923,
      "step": 2577
    },
    {
      "epoch": 0.21111895914586085,
      "grad_norm": 0.1484375,
      "learning_rate": 2.2551678959302912e-05,
      "loss": 0.5952067971229553,
      "step": 2578
    },
    {
      "epoch": 0.2112008516823798,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2550487248511384e-05,
      "loss": 0.7586578726768494,
      "step": 2579
    },
    {
      "epoch": 0.21128274421889875,
      "grad_norm": 0.17578125,
      "learning_rate": 2.2549295079152228e-05,
      "loss": 1.0566492080688477,
      "step": 2580
    },
    {
      "epoch": 0.2113646367554177,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2548102451277254e-05,
      "loss": 0.7290369272232056,
      "step": 2581
    },
    {
      "epoch": 0.21144652929193666,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2546909364938308e-05,
      "loss": 0.8774522542953491,
      "step": 2582
    },
    {
      "epoch": 0.2115284218284556,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.254571582018724e-05,
      "loss": 0.6438607573509216,
      "step": 2583
    },
    {
      "epoch": 0.21161031436497457,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2544521817075928e-05,
      "loss": 1.0225238800048828,
      "step": 2584
    },
    {
      "epoch": 0.2116922069014935,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.254332735565627e-05,
      "loss": 0.9594260454177856,
      "step": 2585
    },
    {
      "epoch": 0.21177409943801248,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.2542132435980177e-05,
      "loss": 0.5366291403770447,
      "step": 2586
    },
    {
      "epoch": 0.21185599197453142,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2540937058099587e-05,
      "loss": 0.9230993986129761,
      "step": 2587
    },
    {
      "epoch": 0.21193788451105036,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.253974122206646e-05,
      "loss": 0.7536305785179138,
      "step": 2588
    },
    {
      "epoch": 0.21201977704756933,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.253854492793277e-05,
      "loss": 1.0587902069091797,
      "step": 2589
    },
    {
      "epoch": 0.21210166958408827,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.253734817575051e-05,
      "loss": 0.7969832420349121,
      "step": 2590
    },
    {
      "epoch": 0.21218356212060724,
      "grad_norm": 0.146484375,
      "learning_rate": 2.2536150965571697e-05,
      "loss": 0.6274747848510742,
      "step": 2591
    },
    {
      "epoch": 0.21226545465712618,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.2534953297448374e-05,
      "loss": 0.9164037704467773,
      "step": 2592
    },
    {
      "epoch": 0.21234734719364515,
      "grad_norm": 0.189453125,
      "learning_rate": 2.253375517143259e-05,
      "loss": 0.8369346857070923,
      "step": 2593
    },
    {
      "epoch": 0.2124292397301641,
      "grad_norm": 0.20703125,
      "learning_rate": 2.2532556587576413e-05,
      "loss": 1.0573561191558838,
      "step": 2594
    },
    {
      "epoch": 0.21251113226668306,
      "grad_norm": 0.18359375,
      "learning_rate": 2.2531357545931954e-05,
      "loss": 0.9581342935562134,
      "step": 2595
    },
    {
      "epoch": 0.212593024803202,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.253015804655132e-05,
      "loss": 0.9424036741256714,
      "step": 2596
    },
    {
      "epoch": 0.21267491733972096,
      "grad_norm": 0.171875,
      "learning_rate": 2.252895808948664e-05,
      "loss": 0.8117066621780396,
      "step": 2597
    },
    {
      "epoch": 0.2127568098762399,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2527757674790082e-05,
      "loss": 0.965522050857544,
      "step": 2598
    },
    {
      "epoch": 0.21283870241275885,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2526556802513812e-05,
      "loss": 1.1489461660385132,
      "step": 2599
    },
    {
      "epoch": 0.2129205949492778,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2525355472710033e-05,
      "loss": 0.87331622838974,
      "step": 2600
    },
    {
      "epoch": 0.21300248748579675,
      "grad_norm": 0.19921875,
      "learning_rate": 2.252415368543095e-05,
      "loss": 0.8089415431022644,
      "step": 2601
    },
    {
      "epoch": 0.21308438002231572,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.25229514407288e-05,
      "loss": 1.0025982856750488,
      "step": 2602
    },
    {
      "epoch": 0.21316627255883466,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2521748738655843e-05,
      "loss": 1.007644534111023,
      "step": 2603
    },
    {
      "epoch": 0.21324816509535363,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.252054557926435e-05,
      "loss": 0.8380034565925598,
      "step": 2604
    },
    {
      "epoch": 0.21333005763187257,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2519341962606614e-05,
      "loss": 0.7382635474205017,
      "step": 2605
    },
    {
      "epoch": 0.21341195016839154,
      "grad_norm": 0.177734375,
      "learning_rate": 2.251813788873495e-05,
      "loss": 1.073710322380066,
      "step": 2606
    },
    {
      "epoch": 0.21349384270491048,
      "grad_norm": 0.1376953125,
      "learning_rate": 2.251693335770169e-05,
      "loss": 0.6477277278900146,
      "step": 2607
    },
    {
      "epoch": 0.21357573524142942,
      "grad_norm": 0.15625,
      "learning_rate": 2.2515728369559196e-05,
      "loss": 0.7063530683517456,
      "step": 2608
    },
    {
      "epoch": 0.2136576277779484,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.251452292435983e-05,
      "loss": 1.0643256902694702,
      "step": 2609
    },
    {
      "epoch": 0.21373952031446733,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2513317022155997e-05,
      "loss": 0.8453289270401001,
      "step": 2610
    },
    {
      "epoch": 0.2138214128509863,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.2512110663000103e-05,
      "loss": 0.8432967662811279,
      "step": 2611
    },
    {
      "epoch": 0.21390330538750524,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.251090384694458e-05,
      "loss": 0.935516893863678,
      "step": 2612
    },
    {
      "epoch": 0.2139851979240242,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.2509696574041887e-05,
      "loss": 1.1079294681549072,
      "step": 2613
    },
    {
      "epoch": 0.21406709046054315,
      "grad_norm": 0.171875,
      "learning_rate": 2.2508488844344495e-05,
      "loss": 0.8772408366203308,
      "step": 2614
    },
    {
      "epoch": 0.21414898299706212,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.25072806579049e-05,
      "loss": 0.6262285113334656,
      "step": 2615
    },
    {
      "epoch": 0.21423087553358106,
      "grad_norm": 0.142578125,
      "learning_rate": 2.2506072014775607e-05,
      "loss": 0.6965766549110413,
      "step": 2616
    },
    {
      "epoch": 0.21431276807010002,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2504862915009155e-05,
      "loss": 0.9412310123443604,
      "step": 2617
    },
    {
      "epoch": 0.21439466060661896,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.250365335865809e-05,
      "loss": 0.8304561972618103,
      "step": 2618
    },
    {
      "epoch": 0.2144765531431379,
      "grad_norm": 0.2109375,
      "learning_rate": 2.2502443345774994e-05,
      "loss": 1.1396684646606445,
      "step": 2619
    },
    {
      "epoch": 0.21455844567965687,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.250123287641245e-05,
      "loss": 0.7396496534347534,
      "step": 2620
    },
    {
      "epoch": 0.2146403382161758,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.2500021950623073e-05,
      "loss": 0.5160247087478638,
      "step": 2621
    },
    {
      "epoch": 0.21472223075269478,
      "grad_norm": 0.15625,
      "learning_rate": 2.2498810568459504e-05,
      "loss": 0.82952880859375,
      "step": 2622
    },
    {
      "epoch": 0.21480412328921372,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.2497598729974377e-05,
      "loss": 0.7126262187957764,
      "step": 2623
    },
    {
      "epoch": 0.2148860158257327,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.2496386435220377e-05,
      "loss": 0.997394323348999,
      "step": 2624
    },
    {
      "epoch": 0.21496790836225163,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.249517368425019e-05,
      "loss": 0.6629370450973511,
      "step": 2625
    },
    {
      "epoch": 0.2150498008987706,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2493960477116528e-05,
      "loss": 0.77293461561203,
      "step": 2626
    },
    {
      "epoch": 0.21513169343528954,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.249274681387212e-05,
      "loss": 0.6796474456787109,
      "step": 2627
    },
    {
      "epoch": 0.2152135859718085,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2491532694569724e-05,
      "loss": 0.9979060292243958,
      "step": 2628
    },
    {
      "epoch": 0.21529547850832745,
      "grad_norm": 0.224609375,
      "learning_rate": 2.2490318119262107e-05,
      "loss": 1.1129047870635986,
      "step": 2629
    },
    {
      "epoch": 0.2153773710448464,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2489103088002054e-05,
      "loss": 0.9572575688362122,
      "step": 2630
    },
    {
      "epoch": 0.21545926358136536,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2487887600842374e-05,
      "loss": 0.6022533774375916,
      "step": 2631
    },
    {
      "epoch": 0.2155411561178843,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2486671657835907e-05,
      "loss": 0.9832905530929565,
      "step": 2632
    },
    {
      "epoch": 0.21562304865440327,
      "grad_norm": 0.19140625,
      "learning_rate": 2.2485455259035503e-05,
      "loss": 0.8883419632911682,
      "step": 2633
    },
    {
      "epoch": 0.2157049411909222,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.2484238404494022e-05,
      "loss": 0.985409140586853,
      "step": 2634
    },
    {
      "epoch": 0.21578683372744117,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.248302109426436e-05,
      "loss": 0.9418313503265381,
      "step": 2635
    },
    {
      "epoch": 0.21586872626396011,
      "grad_norm": 0.193359375,
      "learning_rate": 2.2481803328399428e-05,
      "loss": 0.7812432050704956,
      "step": 2636
    },
    {
      "epoch": 0.21595061880047908,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2480585106952145e-05,
      "loss": 1.1868882179260254,
      "step": 2637
    },
    {
      "epoch": 0.21603251133699802,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2479366429975473e-05,
      "loss": 0.7767732739448547,
      "step": 2638
    },
    {
      "epoch": 0.21611440387351696,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2478147297522372e-05,
      "loss": 1.016821265220642,
      "step": 2639
    },
    {
      "epoch": 0.21619629641003593,
      "grad_norm": 0.1640625,
      "learning_rate": 2.247692770964583e-05,
      "loss": 0.6401596069335938,
      "step": 2640
    },
    {
      "epoch": 0.21627818894655487,
      "grad_norm": 0.201171875,
      "learning_rate": 2.2475707666398863e-05,
      "loss": 1.0144199132919312,
      "step": 2641
    },
    {
      "epoch": 0.21636008148307384,
      "grad_norm": 0.189453125,
      "learning_rate": 2.2474487167834494e-05,
      "loss": 0.8511896133422852,
      "step": 2642
    },
    {
      "epoch": 0.21644197401959278,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.2473266214005772e-05,
      "loss": 0.9372079968452454,
      "step": 2643
    },
    {
      "epoch": 0.21652386655611175,
      "grad_norm": 0.185546875,
      "learning_rate": 2.2472044804965762e-05,
      "loss": 0.9081453680992126,
      "step": 2644
    },
    {
      "epoch": 0.2166057590926307,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2470822940767557e-05,
      "loss": 1.0545884370803833,
      "step": 2645
    },
    {
      "epoch": 0.21668765162914966,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2469600621464257e-05,
      "loss": 0.745715320110321,
      "step": 2646
    },
    {
      "epoch": 0.2167695441656686,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.2468377847108994e-05,
      "loss": 0.6323316097259521,
      "step": 2647
    },
    {
      "epoch": 0.21685143670218757,
      "grad_norm": 0.150390625,
      "learning_rate": 2.2467154617754913e-05,
      "loss": 0.6223617196083069,
      "step": 2648
    },
    {
      "epoch": 0.2169333292387065,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.2465930933455183e-05,
      "loss": 0.5440846681594849,
      "step": 2649
    },
    {
      "epoch": 0.21701522177522545,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.246470679426299e-05,
      "loss": 0.8357701897621155,
      "step": 2650
    },
    {
      "epoch": 0.21709711431174442,
      "grad_norm": 0.1484375,
      "learning_rate": 2.2463482200231535e-05,
      "loss": 0.6614250540733337,
      "step": 2651
    },
    {
      "epoch": 0.21717900684826336,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2462257151414047e-05,
      "loss": 1.0901719331741333,
      "step": 2652
    },
    {
      "epoch": 0.21726089938478232,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2461031647863777e-05,
      "loss": 0.8307749032974243,
      "step": 2653
    },
    {
      "epoch": 0.21734279192130126,
      "grad_norm": 0.126953125,
      "learning_rate": 2.2459805689633983e-05,
      "loss": 0.6122291684150696,
      "step": 2654
    },
    {
      "epoch": 0.21742468445782023,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.2458579276777954e-05,
      "loss": 0.8964424133300781,
      "step": 2655
    },
    {
      "epoch": 0.21750657699433917,
      "grad_norm": 0.185546875,
      "learning_rate": 2.2457352409348995e-05,
      "loss": 0.9349625706672668,
      "step": 2656
    },
    {
      "epoch": 0.21758846953085814,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.2456125087400426e-05,
      "loss": 0.8891605734825134,
      "step": 2657
    },
    {
      "epoch": 0.21767036206737708,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.2454897310985595e-05,
      "loss": 0.7247942686080933,
      "step": 2658
    },
    {
      "epoch": 0.21775225460389605,
      "grad_norm": 0.166015625,
      "learning_rate": 2.245366908015787e-05,
      "loss": 0.7170271873474121,
      "step": 2659
    },
    {
      "epoch": 0.217834147140415,
      "grad_norm": 0.197265625,
      "learning_rate": 2.245244039497063e-05,
      "loss": 0.9139279127120972,
      "step": 2660
    },
    {
      "epoch": 0.21791603967693393,
      "grad_norm": 0.154296875,
      "learning_rate": 2.2451211255477276e-05,
      "loss": 0.7368593811988831,
      "step": 2661
    },
    {
      "epoch": 0.2179979322134529,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.244998166173124e-05,
      "loss": 0.7071901559829712,
      "step": 2662
    },
    {
      "epoch": 0.21807982474997184,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2448751613785963e-05,
      "loss": 0.9164714813232422,
      "step": 2663
    },
    {
      "epoch": 0.2181617172864908,
      "grad_norm": 0.166015625,
      "learning_rate": 2.24475211116949e-05,
      "loss": 0.9555971026420593,
      "step": 2664
    },
    {
      "epoch": 0.21824360982300975,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.244629015551154e-05,
      "loss": 0.6782249808311462,
      "step": 2665
    },
    {
      "epoch": 0.21832550235952872,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2445058745289387e-05,
      "loss": 0.7124421000480652,
      "step": 2666
    },
    {
      "epoch": 0.21840739489604766,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.2443826881081962e-05,
      "loss": 0.8786941766738892,
      "step": 2667
    },
    {
      "epoch": 0.21848928743256663,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2442594562942804e-05,
      "loss": 0.8248938918113708,
      "step": 2668
    },
    {
      "epoch": 0.21857117996908557,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2441361790925476e-05,
      "loss": 0.6030952334403992,
      "step": 2669
    },
    {
      "epoch": 0.21865307250560453,
      "grad_norm": 0.15625,
      "learning_rate": 2.244012856508356e-05,
      "loss": 0.8378911018371582,
      "step": 2670
    },
    {
      "epoch": 0.21873496504212347,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.2438894885470655e-05,
      "loss": 0.8841310143470764,
      "step": 2671
    },
    {
      "epoch": 0.21881685757864242,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2437660752140386e-05,
      "loss": 0.7605337500572205,
      "step": 2672
    },
    {
      "epoch": 0.21889875011516138,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2436426165146392e-05,
      "loss": 0.7846240401268005,
      "step": 2673
    },
    {
      "epoch": 0.21898064265168032,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.243519112454233e-05,
      "loss": 0.8841848969459534,
      "step": 2674
    },
    {
      "epoch": 0.2190625351881993,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.243395563038188e-05,
      "loss": 1.184723973274231,
      "step": 2675
    },
    {
      "epoch": 0.21914442772471823,
      "grad_norm": 0.171875,
      "learning_rate": 2.243271968271875e-05,
      "loss": 0.8364683985710144,
      "step": 2676
    },
    {
      "epoch": 0.2192263202612372,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.243148328160665e-05,
      "loss": 1.0725886821746826,
      "step": 2677
    },
    {
      "epoch": 0.21930821279775614,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.2430246427099325e-05,
      "loss": 0.9857078194618225,
      "step": 2678
    },
    {
      "epoch": 0.2193901053342751,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.242900911925053e-05,
      "loss": 0.6420339345932007,
      "step": 2679
    },
    {
      "epoch": 0.21947199787079405,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2427771358114044e-05,
      "loss": 0.6804531812667847,
      "step": 2680
    },
    {
      "epoch": 0.219553890407313,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.2426533143743666e-05,
      "loss": 0.732304573059082,
      "step": 2681
    },
    {
      "epoch": 0.21963578294383196,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2425294476193213e-05,
      "loss": 1.0372018814086914,
      "step": 2682
    },
    {
      "epoch": 0.2197176754803509,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.2424055355516528e-05,
      "loss": 0.6854705214500427,
      "step": 2683
    },
    {
      "epoch": 0.21979956801686987,
      "grad_norm": 0.19140625,
      "learning_rate": 2.2422815781767455e-05,
      "loss": 0.7514453530311584,
      "step": 2684
    },
    {
      "epoch": 0.2198814605533888,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2421575754999887e-05,
      "loss": 0.917513906955719,
      "step": 2685
    },
    {
      "epoch": 0.21996335308990778,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.242033527526771e-05,
      "loss": 0.6444403529167175,
      "step": 2686
    },
    {
      "epoch": 0.22004524562642672,
      "grad_norm": 0.15625,
      "learning_rate": 2.2419094342624845e-05,
      "loss": 0.8434231877326965,
      "step": 2687
    },
    {
      "epoch": 0.22012713816294568,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.241785295712523e-05,
      "loss": 0.8348996043205261,
      "step": 2688
    },
    {
      "epoch": 0.22020903069946463,
      "grad_norm": 0.162109375,
      "learning_rate": 2.241661111882281e-05,
      "loss": 0.6523064970970154,
      "step": 2689
    },
    {
      "epoch": 0.2202909232359836,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2415368827771575e-05,
      "loss": 0.6707596778869629,
      "step": 2690
    },
    {
      "epoch": 0.22037281577250253,
      "grad_norm": 0.16015625,
      "learning_rate": 2.241412608402551e-05,
      "loss": 0.8914279937744141,
      "step": 2691
    },
    {
      "epoch": 0.22045470830902147,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.241288288763863e-05,
      "loss": 1.177140235900879,
      "step": 2692
    },
    {
      "epoch": 0.22053660084554044,
      "grad_norm": 0.162109375,
      "learning_rate": 2.2411639238664974e-05,
      "loss": 0.6267945170402527,
      "step": 2693
    },
    {
      "epoch": 0.22061849338205938,
      "grad_norm": 0.162109375,
      "learning_rate": 2.24103951371586e-05,
      "loss": 1.0307586193084717,
      "step": 2694
    },
    {
      "epoch": 0.22070038591857835,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.240915058317357e-05,
      "loss": 1.268770694732666,
      "step": 2695
    },
    {
      "epoch": 0.2207822784550973,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2407905576763985e-05,
      "loss": 0.752870500087738,
      "step": 2696
    },
    {
      "epoch": 0.22086417099161626,
      "grad_norm": 0.197265625,
      "learning_rate": 2.240666011798396e-05,
      "loss": 1.4423089027404785,
      "step": 2697
    },
    {
      "epoch": 0.2209460635281352,
      "grad_norm": 0.158203125,
      "learning_rate": 2.2405414206887617e-05,
      "loss": 0.8685122132301331,
      "step": 2698
    },
    {
      "epoch": 0.22102795606465417,
      "grad_norm": 0.1796875,
      "learning_rate": 2.240416784352912e-05,
      "loss": 0.9673736691474915,
      "step": 2699
    },
    {
      "epoch": 0.2211098486011731,
      "grad_norm": 0.212890625,
      "learning_rate": 2.2402921027962637e-05,
      "loss": 0.8751057982444763,
      "step": 2700
    },
    {
      "epoch": 0.22119174113769208,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.240167376024236e-05,
      "loss": 0.9190973043441772,
      "step": 2701
    },
    {
      "epoch": 0.22127363367421102,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.2400426040422497e-05,
      "loss": 0.921265721321106,
      "step": 2702
    },
    {
      "epoch": 0.22135552621072996,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2399177868557285e-05,
      "loss": 0.7625551819801331,
      "step": 2703
    },
    {
      "epoch": 0.22143741874724893,
      "grad_norm": 0.158203125,
      "learning_rate": 2.239792924470097e-05,
      "loss": 0.8799278736114502,
      "step": 2704
    },
    {
      "epoch": 0.22151931128376787,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2396680168907826e-05,
      "loss": 0.663055419921875,
      "step": 2705
    },
    {
      "epoch": 0.22160120382028684,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.239543064123214e-05,
      "loss": 0.9900674819946289,
      "step": 2706
    },
    {
      "epoch": 0.22168309635680578,
      "grad_norm": 0.181640625,
      "learning_rate": 2.2394180661728222e-05,
      "loss": 1.0261987447738647,
      "step": 2707
    },
    {
      "epoch": 0.22176498889332474,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.2392930230450403e-05,
      "loss": 1.237915277481079,
      "step": 2708
    },
    {
      "epoch": 0.22184688142984368,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.239167934745303e-05,
      "loss": 0.6792721152305603,
      "step": 2709
    },
    {
      "epoch": 0.22192877396636265,
      "grad_norm": 0.193359375,
      "learning_rate": 2.2390428012790474e-05,
      "loss": 0.9879810810089111,
      "step": 2710
    },
    {
      "epoch": 0.2220106665028816,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.2389176226517122e-05,
      "loss": 1.0300103425979614,
      "step": 2711
    },
    {
      "epoch": 0.22209255903940053,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.238792398868738e-05,
      "loss": 0.7472534775733948,
      "step": 2712
    },
    {
      "epoch": 0.2221744515759195,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2386671299355678e-05,
      "loss": 0.5477824211120605,
      "step": 2713
    },
    {
      "epoch": 0.22225634411243844,
      "grad_norm": 0.1875,
      "learning_rate": 2.2385418158576457e-05,
      "loss": 1.100843071937561,
      "step": 2714
    },
    {
      "epoch": 0.2223382366489574,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2384164566404196e-05,
      "loss": 0.5873090624809265,
      "step": 2715
    },
    {
      "epoch": 0.22242012918547635,
      "grad_norm": 0.1337890625,
      "learning_rate": 2.2382910522893367e-05,
      "loss": 1.061188817024231,
      "step": 2716
    },
    {
      "epoch": 0.22250202172199532,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2381656028098488e-05,
      "loss": 0.6065191626548767,
      "step": 2717
    },
    {
      "epoch": 0.22258391425851426,
      "grad_norm": 0.171875,
      "learning_rate": 2.2380401082074073e-05,
      "loss": 0.9351382255554199,
      "step": 2718
    },
    {
      "epoch": 0.22266580679503323,
      "grad_norm": 0.14453125,
      "learning_rate": 2.2379145684874677e-05,
      "loss": 0.9476308226585388,
      "step": 2719
    },
    {
      "epoch": 0.22274769933155217,
      "grad_norm": 0.283203125,
      "learning_rate": 2.237788983655486e-05,
      "loss": 0.9022799134254456,
      "step": 2720
    },
    {
      "epoch": 0.22282959186807114,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.237663353716921e-05,
      "loss": 0.6997306942939758,
      "step": 2721
    },
    {
      "epoch": 0.22291148440459008,
      "grad_norm": 0.19140625,
      "learning_rate": 2.2375376786772323e-05,
      "loss": 1.0690778493881226,
      "step": 2722
    },
    {
      "epoch": 0.22299337694110902,
      "grad_norm": 0.197265625,
      "learning_rate": 2.2374119585418834e-05,
      "loss": 1.147035837173462,
      "step": 2723
    },
    {
      "epoch": 0.22307526947762799,
      "grad_norm": 0.154296875,
      "learning_rate": 2.2372861933163374e-05,
      "loss": 1.100911021232605,
      "step": 2724
    },
    {
      "epoch": 0.22315716201414693,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2371603830060615e-05,
      "loss": 0.7608529329299927,
      "step": 2725
    },
    {
      "epoch": 0.2232390545506659,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2370345276165238e-05,
      "loss": 0.9179092645645142,
      "step": 2726
    },
    {
      "epoch": 0.22332094708718483,
      "grad_norm": 0.18359375,
      "learning_rate": 2.236908627153194e-05,
      "loss": 0.5465078353881836,
      "step": 2727
    },
    {
      "epoch": 0.2234028396237038,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2367826816215446e-05,
      "loss": 1.1685466766357422,
      "step": 2728
    },
    {
      "epoch": 0.22348473216022274,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.2366566910270496e-05,
      "loss": 0.9647210240364075,
      "step": 2729
    },
    {
      "epoch": 0.2235666246967417,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2365306553751852e-05,
      "loss": 1.0556919574737549,
      "step": 2730
    },
    {
      "epoch": 0.22364851723326065,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.2364045746714293e-05,
      "loss": 0.4899267852306366,
      "step": 2731
    },
    {
      "epoch": 0.22373040976977962,
      "grad_norm": 0.193359375,
      "learning_rate": 2.2362784489212622e-05,
      "loss": 1.1409335136413574,
      "step": 2732
    },
    {
      "epoch": 0.22381230230629856,
      "grad_norm": 0.1875,
      "learning_rate": 2.2361522781301652e-05,
      "loss": 1.1569693088531494,
      "step": 2733
    },
    {
      "epoch": 0.2238941948428175,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.236026062303623e-05,
      "loss": 0.6331688165664673,
      "step": 2734
    },
    {
      "epoch": 0.22397608737933647,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2358998014471207e-05,
      "loss": 0.7054038643836975,
      "step": 2735
    },
    {
      "epoch": 0.2240579799158554,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2357734955661468e-05,
      "loss": 0.6985722780227661,
      "step": 2736
    },
    {
      "epoch": 0.22413987245237438,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2356471446661906e-05,
      "loss": 0.8991851806640625,
      "step": 2737
    },
    {
      "epoch": 0.22422176498889332,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2355207487527443e-05,
      "loss": 1.0830262899398804,
      "step": 2738
    },
    {
      "epoch": 0.2243036575254123,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2353943078313004e-05,
      "loss": 0.8604142069816589,
      "step": 2739
    },
    {
      "epoch": 0.22438555006193123,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.235267821907356e-05,
      "loss": 0.7714955806732178,
      "step": 2740
    },
    {
      "epoch": 0.2244674425984502,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2351412909864083e-05,
      "loss": 1.056908130645752,
      "step": 2741
    },
    {
      "epoch": 0.22454933513496914,
      "grad_norm": 0.244140625,
      "learning_rate": 2.2350147150739562e-05,
      "loss": 0.9743558168411255,
      "step": 2742
    },
    {
      "epoch": 0.22463122767148808,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.2348880941755016e-05,
      "loss": 1.0073190927505493,
      "step": 2743
    },
    {
      "epoch": 0.22471312020800704,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2347614282965485e-05,
      "loss": 0.8281304836273193,
      "step": 2744
    },
    {
      "epoch": 0.22479501274452598,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.2346347174426017e-05,
      "loss": 0.8440684676170349,
      "step": 2745
    },
    {
      "epoch": 0.22487690528104495,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2345079616191685e-05,
      "loss": 0.7128986120223999,
      "step": 2746
    },
    {
      "epoch": 0.2249587978175639,
      "grad_norm": 0.203125,
      "learning_rate": 2.2343811608317587e-05,
      "loss": 0.8777107000350952,
      "step": 2747
    },
    {
      "epoch": 0.22504069035408286,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2342543150858834e-05,
      "loss": 0.9418197870254517,
      "step": 2748
    },
    {
      "epoch": 0.2251225828906018,
      "grad_norm": 0.357421875,
      "learning_rate": 2.2341274243870555e-05,
      "loss": 1.0588223934173584,
      "step": 2749
    },
    {
      "epoch": 0.22520447542712077,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2340004887407908e-05,
      "loss": 0.9309899806976318,
      "step": 2750
    },
    {
      "epoch": 0.2252863679636397,
      "grad_norm": 0.177734375,
      "learning_rate": 2.233873508152606e-05,
      "loss": 0.6725599765777588,
      "step": 2751
    },
    {
      "epoch": 0.22536826050015868,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2337464826280202e-05,
      "loss": 0.6725819706916809,
      "step": 2752
    },
    {
      "epoch": 0.22545015303667762,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2336194121725548e-05,
      "loss": 1.0403375625610352,
      "step": 2753
    },
    {
      "epoch": 0.22553204557319656,
      "grad_norm": 0.232421875,
      "learning_rate": 2.2334922967917324e-05,
      "loss": 0.9659050703048706,
      "step": 2754
    },
    {
      "epoch": 0.22561393810971553,
      "grad_norm": 0.177734375,
      "learning_rate": 2.233365136491078e-05,
      "loss": 1.0475457906723022,
      "step": 2755
    },
    {
      "epoch": 0.22569583064623447,
      "grad_norm": 0.162109375,
      "learning_rate": 2.233237931276119e-05,
      "loss": 1.0404893159866333,
      "step": 2756
    },
    {
      "epoch": 0.22577772318275344,
      "grad_norm": 0.16015625,
      "learning_rate": 2.233110681152384e-05,
      "loss": 0.8932117223739624,
      "step": 2757
    },
    {
      "epoch": 0.22585961571927238,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.2329833861254035e-05,
      "loss": 0.9532628059387207,
      "step": 2758
    },
    {
      "epoch": 0.22594150825579135,
      "grad_norm": 0.18359375,
      "learning_rate": 2.23285604620071e-05,
      "loss": 0.9125658273696899,
      "step": 2759
    },
    {
      "epoch": 0.2260234007923103,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2327286613838394e-05,
      "loss": 0.612433671951294,
      "step": 2760
    },
    {
      "epoch": 0.22610529332882925,
      "grad_norm": 0.189453125,
      "learning_rate": 2.2326012316803274e-05,
      "loss": 0.8008420467376709,
      "step": 2761
    },
    {
      "epoch": 0.2261871858653482,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.232473757095713e-05,
      "loss": 0.9025201797485352,
      "step": 2762
    },
    {
      "epoch": 0.22626907840186716,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.2323462376355366e-05,
      "loss": 0.9112869501113892,
      "step": 2763
    },
    {
      "epoch": 0.2263509709383861,
      "grad_norm": 0.189453125,
      "learning_rate": 2.2322186733053412e-05,
      "loss": 0.8256112337112427,
      "step": 2764
    },
    {
      "epoch": 0.22643286347490504,
      "grad_norm": 0.15234375,
      "learning_rate": 2.23209106411067e-05,
      "loss": 0.9586428999900818,
      "step": 2765
    },
    {
      "epoch": 0.226514756011424,
      "grad_norm": 0.201171875,
      "learning_rate": 2.231963410057071e-05,
      "loss": 0.8691222071647644,
      "step": 2766
    },
    {
      "epoch": 0.22659664854794295,
      "grad_norm": 0.201171875,
      "learning_rate": 2.2318357111500915e-05,
      "loss": 0.6286084055900574,
      "step": 2767
    },
    {
      "epoch": 0.22667854108446192,
      "grad_norm": 0.1484375,
      "learning_rate": 2.2317079673952823e-05,
      "loss": 0.7256220579147339,
      "step": 2768
    },
    {
      "epoch": 0.22676043362098086,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.2315801787981957e-05,
      "loss": 0.8661752343177795,
      "step": 2769
    },
    {
      "epoch": 0.22684232615749983,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2314523453643858e-05,
      "loss": 0.6361514925956726,
      "step": 2770
    },
    {
      "epoch": 0.22692421869401877,
      "grad_norm": 0.140625,
      "learning_rate": 2.2313244670994085e-05,
      "loss": 0.5595276355743408,
      "step": 2771
    },
    {
      "epoch": 0.22700611123053774,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.231196544008822e-05,
      "loss": 0.6491144299507141,
      "step": 2772
    },
    {
      "epoch": 0.22708800376705668,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2310685760981863e-05,
      "loss": 0.8894050121307373,
      "step": 2773
    },
    {
      "epoch": 0.22716989630357562,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.2309405633730644e-05,
      "loss": 1.0037997961044312,
      "step": 2774
    },
    {
      "epoch": 0.2272517888400946,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.230812505839019e-05,
      "loss": 0.7675134539604187,
      "step": 2775
    },
    {
      "epoch": 0.22733368137661353,
      "grad_norm": 0.146484375,
      "learning_rate": 2.2306844035016164e-05,
      "loss": 0.9257134199142456,
      "step": 2776
    },
    {
      "epoch": 0.2274155739131325,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2305562563664253e-05,
      "loss": 1.1266851425170898,
      "step": 2777
    },
    {
      "epoch": 0.22749746644965144,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.2304280644390138e-05,
      "loss": 1.0576081275939941,
      "step": 2778
    },
    {
      "epoch": 0.2275793589861704,
      "grad_norm": 0.138671875,
      "learning_rate": 2.230299827724955e-05,
      "loss": 0.9829504489898682,
      "step": 2779
    },
    {
      "epoch": 0.22766125152268935,
      "grad_norm": 0.158203125,
      "learning_rate": 2.230171546229822e-05,
      "loss": 0.7380350828170776,
      "step": 2780
    },
    {
      "epoch": 0.2277431440592083,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2300432199591908e-05,
      "loss": 1.1664187908172607,
      "step": 2781
    },
    {
      "epoch": 0.22782503659572725,
      "grad_norm": 0.15625,
      "learning_rate": 2.2299148489186392e-05,
      "loss": 0.8917802572250366,
      "step": 2782
    },
    {
      "epoch": 0.22790692913224622,
      "grad_norm": 0.267578125,
      "learning_rate": 2.229786433113746e-05,
      "loss": 0.7070808410644531,
      "step": 2783
    },
    {
      "epoch": 0.22798882166876516,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.229657972550093e-05,
      "loss": 0.7713556289672852,
      "step": 2784
    },
    {
      "epoch": 0.2280707142052841,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2295294672332636e-05,
      "loss": 0.7796047329902649,
      "step": 2785
    },
    {
      "epoch": 0.22815260674180307,
      "grad_norm": 0.158203125,
      "learning_rate": 2.2294009171688433e-05,
      "loss": 0.9818745851516724,
      "step": 2786
    },
    {
      "epoch": 0.228234499278322,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2292723223624196e-05,
      "loss": 0.9776700735092163,
      "step": 2787
    },
    {
      "epoch": 0.22831639181484098,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2291436828195812e-05,
      "loss": 0.7744762301445007,
      "step": 2788
    },
    {
      "epoch": 0.22839828435135992,
      "grad_norm": 0.15625,
      "learning_rate": 2.2290149985459198e-05,
      "loss": 0.9199493527412415,
      "step": 2789
    },
    {
      "epoch": 0.2284801768878789,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.2288862695470278e-05,
      "loss": 0.9536228179931641,
      "step": 2790
    },
    {
      "epoch": 0.22856206942439783,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2287574958285015e-05,
      "loss": 1.0106019973754883,
      "step": 2791
    },
    {
      "epoch": 0.2286439619609168,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2286286773959373e-05,
      "loss": 0.7199668884277344,
      "step": 2792
    },
    {
      "epoch": 0.22872585449743574,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.2284998142549335e-05,
      "loss": 0.8404042720794678,
      "step": 2793
    },
    {
      "epoch": 0.2288077470339547,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2283709064110924e-05,
      "loss": 0.9001615047454834,
      "step": 2794
    },
    {
      "epoch": 0.22888963957047365,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2282419538700158e-05,
      "loss": 0.622333824634552,
      "step": 2795
    },
    {
      "epoch": 0.2289715321069926,
      "grad_norm": 0.16796875,
      "learning_rate": 2.228112956637309e-05,
      "loss": 0.6969611048698425,
      "step": 2796
    },
    {
      "epoch": 0.22905342464351156,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2279839147185788e-05,
      "loss": 1.0442826747894287,
      "step": 2797
    },
    {
      "epoch": 0.2291353171800305,
      "grad_norm": 0.142578125,
      "learning_rate": 2.2278548281194333e-05,
      "loss": 0.7030993103981018,
      "step": 2798
    },
    {
      "epoch": 0.22921720971654946,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.227725696845484e-05,
      "loss": 0.774620771408081,
      "step": 2799
    },
    {
      "epoch": 0.2292991022530684,
      "grad_norm": 0.171875,
      "learning_rate": 2.2275965209023427e-05,
      "loss": 0.902047872543335,
      "step": 2800
    },
    {
      "epoch": 0.22938099478958737,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.2274673002956245e-05,
      "loss": 0.9336604475975037,
      "step": 2801
    },
    {
      "epoch": 0.2294628873261063,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2273380350309453e-05,
      "loss": 0.5391343235969543,
      "step": 2802
    },
    {
      "epoch": 0.22954477986262528,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2272087251139242e-05,
      "loss": 0.8729950189590454,
      "step": 2803
    },
    {
      "epoch": 0.22962667239914422,
      "grad_norm": 0.197265625,
      "learning_rate": 2.2270793705501807e-05,
      "loss": 0.7882866263389587,
      "step": 2804
    },
    {
      "epoch": 0.2297085649356632,
      "grad_norm": 0.154296875,
      "learning_rate": 2.2269499713453376e-05,
      "loss": 0.8446259498596191,
      "step": 2805
    },
    {
      "epoch": 0.22979045747218213,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2268205275050194e-05,
      "loss": 0.7359898090362549,
      "step": 2806
    },
    {
      "epoch": 0.22987235000870107,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2266910390348516e-05,
      "loss": 0.9246798157691956,
      "step": 2807
    },
    {
      "epoch": 0.22995424254522004,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2265615059404623e-05,
      "loss": 0.8838239908218384,
      "step": 2808
    },
    {
      "epoch": 0.23003613508173898,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.2264319282274826e-05,
      "loss": 0.7450317144393921,
      "step": 2809
    },
    {
      "epoch": 0.23011802761825795,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.226302305901543e-05,
      "loss": 0.9908334016799927,
      "step": 2810
    },
    {
      "epoch": 0.2301999201547769,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2261726389682788e-05,
      "loss": 0.7841396927833557,
      "step": 2811
    },
    {
      "epoch": 0.23028181269129586,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2260429274333247e-05,
      "loss": 0.5980224013328552,
      "step": 2812
    },
    {
      "epoch": 0.2303637052278148,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.2259131713023197e-05,
      "loss": 0.7482721209526062,
      "step": 2813
    },
    {
      "epoch": 0.23044559776433376,
      "grad_norm": 0.2265625,
      "learning_rate": 2.2257833705809024e-05,
      "loss": 0.7392801642417908,
      "step": 2814
    },
    {
      "epoch": 0.2305274903008527,
      "grad_norm": 0.1484375,
      "learning_rate": 2.225653525274715e-05,
      "loss": 0.6493318676948547,
      "step": 2815
    },
    {
      "epoch": 0.23060938283737165,
      "grad_norm": 0.197265625,
      "learning_rate": 2.225523635389401e-05,
      "loss": 1.2641602754592896,
      "step": 2816
    },
    {
      "epoch": 0.23069127537389061,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2253937009306058e-05,
      "loss": 1.003706455230713,
      "step": 2817
    },
    {
      "epoch": 0.23077316791040955,
      "grad_norm": 0.193359375,
      "learning_rate": 2.225263721903977e-05,
      "loss": 0.8137313723564148,
      "step": 2818
    },
    {
      "epoch": 0.23085506044692852,
      "grad_norm": 0.15625,
      "learning_rate": 2.2251336983151643e-05,
      "loss": 0.8125075697898865,
      "step": 2819
    },
    {
      "epoch": 0.23093695298344746,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.225003630169819e-05,
      "loss": 0.7304106950759888,
      "step": 2820
    },
    {
      "epoch": 0.23101884551996643,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.224873517473594e-05,
      "loss": 0.6394543051719666,
      "step": 2821
    },
    {
      "epoch": 0.23110073805648537,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2247433602321453e-05,
      "loss": 0.91314297914505,
      "step": 2822
    },
    {
      "epoch": 0.23118263059300434,
      "grad_norm": 0.171875,
      "learning_rate": 2.224613158451129e-05,
      "loss": 1.0550345182418823,
      "step": 2823
    },
    {
      "epoch": 0.23126452312952328,
      "grad_norm": 0.17578125,
      "learning_rate": 2.224482912136205e-05,
      "loss": 1.013972282409668,
      "step": 2824
    },
    {
      "epoch": 0.23134641566604225,
      "grad_norm": 0.171875,
      "learning_rate": 2.224352621293034e-05,
      "loss": 0.8673343062400818,
      "step": 2825
    },
    {
      "epoch": 0.2314283082025612,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.224222285927279e-05,
      "loss": 0.8091303706169128,
      "step": 2826
    },
    {
      "epoch": 0.23151020073908013,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.224091906044605e-05,
      "loss": 0.9118584990501404,
      "step": 2827
    },
    {
      "epoch": 0.2315920932755991,
      "grad_norm": 0.203125,
      "learning_rate": 2.223961481650679e-05,
      "loss": 1.1091703176498413,
      "step": 2828
    },
    {
      "epoch": 0.23167398581211804,
      "grad_norm": 0.15625,
      "learning_rate": 2.2238310127511697e-05,
      "loss": 0.8564403057098389,
      "step": 2829
    },
    {
      "epoch": 0.231755878348637,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2237004993517475e-05,
      "loss": 0.7855836749076843,
      "step": 2830
    },
    {
      "epoch": 0.23183777088515595,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2235699414580857e-05,
      "loss": 0.9884588718414307,
      "step": 2831
    },
    {
      "epoch": 0.23191966342167492,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2234393390758578e-05,
      "loss": 0.9046828746795654,
      "step": 2832
    },
    {
      "epoch": 0.23200155595819386,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.2233086922107416e-05,
      "loss": 0.9615907669067383,
      "step": 2833
    },
    {
      "epoch": 0.23208344849471282,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2231780008684146e-05,
      "loss": 0.8504703044891357,
      "step": 2834
    },
    {
      "epoch": 0.23216534103123176,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2230472650545573e-05,
      "loss": 0.9791901707649231,
      "step": 2835
    },
    {
      "epoch": 0.23224723356775073,
      "grad_norm": 0.130859375,
      "learning_rate": 2.2229164847748526e-05,
      "loss": 0.6011077165603638,
      "step": 2836
    },
    {
      "epoch": 0.23232912610426967,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2227856600349845e-05,
      "loss": 0.8247321844100952,
      "step": 2837
    },
    {
      "epoch": 0.2324110186407886,
      "grad_norm": 0.15625,
      "learning_rate": 2.222654790840639e-05,
      "loss": 0.8194366693496704,
      "step": 2838
    },
    {
      "epoch": 0.23249291117730758,
      "grad_norm": 0.15625,
      "learning_rate": 2.222523877197504e-05,
      "loss": 0.9967739582061768,
      "step": 2839
    },
    {
      "epoch": 0.23257480371382652,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2223929191112698e-05,
      "loss": 0.6677063703536987,
      "step": 2840
    },
    {
      "epoch": 0.2326566962503455,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.222261916587629e-05,
      "loss": 0.6134150624275208,
      "step": 2841
    },
    {
      "epoch": 0.23273858878686443,
      "grad_norm": 0.1875,
      "learning_rate": 2.2221308696322745e-05,
      "loss": 0.6183879375457764,
      "step": 2842
    },
    {
      "epoch": 0.2328204813233834,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2219997782509026e-05,
      "loss": 0.8526945114135742,
      "step": 2843
    },
    {
      "epoch": 0.23290237385990234,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2218686424492106e-05,
      "loss": 0.6396515965461731,
      "step": 2844
    },
    {
      "epoch": 0.2329842663964213,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2217374622328995e-05,
      "loss": 1.0298415422439575,
      "step": 2845
    },
    {
      "epoch": 0.23306615893294025,
      "grad_norm": 0.16015625,
      "learning_rate": 2.2216062376076694e-05,
      "loss": 0.7058196663856506,
      "step": 2846
    },
    {
      "epoch": 0.2331480514694592,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.2214749685792247e-05,
      "loss": 0.612147331237793,
      "step": 2847
    },
    {
      "epoch": 0.23322994400597816,
      "grad_norm": 0.15625,
      "learning_rate": 2.2213436551532708e-05,
      "loss": 0.9517743587493896,
      "step": 2848
    },
    {
      "epoch": 0.2333118365424971,
      "grad_norm": 0.138671875,
      "learning_rate": 2.221212297335515e-05,
      "loss": 0.6100751161575317,
      "step": 2849
    },
    {
      "epoch": 0.23339372907901607,
      "grad_norm": 0.189453125,
      "learning_rate": 2.221080895131667e-05,
      "loss": 0.8263170719146729,
      "step": 2850
    },
    {
      "epoch": 0.233475621615535,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.220949448547437e-05,
      "loss": 0.7124947309494019,
      "step": 2851
    },
    {
      "epoch": 0.23355751415205397,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2208179575885393e-05,
      "loss": 0.7261850237846375,
      "step": 2852
    },
    {
      "epoch": 0.23363940668857291,
      "grad_norm": 0.150390625,
      "learning_rate": 2.2206864222606893e-05,
      "loss": 0.9134925007820129,
      "step": 2853
    },
    {
      "epoch": 0.23372129922509188,
      "grad_norm": 0.16796875,
      "learning_rate": 2.220554842569603e-05,
      "loss": 1.0967222452163696,
      "step": 2854
    },
    {
      "epoch": 0.23380319176161082,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.220423218521e-05,
      "loss": 0.640552818775177,
      "step": 2855
    },
    {
      "epoch": 0.2338850842981298,
      "grad_norm": 0.236328125,
      "learning_rate": 2.2202915501206012e-05,
      "loss": 0.6707224249839783,
      "step": 2856
    },
    {
      "epoch": 0.23396697683464873,
      "grad_norm": 0.201171875,
      "learning_rate": 2.220159837374129e-05,
      "loss": 0.9944537281990051,
      "step": 2857
    },
    {
      "epoch": 0.23404886937116767,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2200280802873088e-05,
      "loss": 0.9138835668563843,
      "step": 2858
    },
    {
      "epoch": 0.23413076190768664,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.219896278865867e-05,
      "loss": 0.8319474458694458,
      "step": 2859
    },
    {
      "epoch": 0.23421265444420558,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2197644331155323e-05,
      "loss": 1.066777229309082,
      "step": 2860
    },
    {
      "epoch": 0.23429454698072455,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2196325430420353e-05,
      "loss": 0.8615840077400208,
      "step": 2861
    },
    {
      "epoch": 0.2343764395172435,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2195006086511083e-05,
      "loss": 0.8138639330863953,
      "step": 2862
    },
    {
      "epoch": 0.23445833205376246,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2193686299484855e-05,
      "loss": 1.0833104848861694,
      "step": 2863
    },
    {
      "epoch": 0.2345402245902814,
      "grad_norm": 0.15625,
      "learning_rate": 2.219236606939904e-05,
      "loss": 0.9868308305740356,
      "step": 2864
    },
    {
      "epoch": 0.23462211712680037,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2191045396311017e-05,
      "loss": 0.7957302331924438,
      "step": 2865
    },
    {
      "epoch": 0.2347040096633193,
      "grad_norm": 0.189453125,
      "learning_rate": 2.2189724280278185e-05,
      "loss": 0.7406501770019531,
      "step": 2866
    },
    {
      "epoch": 0.23478590219983828,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2188402721357972e-05,
      "loss": 0.9510295987129211,
      "step": 2867
    },
    {
      "epoch": 0.23486779473635722,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.218708071960781e-05,
      "loss": 0.681911051273346,
      "step": 2868
    },
    {
      "epoch": 0.23494968727287616,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.218575827508516e-05,
      "loss": 1.0726261138916016,
      "step": 2869
    },
    {
      "epoch": 0.23503157980939512,
      "grad_norm": 0.193359375,
      "learning_rate": 2.2184435387847506e-05,
      "loss": 0.5477421283721924,
      "step": 2870
    },
    {
      "epoch": 0.23511347234591407,
      "grad_norm": 0.150390625,
      "learning_rate": 2.2183112057952345e-05,
      "loss": 0.6962782740592957,
      "step": 2871
    },
    {
      "epoch": 0.23519536488243303,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.2181788285457193e-05,
      "loss": 0.9193621277809143,
      "step": 2872
    },
    {
      "epoch": 0.23527725741895197,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.2180464070419582e-05,
      "loss": 0.9806286096572876,
      "step": 2873
    },
    {
      "epoch": 0.23535914995547094,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2179139412897075e-05,
      "loss": 0.944311797618866,
      "step": 2874
    },
    {
      "epoch": 0.23544104249198988,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.2177814312947244e-05,
      "loss": 0.9483674764633179,
      "step": 2875
    },
    {
      "epoch": 0.23552293502850885,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.2176488770627683e-05,
      "loss": 0.8573991060256958,
      "step": 2876
    },
    {
      "epoch": 0.2356048275650278,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.2175162785996007e-05,
      "loss": 0.9898126125335693,
      "step": 2877
    },
    {
      "epoch": 0.23568672010154673,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2173836359109846e-05,
      "loss": 1.0606895685195923,
      "step": 2878
    },
    {
      "epoch": 0.2357686126380657,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.217250949002686e-05,
      "loss": 0.8710831999778748,
      "step": 2879
    },
    {
      "epoch": 0.23585050517458464,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2171182178804708e-05,
      "loss": 1.0073729753494263,
      "step": 2880
    },
    {
      "epoch": 0.2359323977111036,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.216985442550109e-05,
      "loss": 0.7897433042526245,
      "step": 2881
    },
    {
      "epoch": 0.23601429024762255,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.216852623017371e-05,
      "loss": 0.8032549619674683,
      "step": 2882
    },
    {
      "epoch": 0.23609618278414152,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.2167197592880297e-05,
      "loss": 0.9509630799293518,
      "step": 2883
    },
    {
      "epoch": 0.23617807532066046,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2165868513678604e-05,
      "loss": 0.8557263612747192,
      "step": 2884
    },
    {
      "epoch": 0.23625996785717943,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2164538992626393e-05,
      "loss": 0.7353136539459229,
      "step": 2885
    },
    {
      "epoch": 0.23634186039369837,
      "grad_norm": 0.173828125,
      "learning_rate": 2.2163209029781458e-05,
      "loss": 0.8399332165718079,
      "step": 2886
    },
    {
      "epoch": 0.23642375293021733,
      "grad_norm": 0.146484375,
      "learning_rate": 2.2161878625201593e-05,
      "loss": 0.6546846628189087,
      "step": 2887
    },
    {
      "epoch": 0.23650564546673628,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.216054777894463e-05,
      "loss": 0.745204508304596,
      "step": 2888
    },
    {
      "epoch": 0.23658753800325522,
      "grad_norm": 0.197265625,
      "learning_rate": 2.2159216491068415e-05,
      "loss": 0.8619378209114075,
      "step": 2889
    },
    {
      "epoch": 0.23666943053977418,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2157884761630806e-05,
      "loss": 0.6461140513420105,
      "step": 2890
    },
    {
      "epoch": 0.23675132307629312,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2156552590689694e-05,
      "loss": 0.6375608444213867,
      "step": 2891
    },
    {
      "epoch": 0.2368332156128121,
      "grad_norm": 0.193359375,
      "learning_rate": 2.215521997830297e-05,
      "loss": 0.5877677798271179,
      "step": 2892
    },
    {
      "epoch": 0.23691510814933103,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.215388692452856e-05,
      "loss": 1.1882790327072144,
      "step": 2893
    },
    {
      "epoch": 0.23699700068585,
      "grad_norm": 0.271484375,
      "learning_rate": 2.2152553429424403e-05,
      "loss": 0.9810208678245544,
      "step": 2894
    },
    {
      "epoch": 0.23707889322236894,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2151219493048457e-05,
      "loss": 1.1147767305374146,
      "step": 2895
    },
    {
      "epoch": 0.2371607857588879,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.2149885115458704e-05,
      "loss": 1.024065375328064,
      "step": 2896
    },
    {
      "epoch": 0.23724267829540685,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2148550296713137e-05,
      "loss": 0.6614758968353271,
      "step": 2897
    },
    {
      "epoch": 0.23732457083192582,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.214721503686978e-05,
      "loss": 0.6425355076789856,
      "step": 2898
    },
    {
      "epoch": 0.23740646336844476,
      "grad_norm": 0.189453125,
      "learning_rate": 2.2145879335986664e-05,
      "loss": 0.8511650562286377,
      "step": 2899
    },
    {
      "epoch": 0.2374883559049637,
      "grad_norm": 0.19140625,
      "learning_rate": 2.214454319412184e-05,
      "loss": 1.4089066982269287,
      "step": 2900
    },
    {
      "epoch": 0.23757024844148267,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2143206611333387e-05,
      "loss": 0.752382218837738,
      "step": 2901
    },
    {
      "epoch": 0.2376521409780016,
      "grad_norm": 0.1953125,
      "learning_rate": 2.21418695876794e-05,
      "loss": 0.9225605130195618,
      "step": 2902
    },
    {
      "epoch": 0.23773403351452058,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.214053212321799e-05,
      "loss": 1.0390173196792603,
      "step": 2903
    },
    {
      "epoch": 0.23781592605103952,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.213919421800729e-05,
      "loss": 1.0006986856460571,
      "step": 2904
    },
    {
      "epoch": 0.23789781858755848,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.213785587210544e-05,
      "loss": 0.6326524615287781,
      "step": 2905
    },
    {
      "epoch": 0.23797971112407743,
      "grad_norm": 0.1484375,
      "learning_rate": 2.213651708557063e-05,
      "loss": 0.7818308472633362,
      "step": 2906
    },
    {
      "epoch": 0.2380616036605964,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2135177858461033e-05,
      "loss": 1.0309195518493652,
      "step": 2907
    },
    {
      "epoch": 0.23814349619711533,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.213383819083486e-05,
      "loss": 0.8918837308883667,
      "step": 2908
    },
    {
      "epoch": 0.23822538873363427,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.2132498082750348e-05,
      "loss": 0.8845656514167786,
      "step": 2909
    },
    {
      "epoch": 0.23830728127015324,
      "grad_norm": 0.17578125,
      "learning_rate": 2.213115753426573e-05,
      "loss": 0.7136186957359314,
      "step": 2910
    },
    {
      "epoch": 0.23838917380667218,
      "grad_norm": 0.15625,
      "learning_rate": 2.212981654543928e-05,
      "loss": 0.569214940071106,
      "step": 2911
    },
    {
      "epoch": 0.23847106634319115,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.2128475116329287e-05,
      "loss": 1.0568721294403076,
      "step": 2912
    },
    {
      "epoch": 0.2385529588797101,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2127133246994046e-05,
      "loss": 1.0686583518981934,
      "step": 2913
    },
    {
      "epoch": 0.23863485141622906,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.212579093749188e-05,
      "loss": 0.9004485607147217,
      "step": 2914
    },
    {
      "epoch": 0.238716743952748,
      "grad_norm": 0.1640625,
      "learning_rate": 2.2124448187881136e-05,
      "loss": 0.7355899810791016,
      "step": 2915
    },
    {
      "epoch": 0.23879863648926697,
      "grad_norm": 0.19140625,
      "learning_rate": 2.2123104998220176e-05,
      "loss": 1.1711084842681885,
      "step": 2916
    },
    {
      "epoch": 0.2388805290257859,
      "grad_norm": 0.166015625,
      "learning_rate": 2.212176136856738e-05,
      "loss": 0.8979333639144897,
      "step": 2917
    },
    {
      "epoch": 0.23896242156230488,
      "grad_norm": 0.17578125,
      "learning_rate": 2.2120417298981142e-05,
      "loss": 1.1877660751342773,
      "step": 2918
    },
    {
      "epoch": 0.23904431409882382,
      "grad_norm": 0.19921875,
      "learning_rate": 2.2119072789519888e-05,
      "loss": 0.8069250583648682,
      "step": 2919
    },
    {
      "epoch": 0.23912620663534276,
      "grad_norm": 0.14453125,
      "learning_rate": 2.211772784024205e-05,
      "loss": 0.7692654132843018,
      "step": 2920
    },
    {
      "epoch": 0.23920809917186173,
      "grad_norm": 0.158203125,
      "learning_rate": 2.211638245120609e-05,
      "loss": 0.5090773701667786,
      "step": 2921
    },
    {
      "epoch": 0.23928999170838067,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2115036622470485e-05,
      "loss": 0.6984267830848694,
      "step": 2922
    },
    {
      "epoch": 0.23937188424489964,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.211369035409373e-05,
      "loss": 0.9082513451576233,
      "step": 2923
    },
    {
      "epoch": 0.23945377678141858,
      "grad_norm": 0.171875,
      "learning_rate": 2.211234364613433e-05,
      "loss": 1.2546738386154175,
      "step": 2924
    },
    {
      "epoch": 0.23953566931793754,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.2110996498650825e-05,
      "loss": 1.1569148302078247,
      "step": 2925
    },
    {
      "epoch": 0.23961756185445648,
      "grad_norm": 0.19140625,
      "learning_rate": 2.2109648911701772e-05,
      "loss": 0.8320716619491577,
      "step": 2926
    },
    {
      "epoch": 0.23969945439097545,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.2108300885345738e-05,
      "loss": 0.9242125749588013,
      "step": 2927
    },
    {
      "epoch": 0.2397813469274944,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2106952419641317e-05,
      "loss": 1.0613276958465576,
      "step": 2928
    },
    {
      "epoch": 0.23986323946401336,
      "grad_norm": 0.154296875,
      "learning_rate": 2.2105603514647114e-05,
      "loss": 0.54240882396698,
      "step": 2929
    },
    {
      "epoch": 0.2399451320005323,
      "grad_norm": 0.162109375,
      "learning_rate": 2.210425417042176e-05,
      "loss": 0.6276845335960388,
      "step": 2930
    },
    {
      "epoch": 0.24002702453705124,
      "grad_norm": 0.134765625,
      "learning_rate": 2.2102904387023905e-05,
      "loss": 0.6514420509338379,
      "step": 2931
    },
    {
      "epoch": 0.2401089170735702,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2101554164512212e-05,
      "loss": 0.5951980352401733,
      "step": 2932
    },
    {
      "epoch": 0.24019080961008915,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.210020350294537e-05,
      "loss": 0.647340714931488,
      "step": 2933
    },
    {
      "epoch": 0.24027270214660812,
      "grad_norm": 0.166015625,
      "learning_rate": 2.2098852402382085e-05,
      "loss": 0.6541335582733154,
      "step": 2934
    },
    {
      "epoch": 0.24035459468312706,
      "grad_norm": 0.1640625,
      "learning_rate": 2.209750086288108e-05,
      "loss": 0.8863160014152527,
      "step": 2935
    },
    {
      "epoch": 0.24043648721964603,
      "grad_norm": 0.169921875,
      "learning_rate": 2.20961488845011e-05,
      "loss": 1.0600528717041016,
      "step": 2936
    },
    {
      "epoch": 0.24051837975616497,
      "grad_norm": 0.208984375,
      "learning_rate": 2.2094796467300907e-05,
      "loss": 0.9757941365242004,
      "step": 2937
    },
    {
      "epoch": 0.24060027229268394,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.2093443611339276e-05,
      "loss": 0.9362033605575562,
      "step": 2938
    },
    {
      "epoch": 0.24068216482920288,
      "grad_norm": 0.169921875,
      "learning_rate": 2.209209031667502e-05,
      "loss": 0.9061464667320251,
      "step": 2939
    },
    {
      "epoch": 0.24076405736572185,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2090736583366953e-05,
      "loss": 0.8506525754928589,
      "step": 2940
    },
    {
      "epoch": 0.24084594990224079,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.208938241147391e-05,
      "loss": 0.760705828666687,
      "step": 2941
    },
    {
      "epoch": 0.24092784243875973,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.208802780105476e-05,
      "loss": 1.064037561416626,
      "step": 2942
    },
    {
      "epoch": 0.2410097349752787,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2086672752168364e-05,
      "loss": 0.9620828628540039,
      "step": 2943
    },
    {
      "epoch": 0.24109162751179763,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2085317264873626e-05,
      "loss": 0.9036626815795898,
      "step": 2944
    },
    {
      "epoch": 0.2411735200483166,
      "grad_norm": 0.17578125,
      "learning_rate": 2.2083961339229466e-05,
      "loss": 0.8733800649642944,
      "step": 2945
    },
    {
      "epoch": 0.24125541258483554,
      "grad_norm": 0.16796875,
      "learning_rate": 2.208260497529481e-05,
      "loss": 0.7264404892921448,
      "step": 2946
    },
    {
      "epoch": 0.2413373051213545,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.2081248173128614e-05,
      "loss": 0.9964972734451294,
      "step": 2947
    },
    {
      "epoch": 0.24141919765787345,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2079890932789854e-05,
      "loss": 0.7170913219451904,
      "step": 2948
    },
    {
      "epoch": 0.24150109019439242,
      "grad_norm": 0.119140625,
      "learning_rate": 2.2078533254337518e-05,
      "loss": 0.8381007313728333,
      "step": 2949
    },
    {
      "epoch": 0.24158298273091136,
      "grad_norm": 0.185546875,
      "learning_rate": 2.2077175137830615e-05,
      "loss": 0.7747131586074829,
      "step": 2950
    },
    {
      "epoch": 0.2416648752674303,
      "grad_norm": 0.185546875,
      "learning_rate": 2.2075816583328175e-05,
      "loss": 1.1531018018722534,
      "step": 2951
    },
    {
      "epoch": 0.24174676780394927,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2074457590889245e-05,
      "loss": 1.0712947845458984,
      "step": 2952
    },
    {
      "epoch": 0.2418286603404682,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.20730981605729e-05,
      "loss": 0.7124852538108826,
      "step": 2953
    },
    {
      "epoch": 0.24191055287698718,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.2071738292438218e-05,
      "loss": 0.8471325635910034,
      "step": 2954
    },
    {
      "epoch": 0.24199244541350612,
      "grad_norm": 0.1796875,
      "learning_rate": 2.2070377986544307e-05,
      "loss": 0.7931849360466003,
      "step": 2955
    },
    {
      "epoch": 0.2420743379500251,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2069017242950294e-05,
      "loss": 0.8124366402626038,
      "step": 2956
    },
    {
      "epoch": 0.24215623048654403,
      "grad_norm": 0.197265625,
      "learning_rate": 2.2067656061715317e-05,
      "loss": 0.7925934195518494,
      "step": 2957
    },
    {
      "epoch": 0.242238123023063,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2066294442898542e-05,
      "loss": 0.8002415299415588,
      "step": 2958
    },
    {
      "epoch": 0.24232001555958194,
      "grad_norm": 0.15234375,
      "learning_rate": 2.2064932386559156e-05,
      "loss": 0.7029062509536743,
      "step": 2959
    },
    {
      "epoch": 0.2424019080961009,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2063569892756347e-05,
      "loss": 0.9557847380638123,
      "step": 2960
    },
    {
      "epoch": 0.24248380063261984,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.206220696154935e-05,
      "loss": 1.0886515378952026,
      "step": 2961
    },
    {
      "epoch": 0.24256569316913879,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.2060843592997387e-05,
      "loss": 0.7827026844024658,
      "step": 2962
    },
    {
      "epoch": 0.24264758570565775,
      "grad_norm": 0.18359375,
      "learning_rate": 2.2059479787159726e-05,
      "loss": 0.9219954609870911,
      "step": 2963
    },
    {
      "epoch": 0.2427294782421767,
      "grad_norm": 0.154296875,
      "learning_rate": 2.205811554409564e-05,
      "loss": 0.9839540719985962,
      "step": 2964
    },
    {
      "epoch": 0.24281137077869566,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.205675086386443e-05,
      "loss": 0.5516818165779114,
      "step": 2965
    },
    {
      "epoch": 0.2428932633152146,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2055385746525405e-05,
      "loss": 0.7227625846862793,
      "step": 2966
    },
    {
      "epoch": 0.24297515585173357,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.2054020192137902e-05,
      "loss": 0.827283501625061,
      "step": 2967
    },
    {
      "epoch": 0.2430570483882525,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2052654200761273e-05,
      "loss": 1.0273233652114868,
      "step": 2968
    },
    {
      "epoch": 0.24313894092477148,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.2051287772454886e-05,
      "loss": 0.8882022500038147,
      "step": 2969
    },
    {
      "epoch": 0.24322083346129042,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.204992090727813e-05,
      "loss": 0.6083340644836426,
      "step": 2970
    },
    {
      "epoch": 0.2433027259978094,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2048553605290424e-05,
      "loss": 0.6950846910476685,
      "step": 2971
    },
    {
      "epoch": 0.24338461853432833,
      "grad_norm": 0.236328125,
      "learning_rate": 2.2047185866551192e-05,
      "loss": 1.3416111469268799,
      "step": 2972
    },
    {
      "epoch": 0.24346651107084727,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.204581769111988e-05,
      "loss": 1.0637092590332031,
      "step": 2973
    },
    {
      "epoch": 0.24354840360736624,
      "grad_norm": 0.19921875,
      "learning_rate": 2.2044449079055956e-05,
      "loss": 0.8769762516021729,
      "step": 2974
    },
    {
      "epoch": 0.24363029614388518,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.2043080030418902e-05,
      "loss": 0.9633129835128784,
      "step": 2975
    },
    {
      "epoch": 0.24371218868040415,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.2041710545268227e-05,
      "loss": 1.1436876058578491,
      "step": 2976
    },
    {
      "epoch": 0.2437940812169231,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.2040340623663453e-05,
      "loss": 0.8332898616790771,
      "step": 2977
    },
    {
      "epoch": 0.24387597375344205,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.2038970265664123e-05,
      "loss": 1.0950275659561157,
      "step": 2978
    },
    {
      "epoch": 0.243957866289961,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.2037599471329797e-05,
      "loss": 0.9903711080551147,
      "step": 2979
    },
    {
      "epoch": 0.24403975882647996,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.203622824072006e-05,
      "loss": 0.7225038409233093,
      "step": 2980
    },
    {
      "epoch": 0.2441216513629989,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.20348565738945e-05,
      "loss": 1.003244400024414,
      "step": 2981
    },
    {
      "epoch": 0.24420354389951784,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.2033484470912746e-05,
      "loss": 0.9090712070465088,
      "step": 2982
    },
    {
      "epoch": 0.2442854364360368,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.2032111931834436e-05,
      "loss": 0.8148490190505981,
      "step": 2983
    },
    {
      "epoch": 0.24436732897255575,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.203073895671922e-05,
      "loss": 0.9793577790260315,
      "step": 2984
    },
    {
      "epoch": 0.24444922150907472,
      "grad_norm": 0.146484375,
      "learning_rate": 2.202936554562677e-05,
      "loss": 1.031598687171936,
      "step": 2985
    },
    {
      "epoch": 0.24453111404559366,
      "grad_norm": 0.185546875,
      "learning_rate": 2.202799169861679e-05,
      "loss": 0.7654539346694946,
      "step": 2986
    },
    {
      "epoch": 0.24461300658211263,
      "grad_norm": 0.162109375,
      "learning_rate": 2.2026617415748992e-05,
      "loss": 0.817785382270813,
      "step": 2987
    },
    {
      "epoch": 0.24469489911863157,
      "grad_norm": 0.158203125,
      "learning_rate": 2.20252426970831e-05,
      "loss": 0.6804990172386169,
      "step": 2988
    },
    {
      "epoch": 0.24477679165515054,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.202386754267887e-05,
      "loss": 0.7387655377388,
      "step": 2989
    },
    {
      "epoch": 0.24485868419166948,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.202249195259607e-05,
      "loss": 0.7371788024902344,
      "step": 2990
    },
    {
      "epoch": 0.24494057672818845,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.2021115926894493e-05,
      "loss": 0.8963190913200378,
      "step": 2991
    },
    {
      "epoch": 0.2450224692647074,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.2019739465633942e-05,
      "loss": 0.8483158946037292,
      "step": 2992
    },
    {
      "epoch": 0.24510436180122633,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2018362568874248e-05,
      "loss": 1.0992926359176636,
      "step": 2993
    },
    {
      "epoch": 0.2451862543377453,
      "grad_norm": 0.173828125,
      "learning_rate": 2.201698523667525e-05,
      "loss": 0.8891171216964722,
      "step": 2994
    },
    {
      "epoch": 0.24526814687426424,
      "grad_norm": 0.177734375,
      "learning_rate": 2.2015607469096823e-05,
      "loss": 0.7438278794288635,
      "step": 2995
    },
    {
      "epoch": 0.2453500394107832,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.2014229266198836e-05,
      "loss": 0.8359938859939575,
      "step": 2996
    },
    {
      "epoch": 0.24543193194730215,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.2012850628041205e-05,
      "loss": 0.7275131940841675,
      "step": 2997
    },
    {
      "epoch": 0.2455138244838211,
      "grad_norm": 0.169921875,
      "learning_rate": 2.2011471554683844e-05,
      "loss": 0.9328632354736328,
      "step": 2998
    },
    {
      "epoch": 0.24559571702034005,
      "grad_norm": 0.1484375,
      "learning_rate": 2.2010092046186697e-05,
      "loss": 0.751015841960907,
      "step": 2999
    },
    {
      "epoch": 0.24567760955685902,
      "grad_norm": 0.1953125,
      "learning_rate": 2.2008712102609718e-05,
      "loss": 0.9326849579811096,
      "step": 3000
    },
    {
      "epoch": 0.24575950209337796,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.200733172401289e-05,
      "loss": 0.8515874743461609,
      "step": 3001
    },
    {
      "epoch": 0.24584139462989693,
      "grad_norm": 0.181640625,
      "learning_rate": 2.200595091045621e-05,
      "loss": 0.753883421421051,
      "step": 3002
    },
    {
      "epoch": 0.24592328716641587,
      "grad_norm": 0.23046875,
      "learning_rate": 2.2004569661999685e-05,
      "loss": 1.2156012058258057,
      "step": 3003
    },
    {
      "epoch": 0.2460051797029348,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.2003187978703362e-05,
      "loss": 0.8289654850959778,
      "step": 3004
    },
    {
      "epoch": 0.24608707223945378,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.200180586062729e-05,
      "loss": 0.8761445879936218,
      "step": 3005
    },
    {
      "epoch": 0.24616896477597272,
      "grad_norm": 0.16796875,
      "learning_rate": 2.2000423307831543e-05,
      "loss": 0.9996823668479919,
      "step": 3006
    },
    {
      "epoch": 0.2462508573124917,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.1999040320376206e-05,
      "loss": 0.8873587846755981,
      "step": 3007
    },
    {
      "epoch": 0.24633274984901063,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.1997656898321393e-05,
      "loss": 0.8999268412590027,
      "step": 3008
    },
    {
      "epoch": 0.2464146423855296,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.1996273041727236e-05,
      "loss": 0.7499046325683594,
      "step": 3009
    },
    {
      "epoch": 0.24649653492204854,
      "grad_norm": 0.171875,
      "learning_rate": 2.199488875065388e-05,
      "loss": 0.9111863374710083,
      "step": 3010
    },
    {
      "epoch": 0.2465784274585675,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1993504025161493e-05,
      "loss": 0.8589242696762085,
      "step": 3011
    },
    {
      "epoch": 0.24666031999508645,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1992118865310263e-05,
      "loss": 0.8668841123580933,
      "step": 3012
    },
    {
      "epoch": 0.2467422125316054,
      "grad_norm": 0.220703125,
      "learning_rate": 2.199073327116039e-05,
      "loss": 0.8567924499511719,
      "step": 3013
    },
    {
      "epoch": 0.24682410506812436,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.1989347242772102e-05,
      "loss": 0.8537012338638306,
      "step": 3014
    },
    {
      "epoch": 0.2469059976046433,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.1987960780205637e-05,
      "loss": 0.9299254417419434,
      "step": 3015
    },
    {
      "epoch": 0.24698789014116226,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.1986573883521263e-05,
      "loss": 0.9946805238723755,
      "step": 3016
    },
    {
      "epoch": 0.2470697826776812,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.1985186552779254e-05,
      "loss": 0.880164384841919,
      "step": 3017
    },
    {
      "epoch": 0.24715167521420017,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.1983798788039914e-05,
      "loss": 0.6238075494766235,
      "step": 3018
    },
    {
      "epoch": 0.2472335677507191,
      "grad_norm": 0.224609375,
      "learning_rate": 2.1982410589363555e-05,
      "loss": 0.8949953317642212,
      "step": 3019
    },
    {
      "epoch": 0.24731546028723808,
      "grad_norm": 0.220703125,
      "learning_rate": 2.198102195681052e-05,
      "loss": 0.963131308555603,
      "step": 3020
    },
    {
      "epoch": 0.24739735282375702,
      "grad_norm": 0.162109375,
      "learning_rate": 2.1979632890441156e-05,
      "loss": 1.1852781772613525,
      "step": 3021
    },
    {
      "epoch": 0.247479245360276,
      "grad_norm": 0.2109375,
      "learning_rate": 2.1978243390315852e-05,
      "loss": 1.0087637901306152,
      "step": 3022
    },
    {
      "epoch": 0.24756113789679493,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.1976853456494988e-05,
      "loss": 0.9421628713607788,
      "step": 3023
    },
    {
      "epoch": 0.24764303043331387,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.197546308903898e-05,
      "loss": 0.950789749622345,
      "step": 3024
    },
    {
      "epoch": 0.24772492296983284,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1974072288008265e-05,
      "loss": 0.8576475977897644,
      "step": 3025
    },
    {
      "epoch": 0.24780681550635178,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1972681053463285e-05,
      "loss": 0.9681447148323059,
      "step": 3026
    },
    {
      "epoch": 0.24788870804287075,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.197128938546451e-05,
      "loss": 0.7783282995223999,
      "step": 3027
    },
    {
      "epoch": 0.2479706005793897,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.196989728407243e-05,
      "loss": 0.9873195290565491,
      "step": 3028
    },
    {
      "epoch": 0.24805249311590866,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.1968504749347557e-05,
      "loss": 0.8307504057884216,
      "step": 3029
    },
    {
      "epoch": 0.2481343856524276,
      "grad_norm": 0.18359375,
      "learning_rate": 2.1967111781350406e-05,
      "loss": 1.3429138660430908,
      "step": 3030
    },
    {
      "epoch": 0.24821627818894657,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.1965718380141524e-05,
      "loss": 0.9864434599876404,
      "step": 3031
    },
    {
      "epoch": 0.2482981707254655,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.1964324545781476e-05,
      "loss": 1.0726772546768188,
      "step": 3032
    },
    {
      "epoch": 0.24838006326198447,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.1962930278330847e-05,
      "loss": 0.9629580974578857,
      "step": 3033
    },
    {
      "epoch": 0.24846195579850341,
      "grad_norm": 0.16796875,
      "learning_rate": 2.1961535577850233e-05,
      "loss": 0.9671548008918762,
      "step": 3034
    },
    {
      "epoch": 0.24854384833502235,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.196014044440025e-05,
      "loss": 1.1271796226501465,
      "step": 3035
    },
    {
      "epoch": 0.24862574087154132,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.1958744878041543e-05,
      "loss": 1.174940824508667,
      "step": 3036
    },
    {
      "epoch": 0.24870763340806026,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.1957348878834767e-05,
      "loss": 1.1124440431594849,
      "step": 3037
    },
    {
      "epoch": 0.24878952594457923,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.1955952446840596e-05,
      "loss": 1.0802549123764038,
      "step": 3038
    },
    {
      "epoch": 0.24887141848109817,
      "grad_norm": 0.18359375,
      "learning_rate": 2.195455558211973e-05,
      "loss": 1.1630305051803589,
      "step": 3039
    },
    {
      "epoch": 0.24895331101761714,
      "grad_norm": 0.1796875,
      "learning_rate": 2.195315828473287e-05,
      "loss": 0.9395103454589844,
      "step": 3040
    },
    {
      "epoch": 0.24903520355413608,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.1951760554740765e-05,
      "loss": 0.8728451728820801,
      "step": 3041
    },
    {
      "epoch": 0.24911709609065505,
      "grad_norm": 0.19140625,
      "learning_rate": 2.1950362392204154e-05,
      "loss": 0.980659008026123,
      "step": 3042
    },
    {
      "epoch": 0.249198988627174,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.1948963797183813e-05,
      "loss": 0.9050116539001465,
      "step": 3043
    },
    {
      "epoch": 0.24928088116369296,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.1947564769740528e-05,
      "loss": 1.059549331665039,
      "step": 3044
    },
    {
      "epoch": 0.2493627737002119,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.1946165309935106e-05,
      "loss": 0.8213518261909485,
      "step": 3045
    },
    {
      "epoch": 0.24944466623673084,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1944765417828374e-05,
      "loss": 1.1724355220794678,
      "step": 3046
    },
    {
      "epoch": 0.2495265587732498,
      "grad_norm": 0.1953125,
      "learning_rate": 2.1943365093481183e-05,
      "loss": 0.6799591779708862,
      "step": 3047
    },
    {
      "epoch": 0.24960845130976875,
      "grad_norm": 0.162109375,
      "learning_rate": 2.1941964336954385e-05,
      "loss": 0.9837350845336914,
      "step": 3048
    },
    {
      "epoch": 0.24969034384628772,
      "grad_norm": 0.1953125,
      "learning_rate": 2.1940563148308872e-05,
      "loss": 0.8224217295646667,
      "step": 3049
    },
    {
      "epoch": 0.24977223638280666,
      "grad_norm": 0.185546875,
      "learning_rate": 2.1939161527605542e-05,
      "loss": 1.083338737487793,
      "step": 3050
    },
    {
      "epoch": 0.24985412891932562,
      "grad_norm": 0.19140625,
      "learning_rate": 2.1937759474905315e-05,
      "loss": 0.946805477142334,
      "step": 3051
    },
    {
      "epoch": 0.24993602145584456,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.193635699026913e-05,
      "loss": 0.6161782741546631,
      "step": 3052
    },
    {
      "epoch": 0.2500179139923635,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.1934954073757945e-05,
      "loss": 0.9599847793579102,
      "step": 3053
    },
    {
      "epoch": 0.2500998065288825,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.193355072543274e-05,
      "loss": 0.5544872283935547,
      "step": 3054
    },
    {
      "epoch": 0.25018169906540144,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.19321469453545e-05,
      "loss": 0.7686026096343994,
      "step": 3055
    },
    {
      "epoch": 0.25026359160192035,
      "grad_norm": 0.150390625,
      "learning_rate": 2.1930742733584256e-05,
      "loss": 0.689558207988739,
      "step": 3056
    },
    {
      "epoch": 0.2503454841384393,
      "grad_norm": 0.1484375,
      "learning_rate": 2.1929338090183026e-05,
      "loss": 0.9226515293121338,
      "step": 3057
    },
    {
      "epoch": 0.2504273766749583,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.192793301521187e-05,
      "loss": 0.9194018244743347,
      "step": 3058
    },
    {
      "epoch": 0.25050926921147726,
      "grad_norm": 0.15625,
      "learning_rate": 2.192652750873185e-05,
      "loss": 1.065134048461914,
      "step": 3059
    },
    {
      "epoch": 0.25059116174799617,
      "grad_norm": 0.181640625,
      "learning_rate": 2.1925121570804063e-05,
      "loss": 1.0438369512557983,
      "step": 3060
    },
    {
      "epoch": 0.25067305428451514,
      "grad_norm": 0.19140625,
      "learning_rate": 2.192371520148961e-05,
      "loss": 0.9493274092674255,
      "step": 3061
    },
    {
      "epoch": 0.2507549468210341,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.192230840084963e-05,
      "loss": 0.7314890623092651,
      "step": 3062
    },
    {
      "epoch": 0.2508368393575531,
      "grad_norm": 0.189453125,
      "learning_rate": 2.1920901168945253e-05,
      "loss": 0.7952975630760193,
      "step": 3063
    },
    {
      "epoch": 0.250918731894072,
      "grad_norm": 0.14453125,
      "learning_rate": 2.191949350583765e-05,
      "loss": 0.8662047386169434,
      "step": 3064
    },
    {
      "epoch": 0.25100062443059096,
      "grad_norm": 0.16015625,
      "learning_rate": 2.1918085411588005e-05,
      "loss": 0.8837013840675354,
      "step": 3065
    },
    {
      "epoch": 0.2510825169671099,
      "grad_norm": 0.16015625,
      "learning_rate": 2.191667688625752e-05,
      "loss": 1.0853338241577148,
      "step": 3066
    },
    {
      "epoch": 0.25116440950362884,
      "grad_norm": 0.203125,
      "learning_rate": 2.1915267929907413e-05,
      "loss": 0.8274083137512207,
      "step": 3067
    },
    {
      "epoch": 0.2512463020401478,
      "grad_norm": 0.16015625,
      "learning_rate": 2.1913858542598924e-05,
      "loss": 0.8375928401947021,
      "step": 3068
    },
    {
      "epoch": 0.2513281945766668,
      "grad_norm": 0.177734375,
      "learning_rate": 2.191244872439331e-05,
      "loss": 0.8726490139961243,
      "step": 3069
    },
    {
      "epoch": 0.25141008711318574,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.191103847535185e-05,
      "loss": 0.7674068212509155,
      "step": 3070
    },
    {
      "epoch": 0.25149197964970466,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1909627795535833e-05,
      "loss": 0.8712756633758545,
      "step": 3071
    },
    {
      "epoch": 0.2515738721862236,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1908216685006578e-05,
      "loss": 0.7865182757377625,
      "step": 3072
    },
    {
      "epoch": 0.2516557647227426,
      "grad_norm": 0.208984375,
      "learning_rate": 2.1906805143825417e-05,
      "loss": 0.9612048864364624,
      "step": 3073
    },
    {
      "epoch": 0.25173765725926156,
      "grad_norm": 0.15234375,
      "learning_rate": 2.1905393172053706e-05,
      "loss": 0.7275028824806213,
      "step": 3074
    },
    {
      "epoch": 0.2518195497957805,
      "grad_norm": 0.20703125,
      "learning_rate": 2.1903980769752806e-05,
      "loss": 1.0381853580474854,
      "step": 3075
    },
    {
      "epoch": 0.25190144233229944,
      "grad_norm": 0.154296875,
      "learning_rate": 2.1902567936984115e-05,
      "loss": 0.6771314740180969,
      "step": 3076
    },
    {
      "epoch": 0.2519833348688184,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.1901154673809028e-05,
      "loss": 0.7741199135780334,
      "step": 3077
    },
    {
      "epoch": 0.2520652274053373,
      "grad_norm": 0.208984375,
      "learning_rate": 2.1899740980288983e-05,
      "loss": 1.65373694896698,
      "step": 3078
    },
    {
      "epoch": 0.2521471199418563,
      "grad_norm": 0.19140625,
      "learning_rate": 2.1898326856485423e-05,
      "loss": 0.7512911558151245,
      "step": 3079
    },
    {
      "epoch": 0.25222901247837526,
      "grad_norm": 0.1875,
      "learning_rate": 2.189691230245981e-05,
      "loss": 0.8491021990776062,
      "step": 3080
    },
    {
      "epoch": 0.2523109050148942,
      "grad_norm": 0.171875,
      "learning_rate": 2.1895497318273622e-05,
      "loss": 1.0949782133102417,
      "step": 3081
    },
    {
      "epoch": 0.25239279755141314,
      "grad_norm": 0.17578125,
      "learning_rate": 2.1894081903988367e-05,
      "loss": 0.7864523530006409,
      "step": 3082
    },
    {
      "epoch": 0.2524746900879321,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.1892666059665558e-05,
      "loss": 0.9498428106307983,
      "step": 3083
    },
    {
      "epoch": 0.2525565826244511,
      "grad_norm": 0.169921875,
      "learning_rate": 2.189124978536674e-05,
      "loss": 0.7808324098587036,
      "step": 3084
    },
    {
      "epoch": 0.25263847516097004,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.1889833081153466e-05,
      "loss": 0.7288825511932373,
      "step": 3085
    },
    {
      "epoch": 0.25272036769748896,
      "grad_norm": 0.169921875,
      "learning_rate": 2.1888415947087315e-05,
      "loss": 0.7504899501800537,
      "step": 3086
    },
    {
      "epoch": 0.2528022602340079,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.188699838322988e-05,
      "loss": 0.8437036871910095,
      "step": 3087
    },
    {
      "epoch": 0.2528841527705269,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.1885580389642774e-05,
      "loss": 0.6429985761642456,
      "step": 3088
    },
    {
      "epoch": 0.2529660453070458,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.1884161966387624e-05,
      "loss": 0.818250834941864,
      "step": 3089
    },
    {
      "epoch": 0.2530479378435648,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.1882743113526085e-05,
      "loss": 1.0685276985168457,
      "step": 3090
    },
    {
      "epoch": 0.25312983038008374,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.1881323831119834e-05,
      "loss": 0.7248176336288452,
      "step": 3091
    },
    {
      "epoch": 0.2532117229166027,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1879904119230544e-05,
      "loss": 1.0803157091140747,
      "step": 3092
    },
    {
      "epoch": 0.2532936154531216,
      "grad_norm": 0.208984375,
      "learning_rate": 2.1878483977919928e-05,
      "loss": 1.0954208374023438,
      "step": 3093
    },
    {
      "epoch": 0.2533755079896406,
      "grad_norm": 0.1953125,
      "learning_rate": 2.1877063407249717e-05,
      "loss": 1.0476826429367065,
      "step": 3094
    },
    {
      "epoch": 0.25345740052615956,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.1875642407281647e-05,
      "loss": 0.9073938131332397,
      "step": 3095
    },
    {
      "epoch": 0.2535392930626785,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1874220978077485e-05,
      "loss": 0.8722144365310669,
      "step": 3096
    },
    {
      "epoch": 0.25362118559919744,
      "grad_norm": 0.15625,
      "learning_rate": 2.187279911969901e-05,
      "loss": 0.6819531321525574,
      "step": 3097
    },
    {
      "epoch": 0.2537030781357164,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.1871376832208017e-05,
      "loss": 1.0163159370422363,
      "step": 3098
    },
    {
      "epoch": 0.2537849706722354,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.1869954115666337e-05,
      "loss": 1.041135549545288,
      "step": 3099
    },
    {
      "epoch": 0.2538668632087543,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.1868530970135797e-05,
      "loss": 0.7326840758323669,
      "step": 3100
    },
    {
      "epoch": 0.25394875574527326,
      "grad_norm": 0.19921875,
      "learning_rate": 2.1867107395678253e-05,
      "loss": 1.2159595489501953,
      "step": 3101
    },
    {
      "epoch": 0.2540306482817922,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1865683392355585e-05,
      "loss": 0.9358042478561401,
      "step": 3102
    },
    {
      "epoch": 0.2541125408183112,
      "grad_norm": 0.203125,
      "learning_rate": 2.1864258960229683e-05,
      "loss": 0.8298425674438477,
      "step": 3103
    },
    {
      "epoch": 0.2541944333548301,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.1862834099362457e-05,
      "loss": 0.7570986747741699,
      "step": 3104
    },
    {
      "epoch": 0.2542763258913491,
      "grad_norm": 0.1640625,
      "learning_rate": 2.186140880981584e-05,
      "loss": 1.0800713300704956,
      "step": 3105
    },
    {
      "epoch": 0.25435821842786804,
      "grad_norm": 0.18359375,
      "learning_rate": 2.1859983091651778e-05,
      "loss": 1.1672977209091187,
      "step": 3106
    },
    {
      "epoch": 0.25444011096438696,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.185855694493224e-05,
      "loss": 1.0439587831497192,
      "step": 3107
    },
    {
      "epoch": 0.2545220035009059,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.185713036971922e-05,
      "loss": 0.7559874653816223,
      "step": 3108
    },
    {
      "epoch": 0.2546038960374249,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.1855703366074707e-05,
      "loss": 0.9461041688919067,
      "step": 3109
    },
    {
      "epoch": 0.25468578857394386,
      "grad_norm": 0.142578125,
      "learning_rate": 2.1854275934060738e-05,
      "loss": 0.9024447202682495,
      "step": 3110
    },
    {
      "epoch": 0.2547676811104628,
      "grad_norm": 0.181640625,
      "learning_rate": 2.1852848073739348e-05,
      "loss": 0.5854197144508362,
      "step": 3111
    },
    {
      "epoch": 0.25484957364698174,
      "grad_norm": 0.2109375,
      "learning_rate": 2.1851419785172604e-05,
      "loss": 0.9766862392425537,
      "step": 3112
    },
    {
      "epoch": 0.2549314661835007,
      "grad_norm": 0.166015625,
      "learning_rate": 2.1849991068422577e-05,
      "loss": 0.9357848763465881,
      "step": 3113
    },
    {
      "epoch": 0.2550133587200197,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.184856192355137e-05,
      "loss": 0.65680330991745,
      "step": 3114
    },
    {
      "epoch": 0.2550952512565386,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.1847132350621096e-05,
      "loss": 0.8248845338821411,
      "step": 3115
    },
    {
      "epoch": 0.25517714379305756,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.1845702349693895e-05,
      "loss": 1.0158073902130127,
      "step": 3116
    },
    {
      "epoch": 0.2552590363295765,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.184427192083192e-05,
      "loss": 0.9877614378929138,
      "step": 3117
    },
    {
      "epoch": 0.25534092886609544,
      "grad_norm": 0.19140625,
      "learning_rate": 2.184284106409734e-05,
      "loss": 0.9232885241508484,
      "step": 3118
    },
    {
      "epoch": 0.2554228214026144,
      "grad_norm": 0.205078125,
      "learning_rate": 2.1841409779552345e-05,
      "loss": 0.760655403137207,
      "step": 3119
    },
    {
      "epoch": 0.2555047139391334,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.1839978067259153e-05,
      "loss": 0.9549834132194519,
      "step": 3120
    },
    {
      "epoch": 0.25558660647565234,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1838545927279982e-05,
      "loss": 0.7998074293136597,
      "step": 3121
    },
    {
      "epoch": 0.25566849901217126,
      "grad_norm": 0.154296875,
      "learning_rate": 2.1837113359677085e-05,
      "loss": 0.8158242702484131,
      "step": 3122
    },
    {
      "epoch": 0.2557503915486902,
      "grad_norm": 0.154296875,
      "learning_rate": 2.1835680364512725e-05,
      "loss": 1.0770370960235596,
      "step": 3123
    },
    {
      "epoch": 0.2558322840852092,
      "grad_norm": 0.212890625,
      "learning_rate": 2.1834246941849183e-05,
      "loss": 1.177646279335022,
      "step": 3124
    },
    {
      "epoch": 0.25591417662172816,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.183281309174877e-05,
      "loss": 0.8634687662124634,
      "step": 3125
    },
    {
      "epoch": 0.2559960691582471,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.1831378814273798e-05,
      "loss": 0.7518569231033325,
      "step": 3126
    },
    {
      "epoch": 0.25607796169476604,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.182994410948661e-05,
      "loss": 0.9456021785736084,
      "step": 3127
    },
    {
      "epoch": 0.256159854231285,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.1828508977449568e-05,
      "loss": 0.8693006038665771,
      "step": 3128
    },
    {
      "epoch": 0.2562417467678039,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.182707341822504e-05,
      "loss": 0.92015540599823,
      "step": 3129
    },
    {
      "epoch": 0.2563236393043229,
      "grad_norm": 0.1796875,
      "learning_rate": 2.1825637431875433e-05,
      "loss": 0.8281747102737427,
      "step": 3130
    },
    {
      "epoch": 0.25640553184084186,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.182420101846315e-05,
      "loss": 0.9466538429260254,
      "step": 3131
    },
    {
      "epoch": 0.25648742437736083,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.1822764178050627e-05,
      "loss": 0.8807326555252075,
      "step": 3132
    },
    {
      "epoch": 0.25656931691387974,
      "grad_norm": 0.212890625,
      "learning_rate": 2.182132691070032e-05,
      "loss": 0.7819794416427612,
      "step": 3133
    },
    {
      "epoch": 0.2566512094503987,
      "grad_norm": 0.16796875,
      "learning_rate": 2.181988921647469e-05,
      "loss": 0.7207423448562622,
      "step": 3134
    },
    {
      "epoch": 0.2567331019869177,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.181845109543623e-05,
      "loss": 0.744535505771637,
      "step": 3135
    },
    {
      "epoch": 0.25681499452343665,
      "grad_norm": 0.19140625,
      "learning_rate": 2.181701254764745e-05,
      "loss": 0.8818186521530151,
      "step": 3136
    },
    {
      "epoch": 0.25689688705995556,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.181557357317087e-05,
      "loss": 0.8681089282035828,
      "step": 3137
    },
    {
      "epoch": 0.2569787795964745,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.1814134172069032e-05,
      "loss": 0.8344646692276001,
      "step": 3138
    },
    {
      "epoch": 0.2570606721329935,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1812694344404503e-05,
      "loss": 0.609687328338623,
      "step": 3139
    },
    {
      "epoch": 0.2571425646695124,
      "grad_norm": 0.18359375,
      "learning_rate": 2.181125409023986e-05,
      "loss": 0.777496874332428,
      "step": 3140
    },
    {
      "epoch": 0.2572244572060314,
      "grad_norm": 0.177734375,
      "learning_rate": 2.180981340963771e-05,
      "loss": 0.958476185798645,
      "step": 3141
    },
    {
      "epoch": 0.25730634974255034,
      "grad_norm": 0.17578125,
      "learning_rate": 2.1808372302660664e-05,
      "loss": 0.8258377313613892,
      "step": 3142
    },
    {
      "epoch": 0.2573882422790693,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.1806930769371358e-05,
      "loss": 0.8648903965950012,
      "step": 3143
    },
    {
      "epoch": 0.2574701348155882,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.180548880983245e-05,
      "loss": 0.9093597531318665,
      "step": 3144
    },
    {
      "epoch": 0.2575520273521072,
      "grad_norm": 0.173828125,
      "learning_rate": 2.180404642410661e-05,
      "loss": 1.144288182258606,
      "step": 3145
    },
    {
      "epoch": 0.25763391988862616,
      "grad_norm": 0.1796875,
      "learning_rate": 2.1802603612256534e-05,
      "loss": 0.6446361541748047,
      "step": 3146
    },
    {
      "epoch": 0.25771581242514513,
      "grad_norm": 0.1796875,
      "learning_rate": 2.1801160374344935e-05,
      "loss": 0.8613916635513306,
      "step": 3147
    },
    {
      "epoch": 0.25779770496166404,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1799716710434534e-05,
      "loss": 1.0618281364440918,
      "step": 3148
    },
    {
      "epoch": 0.257879597498183,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.1798272620588085e-05,
      "loss": 0.9598224759101868,
      "step": 3149
    },
    {
      "epoch": 0.257961490034702,
      "grad_norm": 0.17578125,
      "learning_rate": 2.179682810486835e-05,
      "loss": 0.8121265172958374,
      "step": 3150
    },
    {
      "epoch": 0.2580433825712209,
      "grad_norm": 0.150390625,
      "learning_rate": 2.1795383163338115e-05,
      "loss": 0.9354348182678223,
      "step": 3151
    },
    {
      "epoch": 0.25812527510773986,
      "grad_norm": 0.203125,
      "learning_rate": 2.1793937796060184e-05,
      "loss": 0.9365136027336121,
      "step": 3152
    },
    {
      "epoch": 0.25820716764425883,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.179249200309738e-05,
      "loss": 0.5169033408164978,
      "step": 3153
    },
    {
      "epoch": 0.2582890601807778,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1791045784512542e-05,
      "loss": 0.9792241454124451,
      "step": 3154
    },
    {
      "epoch": 0.2583709527172967,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.1789599140368524e-05,
      "loss": 1.0530011653900146,
      "step": 3155
    },
    {
      "epoch": 0.2584528452538157,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.1788152070728208e-05,
      "loss": 0.5000385642051697,
      "step": 3156
    },
    {
      "epoch": 0.25853473779033465,
      "grad_norm": 0.197265625,
      "learning_rate": 2.1786704575654487e-05,
      "loss": 0.9103628396987915,
      "step": 3157
    },
    {
      "epoch": 0.2586166303268536,
      "grad_norm": 0.1640625,
      "learning_rate": 2.178525665521028e-05,
      "loss": 0.8115824460983276,
      "step": 3158
    },
    {
      "epoch": 0.2586985228633725,
      "grad_norm": 0.197265625,
      "learning_rate": 2.1783808309458515e-05,
      "loss": 0.8792688846588135,
      "step": 3159
    },
    {
      "epoch": 0.2587804153998915,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.1782359538462146e-05,
      "loss": 0.8507426381111145,
      "step": 3160
    },
    {
      "epoch": 0.25886230793641046,
      "grad_norm": 0.19140625,
      "learning_rate": 2.1780910342284138e-05,
      "loss": 0.9073144793510437,
      "step": 3161
    },
    {
      "epoch": 0.2589442004729294,
      "grad_norm": 0.18359375,
      "learning_rate": 2.1779460720987483e-05,
      "loss": 0.7104508876800537,
      "step": 3162
    },
    {
      "epoch": 0.25902609300944834,
      "grad_norm": 0.17578125,
      "learning_rate": 2.1778010674635188e-05,
      "loss": 0.9252887964248657,
      "step": 3163
    },
    {
      "epoch": 0.2591079855459673,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.1776560203290274e-05,
      "loss": 0.7737255096435547,
      "step": 3164
    },
    {
      "epoch": 0.2591898780824863,
      "grad_norm": 0.173828125,
      "learning_rate": 2.177510930701579e-05,
      "loss": 0.8987854719161987,
      "step": 3165
    },
    {
      "epoch": 0.2592717706190052,
      "grad_norm": 0.15625,
      "learning_rate": 2.177365798587479e-05,
      "loss": 0.9809098243713379,
      "step": 3166
    },
    {
      "epoch": 0.25935366315552416,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.177220623993036e-05,
      "loss": 0.9465204477310181,
      "step": 3167
    },
    {
      "epoch": 0.25943555569204313,
      "grad_norm": 0.1953125,
      "learning_rate": 2.1770754069245603e-05,
      "loss": 0.9769243001937866,
      "step": 3168
    },
    {
      "epoch": 0.25951744822856204,
      "grad_norm": 0.177734375,
      "learning_rate": 2.176930147388363e-05,
      "loss": 0.8874969482421875,
      "step": 3169
    },
    {
      "epoch": 0.259599340765081,
      "grad_norm": 0.1640625,
      "learning_rate": 2.1767848453907575e-05,
      "loss": 0.8695144653320312,
      "step": 3170
    },
    {
      "epoch": 0.2596812333016,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1766395009380597e-05,
      "loss": 0.676611602306366,
      "step": 3171
    },
    {
      "epoch": 0.25976312583811895,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.1764941140365866e-05,
      "loss": 0.8168504238128662,
      "step": 3172
    },
    {
      "epoch": 0.25984501837463786,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.1763486846926574e-05,
      "loss": 0.8248706459999084,
      "step": 3173
    },
    {
      "epoch": 0.2599269109111568,
      "grad_norm": 0.166015625,
      "learning_rate": 2.176203212912593e-05,
      "loss": 0.7082149982452393,
      "step": 3174
    },
    {
      "epoch": 0.2600088034476758,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.1760576987027165e-05,
      "loss": 0.6691633462905884,
      "step": 3175
    },
    {
      "epoch": 0.26009069598419476,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.1759121420693523e-05,
      "loss": 0.9209831357002258,
      "step": 3176
    },
    {
      "epoch": 0.2601725885207137,
      "grad_norm": 0.17578125,
      "learning_rate": 2.1757665430188272e-05,
      "loss": 0.7841283082962036,
      "step": 3177
    },
    {
      "epoch": 0.26025448105723265,
      "grad_norm": 0.2578125,
      "learning_rate": 2.1756209015574688e-05,
      "loss": 1.1342308521270752,
      "step": 3178
    },
    {
      "epoch": 0.2603363735937516,
      "grad_norm": 0.193359375,
      "learning_rate": 2.175475217691608e-05,
      "loss": 0.902176022529602,
      "step": 3179
    },
    {
      "epoch": 0.2604182661302705,
      "grad_norm": 0.158203125,
      "learning_rate": 2.1753294914275763e-05,
      "loss": 0.7692538499832153,
      "step": 3180
    },
    {
      "epoch": 0.2605001586667895,
      "grad_norm": 0.1796875,
      "learning_rate": 2.175183722771708e-05,
      "loss": 0.8452116250991821,
      "step": 3181
    },
    {
      "epoch": 0.26058205120330846,
      "grad_norm": 0.169921875,
      "learning_rate": 2.1750379117303387e-05,
      "loss": 0.968145489692688,
      "step": 3182
    },
    {
      "epoch": 0.26066394373982743,
      "grad_norm": 0.150390625,
      "learning_rate": 2.1748920583098056e-05,
      "loss": 0.7973054647445679,
      "step": 3183
    },
    {
      "epoch": 0.26074583627634634,
      "grad_norm": 0.16015625,
      "learning_rate": 2.1747461625164486e-05,
      "loss": 0.5918578505516052,
      "step": 3184
    },
    {
      "epoch": 0.2608277288128653,
      "grad_norm": 0.177734375,
      "learning_rate": 2.1746002243566086e-05,
      "loss": 1.0115458965301514,
      "step": 3185
    },
    {
      "epoch": 0.2609096213493843,
      "grad_norm": 0.203125,
      "learning_rate": 2.1744542438366286e-05,
      "loss": 1.1336578130722046,
      "step": 3186
    },
    {
      "epoch": 0.26099151388590325,
      "grad_norm": 0.1953125,
      "learning_rate": 2.174308220962854e-05,
      "loss": 1.0043970346450806,
      "step": 3187
    },
    {
      "epoch": 0.26107340642242216,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.174162155741631e-05,
      "loss": 0.6462467908859253,
      "step": 3188
    },
    {
      "epoch": 0.26115529895894113,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1740160481793085e-05,
      "loss": 0.8580551743507385,
      "step": 3189
    },
    {
      "epoch": 0.2612371914954601,
      "grad_norm": 0.20703125,
      "learning_rate": 2.173869898282237e-05,
      "loss": 1.1722352504730225,
      "step": 3190
    },
    {
      "epoch": 0.261319084031979,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.1737237060567688e-05,
      "loss": 0.743541419506073,
      "step": 3191
    },
    {
      "epoch": 0.261400976568498,
      "grad_norm": 0.173828125,
      "learning_rate": 2.173577471509257e-05,
      "loss": 0.8838053941726685,
      "step": 3192
    },
    {
      "epoch": 0.26148286910501695,
      "grad_norm": 0.17578125,
      "learning_rate": 2.173431194646059e-05,
      "loss": 1.0029171705245972,
      "step": 3193
    },
    {
      "epoch": 0.2615647616415359,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1732848754735322e-05,
      "loss": 1.2409414052963257,
      "step": 3194
    },
    {
      "epoch": 0.2616466541780548,
      "grad_norm": 0.169921875,
      "learning_rate": 2.173138513998036e-05,
      "loss": 0.846108078956604,
      "step": 3195
    },
    {
      "epoch": 0.2617285467145738,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.1729921102259317e-05,
      "loss": 1.0626165866851807,
      "step": 3196
    },
    {
      "epoch": 0.26181043925109276,
      "grad_norm": 0.1875,
      "learning_rate": 2.172845664163583e-05,
      "loss": 1.053910732269287,
      "step": 3197
    },
    {
      "epoch": 0.26189233178761173,
      "grad_norm": 0.189453125,
      "learning_rate": 2.172699175817355e-05,
      "loss": 0.8863646388053894,
      "step": 3198
    },
    {
      "epoch": 0.26197422432413064,
      "grad_norm": 0.171875,
      "learning_rate": 2.1725526451936142e-05,
      "loss": 1.0099010467529297,
      "step": 3199
    },
    {
      "epoch": 0.2620561168606496,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.17240607229873e-05,
      "loss": 0.9404041767120361,
      "step": 3200
    },
    {
      "epoch": 0.2621380093971686,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.1722594571390728e-05,
      "loss": 0.5972029566764832,
      "step": 3201
    },
    {
      "epoch": 0.2622199019336875,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.172112799721015e-05,
      "loss": 0.7316563129425049,
      "step": 3202
    },
    {
      "epoch": 0.26230179447020646,
      "grad_norm": 0.177734375,
      "learning_rate": 2.1719661000509316e-05,
      "loss": 1.034212589263916,
      "step": 3203
    },
    {
      "epoch": 0.26238368700672543,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.1718193581351975e-05,
      "loss": 0.9521102905273438,
      "step": 3204
    },
    {
      "epoch": 0.2624655795432444,
      "grad_norm": 0.169921875,
      "learning_rate": 2.171672573980192e-05,
      "loss": 0.871894359588623,
      "step": 3205
    },
    {
      "epoch": 0.2625474720797633,
      "grad_norm": 0.150390625,
      "learning_rate": 2.1715257475922942e-05,
      "loss": 0.6355702877044678,
      "step": 3206
    },
    {
      "epoch": 0.2626293646162823,
      "grad_norm": 0.158203125,
      "learning_rate": 2.1713788789778862e-05,
      "loss": 0.8873429894447327,
      "step": 3207
    },
    {
      "epoch": 0.26271125715280125,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.171231968143351e-05,
      "loss": 1.0372215509414673,
      "step": 3208
    },
    {
      "epoch": 0.2627931496893202,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1710850150950747e-05,
      "loss": 0.7492622137069702,
      "step": 3209
    },
    {
      "epoch": 0.26287504222583913,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.1709380198394437e-05,
      "loss": 0.9619871973991394,
      "step": 3210
    },
    {
      "epoch": 0.2629569347623581,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.1707909823828473e-05,
      "loss": 0.6420440673828125,
      "step": 3211
    },
    {
      "epoch": 0.26303882729887706,
      "grad_norm": 0.1640625,
      "learning_rate": 2.1706439027316765e-05,
      "loss": 0.8716416954994202,
      "step": 3212
    },
    {
      "epoch": 0.263120719835396,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.170496780892324e-05,
      "loss": 0.6702097058296204,
      "step": 3213
    },
    {
      "epoch": 0.26320261237191495,
      "grad_norm": 0.1640625,
      "learning_rate": 2.1703496168711842e-05,
      "loss": 1.0833449363708496,
      "step": 3214
    },
    {
      "epoch": 0.2632845049084339,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.170202410674653e-05,
      "loss": 0.8423060178756714,
      "step": 3215
    },
    {
      "epoch": 0.2633663974449529,
      "grad_norm": 0.14453125,
      "learning_rate": 2.1700551623091296e-05,
      "loss": 0.9879197478294373,
      "step": 3216
    },
    {
      "epoch": 0.2634482899814718,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1699078717810133e-05,
      "loss": 0.6664204597473145,
      "step": 3217
    },
    {
      "epoch": 0.26353018251799076,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.169760539096706e-05,
      "loss": 0.8965526819229126,
      "step": 3218
    },
    {
      "epoch": 0.26361207505450973,
      "grad_norm": 0.1279296875,
      "learning_rate": 2.169613164262612e-05,
      "loss": 0.7042847871780396,
      "step": 3219
    },
    {
      "epoch": 0.2636939675910287,
      "grad_norm": 0.16015625,
      "learning_rate": 2.1694657472851362e-05,
      "loss": 0.5074277520179749,
      "step": 3220
    },
    {
      "epoch": 0.2637758601275476,
      "grad_norm": 0.1796875,
      "learning_rate": 2.169318288170686e-05,
      "loss": 0.9434421062469482,
      "step": 3221
    },
    {
      "epoch": 0.2638577526640666,
      "grad_norm": 0.16796875,
      "learning_rate": 2.1691707869256703e-05,
      "loss": 1.2658205032348633,
      "step": 3222
    },
    {
      "epoch": 0.26393964520058555,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.169023243556501e-05,
      "loss": 1.0743223428726196,
      "step": 3223
    },
    {
      "epoch": 0.26402153773710446,
      "grad_norm": 0.158203125,
      "learning_rate": 2.16887565806959e-05,
      "loss": 0.8824954628944397,
      "step": 3224
    },
    {
      "epoch": 0.26410343027362343,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.168728030471353e-05,
      "loss": 0.6488144397735596,
      "step": 3225
    },
    {
      "epoch": 0.2641853228101424,
      "grad_norm": 0.1484375,
      "learning_rate": 2.1685803607682053e-05,
      "loss": 0.5821422934532166,
      "step": 3226
    },
    {
      "epoch": 0.26426721534666137,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.168432648966566e-05,
      "loss": 0.9351921081542969,
      "step": 3227
    },
    {
      "epoch": 0.2643491078831803,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.168284895072855e-05,
      "loss": 1.092302680015564,
      "step": 3228
    },
    {
      "epoch": 0.26443100041969925,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.1681370990934944e-05,
      "loss": 0.7790848612785339,
      "step": 3229
    },
    {
      "epoch": 0.2645128929562182,
      "grad_norm": 0.197265625,
      "learning_rate": 2.1679892610349086e-05,
      "loss": 0.8729329109191895,
      "step": 3230
    },
    {
      "epoch": 0.2645947854927372,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.1678413809035217e-05,
      "loss": 0.8801979422569275,
      "step": 3231
    },
    {
      "epoch": 0.2646766780292561,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.1676934587057626e-05,
      "loss": 0.9245819449424744,
      "step": 3232
    },
    {
      "epoch": 0.26475857056577506,
      "grad_norm": 0.19140625,
      "learning_rate": 2.1675454944480605e-05,
      "loss": 0.9424863457679749,
      "step": 3233
    },
    {
      "epoch": 0.26484046310229403,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.1673974881368456e-05,
      "loss": 0.9543589353561401,
      "step": 3234
    },
    {
      "epoch": 0.26492235563881295,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.1672494397785517e-05,
      "loss": 0.6458380818367004,
      "step": 3235
    },
    {
      "epoch": 0.2650042481753319,
      "grad_norm": 0.208984375,
      "learning_rate": 2.1671013493796134e-05,
      "loss": 0.9891178011894226,
      "step": 3236
    },
    {
      "epoch": 0.2650861407118509,
      "grad_norm": 0.1337890625,
      "learning_rate": 2.1669532169464673e-05,
      "loss": 0.9154999852180481,
      "step": 3237
    },
    {
      "epoch": 0.26516803324836985,
      "grad_norm": 0.17578125,
      "learning_rate": 2.1668050424855518e-05,
      "loss": 0.8810940384864807,
      "step": 3238
    },
    {
      "epoch": 0.26524992578488876,
      "grad_norm": 0.18359375,
      "learning_rate": 2.166656826003307e-05,
      "loss": 0.8166366219520569,
      "step": 3239
    },
    {
      "epoch": 0.26533181832140773,
      "grad_norm": 0.1640625,
      "learning_rate": 2.166508567506175e-05,
      "loss": 0.7319273352622986,
      "step": 3240
    },
    {
      "epoch": 0.2654137108579267,
      "grad_norm": 0.1875,
      "learning_rate": 2.1663602670006e-05,
      "loss": 0.8784331679344177,
      "step": 3241
    },
    {
      "epoch": 0.2654956033944456,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.1662119244930282e-05,
      "loss": 0.9117374420166016,
      "step": 3242
    },
    {
      "epoch": 0.2655774959309646,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.166063539989906e-05,
      "loss": 0.9602220058441162,
      "step": 3243
    },
    {
      "epoch": 0.26565938846748355,
      "grad_norm": 0.185546875,
      "learning_rate": 2.1659151134976838e-05,
      "loss": 0.9221316576004028,
      "step": 3244
    },
    {
      "epoch": 0.2657412810040025,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1657666450228123e-05,
      "loss": 1.1194803714752197,
      "step": 3245
    },
    {
      "epoch": 0.26582317354052143,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.165618134571745e-05,
      "loss": 1.4498815536499023,
      "step": 3246
    },
    {
      "epoch": 0.2659050660770404,
      "grad_norm": 0.1298828125,
      "learning_rate": 2.1654695821509362e-05,
      "loss": 0.7133536338806152,
      "step": 3247
    },
    {
      "epoch": 0.26598695861355937,
      "grad_norm": 0.1875,
      "learning_rate": 2.1653209877668427e-05,
      "loss": 0.7237063050270081,
      "step": 3248
    },
    {
      "epoch": 0.26606885115007833,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.1651723514259233e-05,
      "loss": 0.9974046349525452,
      "step": 3249
    },
    {
      "epoch": 0.26615074368659725,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1650236731346383e-05,
      "loss": 0.8866168260574341,
      "step": 3250
    },
    {
      "epoch": 0.2662326362231162,
      "grad_norm": 0.169921875,
      "learning_rate": 2.16487495289945e-05,
      "loss": 0.8706800937652588,
      "step": 3251
    },
    {
      "epoch": 0.2663145287596352,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.1647261907268223e-05,
      "loss": 1.0852530002593994,
      "step": 3252
    },
    {
      "epoch": 0.2663964212961541,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1645773866232204e-05,
      "loss": 0.8357301950454712,
      "step": 3253
    },
    {
      "epoch": 0.26647831383267306,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.164428540595113e-05,
      "loss": 0.9736103415489197,
      "step": 3254
    },
    {
      "epoch": 0.26656020636919203,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.1642796526489688e-05,
      "loss": 0.9015145897865295,
      "step": 3255
    },
    {
      "epoch": 0.266642098905711,
      "grad_norm": 0.125,
      "learning_rate": 2.1641307227912593e-05,
      "loss": 0.7823876142501831,
      "step": 3256
    },
    {
      "epoch": 0.2667239914422299,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.1639817510284576e-05,
      "loss": 1.0863451957702637,
      "step": 3257
    },
    {
      "epoch": 0.2668058839787489,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.1638327373670387e-05,
      "loss": 0.5649192929267883,
      "step": 3258
    },
    {
      "epoch": 0.26688777651526785,
      "grad_norm": 0.224609375,
      "learning_rate": 2.1636836818134793e-05,
      "loss": 0.8872652053833008,
      "step": 3259
    },
    {
      "epoch": 0.2669696690517868,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.1635345843742583e-05,
      "loss": 0.5855347514152527,
      "step": 3260
    },
    {
      "epoch": 0.26705156158830573,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.1633854450558553e-05,
      "loss": 0.8995658159255981,
      "step": 3261
    },
    {
      "epoch": 0.2671334541248247,
      "grad_norm": 0.185546875,
      "learning_rate": 2.163236263864753e-05,
      "loss": 0.9249913096427917,
      "step": 3262
    },
    {
      "epoch": 0.26721534666134367,
      "grad_norm": 0.181640625,
      "learning_rate": 2.1630870408074356e-05,
      "loss": 0.9172536730766296,
      "step": 3263
    },
    {
      "epoch": 0.2672972391978626,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1629377758903886e-05,
      "loss": 1.160362720489502,
      "step": 3264
    },
    {
      "epoch": 0.26737913173438155,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1627884691201e-05,
      "loss": 1.0079379081726074,
      "step": 3265
    },
    {
      "epoch": 0.2674610242709005,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.162639120503059e-05,
      "loss": 0.8354660272598267,
      "step": 3266
    },
    {
      "epoch": 0.2675429168074195,
      "grad_norm": 0.1728515625,
      "learning_rate": 2.162489730045757e-05,
      "loss": 0.83490389585495,
      "step": 3267
    },
    {
      "epoch": 0.2676248093439384,
      "grad_norm": 0.177734375,
      "learning_rate": 2.1623402977546866e-05,
      "loss": 0.8092930316925049,
      "step": 3268
    },
    {
      "epoch": 0.26770670188045737,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.162190823636344e-05,
      "loss": 0.947308361530304,
      "step": 3269
    },
    {
      "epoch": 0.26778859441697633,
      "grad_norm": 0.212890625,
      "learning_rate": 2.162041307697225e-05,
      "loss": 0.8496508598327637,
      "step": 3270
    },
    {
      "epoch": 0.2678704869534953,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.1618917499438284e-05,
      "loss": 0.9197134971618652,
      "step": 3271
    },
    {
      "epoch": 0.2679523794900142,
      "grad_norm": 0.1484375,
      "learning_rate": 2.1617421503826545e-05,
      "loss": 0.6617064476013184,
      "step": 3272
    },
    {
      "epoch": 0.2680342720265332,
      "grad_norm": 0.18359375,
      "learning_rate": 2.161592509020206e-05,
      "loss": 0.714332640171051,
      "step": 3273
    },
    {
      "epoch": 0.26811616456305215,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.161442825862986e-05,
      "loss": 0.6404594779014587,
      "step": 3274
    },
    {
      "epoch": 0.26819805709957106,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.1612931009175013e-05,
      "loss": 0.7566066384315491,
      "step": 3275
    },
    {
      "epoch": 0.26827994963609003,
      "grad_norm": 0.158203125,
      "learning_rate": 2.161143334190259e-05,
      "loss": 0.8044634461402893,
      "step": 3276
    },
    {
      "epoch": 0.268361842172609,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.1609935256877687e-05,
      "loss": 1.4273775815963745,
      "step": 3277
    },
    {
      "epoch": 0.26844373470912797,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.1608436754165417e-05,
      "loss": 0.7233834266662598,
      "step": 3278
    },
    {
      "epoch": 0.2685256272456469,
      "grad_norm": 0.166015625,
      "learning_rate": 2.1606937833830915e-05,
      "loss": 0.8778788447380066,
      "step": 3279
    },
    {
      "epoch": 0.26860751978216585,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.1605438495939324e-05,
      "loss": 1.1701123714447021,
      "step": 3280
    },
    {
      "epoch": 0.2686894123186848,
      "grad_norm": 0.16796875,
      "learning_rate": 2.1603938740555812e-05,
      "loss": 0.935649573802948,
      "step": 3281
    },
    {
      "epoch": 0.2687713048552038,
      "grad_norm": 0.2353515625,
      "learning_rate": 2.160243856774557e-05,
      "loss": 0.7498301863670349,
      "step": 3282
    },
    {
      "epoch": 0.2688531973917227,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1600937977573796e-05,
      "loss": 0.7776157855987549,
      "step": 3283
    },
    {
      "epoch": 0.26893508992824167,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.1599436970105717e-05,
      "loss": 0.9413073062896729,
      "step": 3284
    },
    {
      "epoch": 0.26901698246476063,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.159793554540657e-05,
      "loss": 1.0632649660110474,
      "step": 3285
    },
    {
      "epoch": 0.26909887500127955,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.1596433703541608e-05,
      "loss": 0.7193772792816162,
      "step": 3286
    },
    {
      "epoch": 0.2691807675377985,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1594931444576115e-05,
      "loss": 1.1338950395584106,
      "step": 3287
    },
    {
      "epoch": 0.2692626600743175,
      "grad_norm": 0.18359375,
      "learning_rate": 2.1593428768575388e-05,
      "loss": 0.944406270980835,
      "step": 3288
    },
    {
      "epoch": 0.26934455261083645,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.1591925675604725e-05,
      "loss": 0.9453411102294922,
      "step": 3289
    },
    {
      "epoch": 0.26942644514735536,
      "grad_norm": 0.19921875,
      "learning_rate": 2.1590422165729473e-05,
      "loss": 0.635551929473877,
      "step": 3290
    },
    {
      "epoch": 0.26950833768387433,
      "grad_norm": 0.2265625,
      "learning_rate": 2.1588918239014973e-05,
      "loss": 0.8830382227897644,
      "step": 3291
    },
    {
      "epoch": 0.2695902302203933,
      "grad_norm": 0.158203125,
      "learning_rate": 2.158741389552659e-05,
      "loss": 0.906486451625824,
      "step": 3292
    },
    {
      "epoch": 0.26967212275691227,
      "grad_norm": 0.15625,
      "learning_rate": 2.1585909135329713e-05,
      "loss": 0.9953230023384094,
      "step": 3293
    },
    {
      "epoch": 0.2697540152934312,
      "grad_norm": 0.2265625,
      "learning_rate": 2.158440395848974e-05,
      "loss": 0.9503775238990784,
      "step": 3294
    },
    {
      "epoch": 0.26983590782995015,
      "grad_norm": 0.146484375,
      "learning_rate": 2.1582898365072102e-05,
      "loss": 0.5092971324920654,
      "step": 3295
    },
    {
      "epoch": 0.2699178003664691,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1581392355142227e-05,
      "loss": 0.9471458196640015,
      "step": 3296
    },
    {
      "epoch": 0.26999969290298803,
      "grad_norm": 0.1572265625,
      "learning_rate": 2.157988592876558e-05,
      "loss": 0.4898902177810669,
      "step": 3297
    },
    {
      "epoch": 0.270081585439507,
      "grad_norm": 0.16015625,
      "learning_rate": 2.1578379086007633e-05,
      "loss": 0.6574862599372864,
      "step": 3298
    },
    {
      "epoch": 0.27016347797602597,
      "grad_norm": 0.1787109375,
      "learning_rate": 2.157687182693388e-05,
      "loss": 0.9062574505805969,
      "step": 3299
    },
    {
      "epoch": 0.27024537051254494,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.1575364151609833e-05,
      "loss": 1.3011393547058105,
      "step": 3300
    },
    {
      "epoch": 0.27032726304906385,
      "grad_norm": 0.130859375,
      "learning_rate": 2.1573856060101023e-05,
      "loss": 0.8509037494659424,
      "step": 3301
    },
    {
      "epoch": 0.2704091555855828,
      "grad_norm": 0.16015625,
      "learning_rate": 2.1572347552472993e-05,
      "loss": 0.9384050369262695,
      "step": 3302
    },
    {
      "epoch": 0.2704910481221018,
      "grad_norm": 0.166015625,
      "learning_rate": 2.1570838628791318e-05,
      "loss": 0.9923960566520691,
      "step": 3303
    },
    {
      "epoch": 0.2705729406586207,
      "grad_norm": 0.19140625,
      "learning_rate": 2.1569329289121577e-05,
      "loss": 0.8370259404182434,
      "step": 3304
    },
    {
      "epoch": 0.27065483319513967,
      "grad_norm": 0.16796875,
      "learning_rate": 2.156781953352937e-05,
      "loss": 0.6945908069610596,
      "step": 3305
    },
    {
      "epoch": 0.27073672573165863,
      "grad_norm": 0.22265625,
      "learning_rate": 2.156630936208032e-05,
      "loss": 0.7941571474075317,
      "step": 3306
    },
    {
      "epoch": 0.2708186182681776,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.156479877484006e-05,
      "loss": 1.0198163986206055,
      "step": 3307
    },
    {
      "epoch": 0.2709005108046965,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.1563287771874255e-05,
      "loss": 0.6542173027992249,
      "step": 3308
    },
    {
      "epoch": 0.2709824033412155,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.1561776353248574e-05,
      "loss": 0.743459165096283,
      "step": 3309
    },
    {
      "epoch": 0.27106429587773445,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.156026451902871e-05,
      "loss": 0.837165892124176,
      "step": 3310
    },
    {
      "epoch": 0.2711461884142534,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.155875226928038e-05,
      "loss": 0.7349922060966492,
      "step": 3311
    },
    {
      "epoch": 0.27122808095077233,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.15572396040693e-05,
      "loss": 0.9216521382331848,
      "step": 3312
    },
    {
      "epoch": 0.2713099734872913,
      "grad_norm": 0.1640625,
      "learning_rate": 2.1555726523461224e-05,
      "loss": 0.7761582732200623,
      "step": 3313
    },
    {
      "epoch": 0.27139186602381027,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.1554213027521916e-05,
      "loss": 0.8121665120124817,
      "step": 3314
    },
    {
      "epoch": 0.2714737585603292,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.155269911631716e-05,
      "loss": 0.8760371804237366,
      "step": 3315
    },
    {
      "epoch": 0.27155565109684815,
      "grad_norm": 0.162109375,
      "learning_rate": 2.1551184789912756e-05,
      "loss": 0.5040208101272583,
      "step": 3316
    },
    {
      "epoch": 0.2716375436333671,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.1549670048374522e-05,
      "loss": 0.5382332801818848,
      "step": 3317
    },
    {
      "epoch": 0.2717194361698861,
      "grad_norm": 0.18359375,
      "learning_rate": 2.1548154891768297e-05,
      "loss": 0.9676521420478821,
      "step": 3318
    },
    {
      "epoch": 0.271801328706405,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.1546639320159933e-05,
      "loss": 0.6809881329536438,
      "step": 3319
    },
    {
      "epoch": 0.27188322124292397,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.1545123333615305e-05,
      "loss": 0.9668429493904114,
      "step": 3320
    },
    {
      "epoch": 0.27196511377944294,
      "grad_norm": 0.17578125,
      "learning_rate": 2.1543606932200302e-05,
      "loss": 0.703383207321167,
      "step": 3321
    },
    {
      "epoch": 0.2720470063159619,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.154209011598084e-05,
      "loss": 0.8019699454307556,
      "step": 3322
    },
    {
      "epoch": 0.2721288988524808,
      "grad_norm": 0.146484375,
      "learning_rate": 2.1540572885022834e-05,
      "loss": 0.679069995880127,
      "step": 3323
    },
    {
      "epoch": 0.2722107913889998,
      "grad_norm": 0.158203125,
      "learning_rate": 2.1539055239392238e-05,
      "loss": 0.9363023042678833,
      "step": 3324
    },
    {
      "epoch": 0.27229268392551875,
      "grad_norm": 0.1640625,
      "learning_rate": 2.1537537179155017e-05,
      "loss": 0.6437115669250488,
      "step": 3325
    },
    {
      "epoch": 0.27237457646203767,
      "grad_norm": 0.1533203125,
      "learning_rate": 2.153601870437714e-05,
      "loss": 0.7268803715705872,
      "step": 3326
    },
    {
      "epoch": 0.27245646899855663,
      "grad_norm": 0.162109375,
      "learning_rate": 2.153449981512462e-05,
      "loss": 0.8665785193443298,
      "step": 3327
    },
    {
      "epoch": 0.2725383615350756,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.153298051146346e-05,
      "loss": 0.865010678768158,
      "step": 3328
    },
    {
      "epoch": 0.27262025407159457,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.1531460793459715e-05,
      "loss": 0.7722593545913696,
      "step": 3329
    },
    {
      "epoch": 0.2727021466081135,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.1529940661179422e-05,
      "loss": 0.8160887956619263,
      "step": 3330
    },
    {
      "epoch": 0.27278403914463245,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.1528420114688654e-05,
      "loss": 0.9909271001815796,
      "step": 3331
    },
    {
      "epoch": 0.2728659316811514,
      "grad_norm": 0.1376953125,
      "learning_rate": 2.1526899154053507e-05,
      "loss": 0.8764643669128418,
      "step": 3332
    },
    {
      "epoch": 0.2729478242176704,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1525377779340086e-05,
      "loss": 0.9621984958648682,
      "step": 3333
    },
    {
      "epoch": 0.2730297167541893,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1523855990614508e-05,
      "loss": 0.8603249192237854,
      "step": 3334
    },
    {
      "epoch": 0.27311160929070827,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.152233378794293e-05,
      "loss": 0.7865197658538818,
      "step": 3335
    },
    {
      "epoch": 0.27319350182722724,
      "grad_norm": 0.1796875,
      "learning_rate": 2.15208111713915e-05,
      "loss": 0.8992390036582947,
      "step": 3336
    },
    {
      "epoch": 0.27327539436374615,
      "grad_norm": 0.166015625,
      "learning_rate": 2.1519288141026405e-05,
      "loss": 0.921657919883728,
      "step": 3337
    },
    {
      "epoch": 0.2733572869002651,
      "grad_norm": 0.16796875,
      "learning_rate": 2.151776469691384e-05,
      "loss": 0.5226486325263977,
      "step": 3338
    },
    {
      "epoch": 0.2734391794367841,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.151624083912002e-05,
      "loss": 0.7886949777603149,
      "step": 3339
    },
    {
      "epoch": 0.27352107197330305,
      "grad_norm": 0.15625,
      "learning_rate": 2.151471656771118e-05,
      "loss": 0.9264416098594666,
      "step": 3340
    },
    {
      "epoch": 0.27360296450982197,
      "grad_norm": 0.1796875,
      "learning_rate": 2.151319188275357e-05,
      "loss": 0.6940490007400513,
      "step": 3341
    },
    {
      "epoch": 0.27368485704634093,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.151166678431346e-05,
      "loss": 0.7757836580276489,
      "step": 3342
    },
    {
      "epoch": 0.2737667495828599,
      "grad_norm": 0.1796875,
      "learning_rate": 2.1510141272457134e-05,
      "loss": 0.8390427827835083,
      "step": 3343
    },
    {
      "epoch": 0.27384864211937887,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.1508615347250898e-05,
      "loss": 0.9607489109039307,
      "step": 3344
    },
    {
      "epoch": 0.2739305346558978,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.1507089008761078e-05,
      "loss": 0.783923864364624,
      "step": 3345
    },
    {
      "epoch": 0.27401242719241675,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.150556225705401e-05,
      "loss": 0.7999295592308044,
      "step": 3346
    },
    {
      "epoch": 0.2740943197289357,
      "grad_norm": 0.158203125,
      "learning_rate": 2.150403509219606e-05,
      "loss": 0.8167508244514465,
      "step": 3347
    },
    {
      "epoch": 0.27417621226545463,
      "grad_norm": 0.1875,
      "learning_rate": 2.1502507514253596e-05,
      "loss": 0.8116077184677124,
      "step": 3348
    },
    {
      "epoch": 0.2742581048019736,
      "grad_norm": 0.18359375,
      "learning_rate": 2.150097952329302e-05,
      "loss": 1.3164403438568115,
      "step": 3349
    },
    {
      "epoch": 0.27433999733849257,
      "grad_norm": 0.18359375,
      "learning_rate": 2.1499451119380745e-05,
      "loss": 0.670732855796814,
      "step": 3350
    },
    {
      "epoch": 0.27442188987501154,
      "grad_norm": 0.17578125,
      "learning_rate": 2.149792230258319e-05,
      "loss": 1.1000217199325562,
      "step": 3351
    },
    {
      "epoch": 0.27450378241153045,
      "grad_norm": 0.18359375,
      "learning_rate": 2.1496393072966824e-05,
      "loss": 0.8944618701934814,
      "step": 3352
    },
    {
      "epoch": 0.2745856749480494,
      "grad_norm": 0.17578125,
      "learning_rate": 2.1494863430598095e-05,
      "loss": 0.9926925301551819,
      "step": 3353
    },
    {
      "epoch": 0.2746675674845684,
      "grad_norm": 0.1640625,
      "learning_rate": 2.1493333375543495e-05,
      "loss": 0.9102782607078552,
      "step": 3354
    },
    {
      "epoch": 0.27474946002108735,
      "grad_norm": 0.154296875,
      "learning_rate": 2.149180290786953e-05,
      "loss": 1.0286027193069458,
      "step": 3355
    },
    {
      "epoch": 0.27483135255760627,
      "grad_norm": 0.15625,
      "learning_rate": 2.1490272027642717e-05,
      "loss": 0.6212396621704102,
      "step": 3356
    },
    {
      "epoch": 0.27491324509412524,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.1488740734929593e-05,
      "loss": 0.8314988613128662,
      "step": 3357
    },
    {
      "epoch": 0.2749951376306442,
      "grad_norm": 0.1708984375,
      "learning_rate": 2.148720902979671e-05,
      "loss": 0.8606696128845215,
      "step": 3358
    },
    {
      "epoch": 0.2750770301671631,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.1485676912310654e-05,
      "loss": 0.8349741697311401,
      "step": 3359
    },
    {
      "epoch": 0.2751589227036821,
      "grad_norm": 0.15234375,
      "learning_rate": 2.148414438253801e-05,
      "loss": 0.7126782536506653,
      "step": 3360
    },
    {
      "epoch": 0.27524081524020105,
      "grad_norm": 0.150390625,
      "learning_rate": 2.1482611440545387e-05,
      "loss": 0.8179392218589783,
      "step": 3361
    },
    {
      "epoch": 0.27532270777672,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.1481078086399414e-05,
      "loss": 0.892266571521759,
      "step": 3362
    },
    {
      "epoch": 0.27540460031323893,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1479544320166738e-05,
      "loss": 1.2144250869750977,
      "step": 3363
    },
    {
      "epoch": 0.2754864928497579,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.1478010141914025e-05,
      "loss": 0.7363628149032593,
      "step": 3364
    },
    {
      "epoch": 0.27556838538627687,
      "grad_norm": 0.1640625,
      "learning_rate": 2.147647555170795e-05,
      "loss": 0.9002977013587952,
      "step": 3365
    },
    {
      "epoch": 0.27565027792279584,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.1474940549615223e-05,
      "loss": 0.7352514863014221,
      "step": 3366
    },
    {
      "epoch": 0.27573217045931475,
      "grad_norm": 0.201171875,
      "learning_rate": 2.1473405135702553e-05,
      "loss": 0.6047722101211548,
      "step": 3367
    },
    {
      "epoch": 0.2758140629958337,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.1471869310036673e-05,
      "loss": 0.7658945322036743,
      "step": 3368
    },
    {
      "epoch": 0.2758959555323527,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.1470333072684345e-05,
      "loss": 0.7119348049163818,
      "step": 3369
    },
    {
      "epoch": 0.2759778480688716,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.1468796423712337e-05,
      "loss": 1.169087529182434,
      "step": 3370
    },
    {
      "epoch": 0.27605974060539057,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.1467259363187434e-05,
      "loss": 0.9668115973472595,
      "step": 3371
    },
    {
      "epoch": 0.27614163314190954,
      "grad_norm": 0.162109375,
      "learning_rate": 2.146572189117645e-05,
      "loss": 1.1241261959075928,
      "step": 3372
    },
    {
      "epoch": 0.2762235256784285,
      "grad_norm": 0.208984375,
      "learning_rate": 2.1464184007746204e-05,
      "loss": 1.1234874725341797,
      "step": 3373
    },
    {
      "epoch": 0.2763054182149474,
      "grad_norm": 0.19140625,
      "learning_rate": 2.1462645712963542e-05,
      "loss": 1.0917906761169434,
      "step": 3374
    },
    {
      "epoch": 0.2763873107514664,
      "grad_norm": 0.166015625,
      "learning_rate": 2.1461107006895322e-05,
      "loss": 0.7701463103294373,
      "step": 3375
    },
    {
      "epoch": 0.27646920328798535,
      "grad_norm": 0.197265625,
      "learning_rate": 2.1459567889608425e-05,
      "loss": 0.8444384336471558,
      "step": 3376
    },
    {
      "epoch": 0.27655109582450427,
      "grad_norm": 0.21484375,
      "learning_rate": 2.1458028361169746e-05,
      "loss": 1.0009405612945557,
      "step": 3377
    },
    {
      "epoch": 0.27663298836102324,
      "grad_norm": 0.162109375,
      "learning_rate": 2.1456488421646202e-05,
      "loss": 0.7138739824295044,
      "step": 3378
    },
    {
      "epoch": 0.2767148808975422,
      "grad_norm": 0.173828125,
      "learning_rate": 2.145494807110472e-05,
      "loss": 0.7219703793525696,
      "step": 3379
    },
    {
      "epoch": 0.27679677343406117,
      "grad_norm": 0.171875,
      "learning_rate": 2.1453407309612253e-05,
      "loss": 0.7128299474716187,
      "step": 3380
    },
    {
      "epoch": 0.2768786659705801,
      "grad_norm": 0.1640625,
      "learning_rate": 2.145186613723577e-05,
      "loss": 0.6667338609695435,
      "step": 3381
    },
    {
      "epoch": 0.27696055850709905,
      "grad_norm": 0.154296875,
      "learning_rate": 2.1450324554042254e-05,
      "loss": 0.8319555521011353,
      "step": 3382
    },
    {
      "epoch": 0.277042451043618,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.144878256009871e-05,
      "loss": 0.8431136608123779,
      "step": 3383
    },
    {
      "epoch": 0.277124343580137,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.144724015547216e-05,
      "loss": 0.7516739368438721,
      "step": 3384
    },
    {
      "epoch": 0.2772062361166559,
      "grad_norm": 0.1328125,
      "learning_rate": 2.144569734022964e-05,
      "loss": 0.5456576347351074,
      "step": 3385
    },
    {
      "epoch": 0.27728812865317487,
      "grad_norm": 0.16796875,
      "learning_rate": 2.1444154114438215e-05,
      "loss": 0.8720216155052185,
      "step": 3386
    },
    {
      "epoch": 0.27737002118969384,
      "grad_norm": 0.255859375,
      "learning_rate": 2.144261047816495e-05,
      "loss": 1.2837430238723755,
      "step": 3387
    },
    {
      "epoch": 0.27745191372621275,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.1441066431476944e-05,
      "loss": 0.7659080624580383,
      "step": 3388
    },
    {
      "epoch": 0.2775338062627317,
      "grad_norm": 0.181640625,
      "learning_rate": 2.1439521974441306e-05,
      "loss": 1.1645841598510742,
      "step": 3389
    },
    {
      "epoch": 0.2776156987992507,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.1437977107125166e-05,
      "loss": 0.9932535886764526,
      "step": 3390
    },
    {
      "epoch": 0.27769759133576966,
      "grad_norm": 0.166015625,
      "learning_rate": 2.1436431829595665e-05,
      "loss": 1.0030277967453003,
      "step": 3391
    },
    {
      "epoch": 0.27777948387228857,
      "grad_norm": 0.166015625,
      "learning_rate": 2.1434886141919973e-05,
      "loss": 0.6630159020423889,
      "step": 3392
    },
    {
      "epoch": 0.27786137640880754,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.143334004416527e-05,
      "loss": 0.6942229270935059,
      "step": 3393
    },
    {
      "epoch": 0.2779432689453265,
      "grad_norm": 0.1826171875,
      "learning_rate": 2.1431793536398753e-05,
      "loss": 0.7576448917388916,
      "step": 3394
    },
    {
      "epoch": 0.2780251614818455,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1430246618687646e-05,
      "loss": 0.9480331540107727,
      "step": 3395
    },
    {
      "epoch": 0.2781070540183644,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.1428699291099177e-05,
      "loss": 0.7499651908874512,
      "step": 3396
    },
    {
      "epoch": 0.27818894655488335,
      "grad_norm": 0.171875,
      "learning_rate": 2.1427151553700606e-05,
      "loss": 0.9190750122070312,
      "step": 3397
    },
    {
      "epoch": 0.2782708390914023,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.1425603406559198e-05,
      "loss": 0.8124038577079773,
      "step": 3398
    },
    {
      "epoch": 0.27835273162792123,
      "grad_norm": 0.173828125,
      "learning_rate": 2.1424054849742248e-05,
      "loss": 1.122057318687439,
      "step": 3399
    },
    {
      "epoch": 0.2784346241644402,
      "grad_norm": 0.171875,
      "learning_rate": 2.1422505883317058e-05,
      "loss": 0.845874547958374,
      "step": 3400
    },
    {
      "epoch": 0.27851651670095917,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.1420956507350953e-05,
      "loss": 0.9810129404067993,
      "step": 3401
    },
    {
      "epoch": 0.27859840923747814,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.1419406721911277e-05,
      "loss": 0.7269851565361023,
      "step": 3402
    },
    {
      "epoch": 0.27868030177399705,
      "grad_norm": 0.1630859375,
      "learning_rate": 2.141785652706539e-05,
      "loss": 0.7321099042892456,
      "step": 3403
    },
    {
      "epoch": 0.278762194310516,
      "grad_norm": 0.158203125,
      "learning_rate": 2.141630592288067e-05,
      "loss": 0.5046508312225342,
      "step": 3404
    },
    {
      "epoch": 0.278844086847035,
      "grad_norm": 0.166015625,
      "learning_rate": 2.1414754909424508e-05,
      "loss": 0.6796903610229492,
      "step": 3405
    },
    {
      "epoch": 0.27892597938355396,
      "grad_norm": 0.1484375,
      "learning_rate": 2.1413203486764324e-05,
      "loss": 0.7862737774848938,
      "step": 3406
    },
    {
      "epoch": 0.27900787192007287,
      "grad_norm": 0.154296875,
      "learning_rate": 2.141165165496754e-05,
      "loss": 0.8339933156967163,
      "step": 3407
    },
    {
      "epoch": 0.27908976445659184,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.141009941410162e-05,
      "loss": 0.9509305953979492,
      "step": 3408
    },
    {
      "epoch": 0.2791716569931108,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.140854676423402e-05,
      "loss": 1.0637438297271729,
      "step": 3409
    },
    {
      "epoch": 0.2792535495296297,
      "grad_norm": 0.19140625,
      "learning_rate": 2.1406993705432225e-05,
      "loss": 1.1327433586120605,
      "step": 3410
    },
    {
      "epoch": 0.2793354420661487,
      "grad_norm": 0.181640625,
      "learning_rate": 2.1405440237763737e-05,
      "loss": 0.9173071384429932,
      "step": 3411
    },
    {
      "epoch": 0.27941733460266766,
      "grad_norm": 0.197265625,
      "learning_rate": 2.140388636129608e-05,
      "loss": 1.0329043865203857,
      "step": 3412
    },
    {
      "epoch": 0.2794992271391866,
      "grad_norm": 0.208984375,
      "learning_rate": 2.1402332076096794e-05,
      "loss": 1.0449938774108887,
      "step": 3413
    },
    {
      "epoch": 0.27958111967570554,
      "grad_norm": 0.150390625,
      "learning_rate": 2.1400777382233423e-05,
      "loss": 0.7545625567436218,
      "step": 3414
    },
    {
      "epoch": 0.2796630122122245,
      "grad_norm": 0.150390625,
      "learning_rate": 2.1399222279773553e-05,
      "loss": 0.8826322555541992,
      "step": 3415
    },
    {
      "epoch": 0.2797449047487435,
      "grad_norm": 0.1796875,
      "learning_rate": 2.1397666768784766e-05,
      "loss": 1.063607096672058,
      "step": 3416
    },
    {
      "epoch": 0.27982679728526244,
      "grad_norm": 0.16796875,
      "learning_rate": 2.1396110849334682e-05,
      "loss": 0.9113302826881409,
      "step": 3417
    },
    {
      "epoch": 0.27990868982178135,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.1394554521490916e-05,
      "loss": 0.9686901569366455,
      "step": 3418
    },
    {
      "epoch": 0.2799905823583003,
      "grad_norm": 0.15234375,
      "learning_rate": 2.1392997785321117e-05,
      "loss": 0.7442489266395569,
      "step": 3419
    },
    {
      "epoch": 0.2800724748948193,
      "grad_norm": 0.1640625,
      "learning_rate": 2.139144064089295e-05,
      "loss": 0.9052310585975647,
      "step": 3420
    },
    {
      "epoch": 0.2801543674313382,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.138988308827409e-05,
      "loss": 0.8613073229789734,
      "step": 3421
    },
    {
      "epoch": 0.28023625996785717,
      "grad_norm": 0.1806640625,
      "learning_rate": 2.1388325127532237e-05,
      "loss": 0.8392568826675415,
      "step": 3422
    },
    {
      "epoch": 0.28031815250437614,
      "grad_norm": 0.5545027256011963,
      "learning_rate": 2.138676675873511e-05,
      "loss": 1.0198066234588623,
      "step": 3423
    },
    {
      "epoch": 0.2804000450408951,
      "grad_norm": 0.4595646262168884,
      "learning_rate": 2.138520798195043e-05,
      "loss": 0.7996975779533386,
      "step": 3424
    },
    {
      "epoch": 0.280481937577414,
      "grad_norm": 0.30966249108314514,
      "learning_rate": 2.1383648797245964e-05,
      "loss": 1.255074381828308,
      "step": 3425
    },
    {
      "epoch": 0.280563830113933,
      "grad_norm": 0.3617246747016907,
      "learning_rate": 2.1382089204689472e-05,
      "loss": 0.9435317516326904,
      "step": 3426
    },
    {
      "epoch": 0.28064572265045196,
      "grad_norm": 0.4147423207759857,
      "learning_rate": 2.1380529204348737e-05,
      "loss": 1.2138372659683228,
      "step": 3427
    },
    {
      "epoch": 0.2807276151869709,
      "grad_norm": 0.3086487054824829,
      "learning_rate": 2.1378968796291573e-05,
      "loss": 1.2042549848556519,
      "step": 3428
    },
    {
      "epoch": 0.28080950772348984,
      "grad_norm": 0.2678821384906769,
      "learning_rate": 2.1377407980585793e-05,
      "loss": 1.2063226699829102,
      "step": 3429
    },
    {
      "epoch": 0.2808914002600088,
      "grad_norm": 0.2660142183303833,
      "learning_rate": 2.1375846757299238e-05,
      "loss": 1.0376510620117188,
      "step": 3430
    },
    {
      "epoch": 0.2809732927965278,
      "grad_norm": 0.3243228793144226,
      "learning_rate": 2.1374285126499768e-05,
      "loss": 1.0781488418579102,
      "step": 3431
    },
    {
      "epoch": 0.2810551853330467,
      "grad_norm": 0.2732582688331604,
      "learning_rate": 2.1372723088255255e-05,
      "loss": 1.0324527025222778,
      "step": 3432
    },
    {
      "epoch": 0.28113707786956565,
      "grad_norm": 0.2861688435077667,
      "learning_rate": 2.1371160642633597e-05,
      "loss": 0.7667874693870544,
      "step": 3433
    },
    {
      "epoch": 0.2812189704060846,
      "grad_norm": 0.292312890291214,
      "learning_rate": 2.1369597789702696e-05,
      "loss": 1.2269484996795654,
      "step": 3434
    },
    {
      "epoch": 0.2813008629426036,
      "grad_norm": 0.30376964807510376,
      "learning_rate": 2.1368034529530486e-05,
      "loss": 0.9459224939346313,
      "step": 3435
    },
    {
      "epoch": 0.2813827554791225,
      "grad_norm": 0.2555980682373047,
      "learning_rate": 2.136647086218491e-05,
      "loss": 1.2257999181747437,
      "step": 3436
    },
    {
      "epoch": 0.28146464801564147,
      "grad_norm": 0.3167591392993927,
      "learning_rate": 2.136490678773394e-05,
      "loss": 0.9251630306243896,
      "step": 3437
    },
    {
      "epoch": 0.28154654055216044,
      "grad_norm": 0.2513663172721863,
      "learning_rate": 2.1363342306245542e-05,
      "loss": 0.9764858484268188,
      "step": 3438
    },
    {
      "epoch": 0.28162843308867935,
      "grad_norm": 0.2653743624687195,
      "learning_rate": 2.1361777417787722e-05,
      "loss": 0.9525572061538696,
      "step": 3439
    },
    {
      "epoch": 0.2817103256251983,
      "grad_norm": 0.25401419401168823,
      "learning_rate": 2.1360212122428504e-05,
      "loss": 0.958200216293335,
      "step": 3440
    },
    {
      "epoch": 0.2817922181617173,
      "grad_norm": 0.34562209248542786,
      "learning_rate": 2.135864642023591e-05,
      "loss": 0.9576067924499512,
      "step": 3441
    },
    {
      "epoch": 0.28187411069823626,
      "grad_norm": 0.24991260468959808,
      "learning_rate": 2.1357080311278e-05,
      "loss": 0.7946743369102478,
      "step": 3442
    },
    {
      "epoch": 0.28195600323475517,
      "grad_norm": 0.31108707189559937,
      "learning_rate": 2.135551379562284e-05,
      "loss": 0.7807180881500244,
      "step": 3443
    },
    {
      "epoch": 0.28203789577127414,
      "grad_norm": 0.3037821352481842,
      "learning_rate": 2.1353946873338516e-05,
      "loss": 0.9542021751403809,
      "step": 3444
    },
    {
      "epoch": 0.2821197883077931,
      "grad_norm": 0.26082366704940796,
      "learning_rate": 2.1352379544493136e-05,
      "loss": 0.8835973143577576,
      "step": 3445
    },
    {
      "epoch": 0.2822016808443121,
      "grad_norm": 0.24040617048740387,
      "learning_rate": 2.135081180915482e-05,
      "loss": 0.9280560612678528,
      "step": 3446
    },
    {
      "epoch": 0.282283573380831,
      "grad_norm": 0.24738606810569763,
      "learning_rate": 2.134924366739171e-05,
      "loss": 0.9297424554824829,
      "step": 3447
    },
    {
      "epoch": 0.28236546591734996,
      "grad_norm": 0.2396710366010666,
      "learning_rate": 2.1347675119271963e-05,
      "loss": 0.9982947707176208,
      "step": 3448
    },
    {
      "epoch": 0.2824473584538689,
      "grad_norm": 0.22884425520896912,
      "learning_rate": 2.1346106164863757e-05,
      "loss": 0.8980881571769714,
      "step": 3449
    },
    {
      "epoch": 0.28252925099038784,
      "grad_norm": 0.24518036842346191,
      "learning_rate": 2.1344536804235283e-05,
      "loss": 0.8322001099586487,
      "step": 3450
    },
    {
      "epoch": 0.2826111435269068,
      "grad_norm": 0.20058652758598328,
      "learning_rate": 2.1342967037454752e-05,
      "loss": 0.9090523719787598,
      "step": 3451
    },
    {
      "epoch": 0.2826930360634258,
      "grad_norm": 0.20916317403316498,
      "learning_rate": 2.1341396864590393e-05,
      "loss": 1.0050197839736938,
      "step": 3452
    },
    {
      "epoch": 0.28277492859994474,
      "grad_norm": 0.1953408569097519,
      "learning_rate": 2.133982628571045e-05,
      "loss": 0.9561449289321899,
      "step": 3453
    },
    {
      "epoch": 0.28285682113646365,
      "grad_norm": 0.21356765925884247,
      "learning_rate": 2.133825530088319e-05,
      "loss": 1.0452392101287842,
      "step": 3454
    },
    {
      "epoch": 0.2829387136729826,
      "grad_norm": 0.24609620869159698,
      "learning_rate": 2.1336683910176894e-05,
      "loss": 1.0572271347045898,
      "step": 3455
    },
    {
      "epoch": 0.2830206062095016,
      "grad_norm": 0.22513464093208313,
      "learning_rate": 2.133511211365986e-05,
      "loss": 0.8402677178382874,
      "step": 3456
    },
    {
      "epoch": 0.28310249874602056,
      "grad_norm": 0.24858179688453674,
      "learning_rate": 2.1333539911400406e-05,
      "loss": 1.086310625076294,
      "step": 3457
    },
    {
      "epoch": 0.28318439128253947,
      "grad_norm": 0.22680412232875824,
      "learning_rate": 2.133196730346686e-05,
      "loss": 0.6389338970184326,
      "step": 3458
    },
    {
      "epoch": 0.28326628381905844,
      "grad_norm": 0.1937006264925003,
      "learning_rate": 2.1330394289927586e-05,
      "loss": 1.0341110229492188,
      "step": 3459
    },
    {
      "epoch": 0.2833481763555774,
      "grad_norm": 0.21183858811855316,
      "learning_rate": 2.1328820870850944e-05,
      "loss": 0.8572084307670593,
      "step": 3460
    },
    {
      "epoch": 0.2834300688920963,
      "grad_norm": 0.23052723705768585,
      "learning_rate": 2.1327247046305326e-05,
      "loss": 1.122206449508667,
      "step": 3461
    },
    {
      "epoch": 0.2835119614286153,
      "grad_norm": 0.2200106531381607,
      "learning_rate": 2.1325672816359135e-05,
      "loss": 0.8906905651092529,
      "step": 3462
    },
    {
      "epoch": 0.28359385396513426,
      "grad_norm": 0.24819497764110565,
      "learning_rate": 2.1324098181080793e-05,
      "loss": 1.2684714794158936,
      "step": 3463
    },
    {
      "epoch": 0.2836757465016532,
      "grad_norm": 0.17595423758029938,
      "learning_rate": 2.132252314053874e-05,
      "loss": 0.580949068069458,
      "step": 3464
    },
    {
      "epoch": 0.28375763903817214,
      "grad_norm": 0.20759359002113342,
      "learning_rate": 2.1320947694801438e-05,
      "loss": 1.0530781745910645,
      "step": 3465
    },
    {
      "epoch": 0.2838395315746911,
      "grad_norm": 0.164481982588768,
      "learning_rate": 2.1319371843937355e-05,
      "loss": 0.520302951335907,
      "step": 3466
    },
    {
      "epoch": 0.2839214241112101,
      "grad_norm": 0.21060939133167267,
      "learning_rate": 2.131779558801499e-05,
      "loss": 0.9476941227912903,
      "step": 3467
    },
    {
      "epoch": 0.28400331664772904,
      "grad_norm": 0.2568098306655884,
      "learning_rate": 2.131621892710285e-05,
      "loss": 0.8709134459495544,
      "step": 3468
    },
    {
      "epoch": 0.28408520918424796,
      "grad_norm": 0.20867551863193512,
      "learning_rate": 2.1314641861269463e-05,
      "loss": 0.9996675252914429,
      "step": 3469
    },
    {
      "epoch": 0.2841671017207669,
      "grad_norm": 0.19812415540218353,
      "learning_rate": 2.131306439058338e-05,
      "loss": 0.851952075958252,
      "step": 3470
    },
    {
      "epoch": 0.2842489942572859,
      "grad_norm": 0.19411563873291016,
      "learning_rate": 2.1311486515113162e-05,
      "loss": 0.6276647448539734,
      "step": 3471
    },
    {
      "epoch": 0.2843308867938048,
      "grad_norm": 0.24754750728607178,
      "learning_rate": 2.130990823492739e-05,
      "loss": 0.8451815247535706,
      "step": 3472
    },
    {
      "epoch": 0.2844127793303238,
      "grad_norm": 0.20604436099529266,
      "learning_rate": 2.1308329550094658e-05,
      "loss": 0.9173741340637207,
      "step": 3473
    },
    {
      "epoch": 0.28449467186684274,
      "grad_norm": 0.21023263037204742,
      "learning_rate": 2.1306750460683585e-05,
      "loss": 1.026425838470459,
      "step": 3474
    },
    {
      "epoch": 0.2845765644033617,
      "grad_norm": 0.19120372831821442,
      "learning_rate": 2.1305170966762805e-05,
      "loss": 0.8910130262374878,
      "step": 3475
    },
    {
      "epoch": 0.2846584569398806,
      "grad_norm": 0.18901297450065613,
      "learning_rate": 2.130359106840097e-05,
      "loss": 0.6302460432052612,
      "step": 3476
    },
    {
      "epoch": 0.2847403494763996,
      "grad_norm": 0.21456506848335266,
      "learning_rate": 2.130201076566675e-05,
      "loss": 1.1058099269866943,
      "step": 3477
    },
    {
      "epoch": 0.28482224201291856,
      "grad_norm": 0.18502090871334076,
      "learning_rate": 2.130043005862883e-05,
      "loss": 0.6995141506195068,
      "step": 3478
    },
    {
      "epoch": 0.2849041345494375,
      "grad_norm": 0.2526298761367798,
      "learning_rate": 2.1298848947355914e-05,
      "loss": 1.0816057920455933,
      "step": 3479
    },
    {
      "epoch": 0.28498602708595644,
      "grad_norm": 0.2092422991991043,
      "learning_rate": 2.1297267431916723e-05,
      "loss": 0.7715293169021606,
      "step": 3480
    },
    {
      "epoch": 0.2850679196224754,
      "grad_norm": 0.22245262563228607,
      "learning_rate": 2.1295685512379997e-05,
      "loss": 1.4129287004470825,
      "step": 3481
    },
    {
      "epoch": 0.2851498121589944,
      "grad_norm": 0.252566933631897,
      "learning_rate": 2.129410318881449e-05,
      "loss": 0.9743900299072266,
      "step": 3482
    },
    {
      "epoch": 0.2852317046955133,
      "grad_norm": 0.16718724370002747,
      "learning_rate": 2.1292520461288984e-05,
      "loss": 0.48105719685554504,
      "step": 3483
    },
    {
      "epoch": 0.28531359723203226,
      "grad_norm": 0.21920624375343323,
      "learning_rate": 2.1290937329872263e-05,
      "loss": 0.7706964015960693,
      "step": 3484
    },
    {
      "epoch": 0.2853954897685512,
      "grad_norm": 0.20677447319030762,
      "learning_rate": 2.128935379463314e-05,
      "loss": 1.0058112144470215,
      "step": 3485
    },
    {
      "epoch": 0.2854773823050702,
      "grad_norm": 0.20283395051956177,
      "learning_rate": 2.1287769855640437e-05,
      "loss": 0.8632333874702454,
      "step": 3486
    },
    {
      "epoch": 0.2855592748415891,
      "grad_norm": 0.20556992292404175,
      "learning_rate": 2.1286185512963006e-05,
      "loss": 0.729941725730896,
      "step": 3487
    },
    {
      "epoch": 0.2856411673781081,
      "grad_norm": 0.20022818446159363,
      "learning_rate": 2.1284600766669703e-05,
      "loss": 1.026996374130249,
      "step": 3488
    },
    {
      "epoch": 0.28572305991462704,
      "grad_norm": 0.1891862154006958,
      "learning_rate": 2.1283015616829413e-05,
      "loss": 0.8198191523551941,
      "step": 3489
    },
    {
      "epoch": 0.285804952451146,
      "grad_norm": 0.26199573278427124,
      "learning_rate": 2.128143006351103e-05,
      "loss": 0.8752697110176086,
      "step": 3490
    },
    {
      "epoch": 0.2858868449876649,
      "grad_norm": 0.25328126549720764,
      "learning_rate": 2.1279844106783468e-05,
      "loss": 0.7696465253829956,
      "step": 3491
    },
    {
      "epoch": 0.2859687375241839,
      "grad_norm": 0.1954895406961441,
      "learning_rate": 2.127825774671566e-05,
      "loss": 0.9152534008026123,
      "step": 3492
    },
    {
      "epoch": 0.28605063006070286,
      "grad_norm": 0.16714297235012054,
      "learning_rate": 2.1276670983376558e-05,
      "loss": 0.6931283473968506,
      "step": 3493
    },
    {
      "epoch": 0.2861325225972218,
      "grad_norm": 0.21672800183296204,
      "learning_rate": 2.1275083816835127e-05,
      "loss": 1.0445926189422607,
      "step": 3494
    },
    {
      "epoch": 0.28621441513374074,
      "grad_norm": 0.19939132034778595,
      "learning_rate": 2.127349624716035e-05,
      "loss": 0.9226009249687195,
      "step": 3495
    },
    {
      "epoch": 0.2862963076702597,
      "grad_norm": 0.17384614050388336,
      "learning_rate": 2.1271908274421233e-05,
      "loss": 0.6694759726524353,
      "step": 3496
    },
    {
      "epoch": 0.2863782002067787,
      "grad_norm": 0.21277102828025818,
      "learning_rate": 2.1270319898686792e-05,
      "loss": 0.7632806301116943,
      "step": 3497
    },
    {
      "epoch": 0.2864600927432976,
      "grad_norm": 0.18788889050483704,
      "learning_rate": 2.126873112002607e-05,
      "loss": 0.8672062158584595,
      "step": 3498
    },
    {
      "epoch": 0.28654198527981656,
      "grad_norm": 0.21760745346546173,
      "learning_rate": 2.1267141938508115e-05,
      "loss": 0.6594892740249634,
      "step": 3499
    },
    {
      "epoch": 0.2866238778163355,
      "grad_norm": 0.215936541557312,
      "learning_rate": 2.126555235420201e-05,
      "loss": 0.6994695067405701,
      "step": 3500
    },
    {
      "epoch": 0.2867057703528545,
      "grad_norm": 0.1899453103542328,
      "learning_rate": 2.126396236717683e-05,
      "loss": 1.0494346618652344,
      "step": 3501
    },
    {
      "epoch": 0.2867876628893734,
      "grad_norm": 0.19823096692562103,
      "learning_rate": 2.1262371977501692e-05,
      "loss": 0.7733740210533142,
      "step": 3502
    },
    {
      "epoch": 0.2868695554258924,
      "grad_norm": 0.22605086863040924,
      "learning_rate": 2.1260781185245722e-05,
      "loss": 0.7243752479553223,
      "step": 3503
    },
    {
      "epoch": 0.28695144796241134,
      "grad_norm": 0.16761215031147003,
      "learning_rate": 2.1259189990478053e-05,
      "loss": 0.7263807654380798,
      "step": 3504
    },
    {
      "epoch": 0.28703334049893026,
      "grad_norm": 0.21117354929447174,
      "learning_rate": 2.1257598393267856e-05,
      "loss": 0.916650652885437,
      "step": 3505
    },
    {
      "epoch": 0.2871152330354492,
      "grad_norm": 0.17035061120986938,
      "learning_rate": 2.12560063936843e-05,
      "loss": 0.5386192202568054,
      "step": 3506
    },
    {
      "epoch": 0.2871971255719682,
      "grad_norm": 0.20572909712791443,
      "learning_rate": 2.1254413991796587e-05,
      "loss": 0.774568498134613,
      "step": 3507
    },
    {
      "epoch": 0.28727901810848716,
      "grad_norm": 0.19865141808986664,
      "learning_rate": 2.1252821187673924e-05,
      "loss": 0.8318107724189758,
      "step": 3508
    },
    {
      "epoch": 0.2873609106450061,
      "grad_norm": 0.19003914296627045,
      "learning_rate": 2.1251227981385544e-05,
      "loss": 0.9233892560005188,
      "step": 3509
    },
    {
      "epoch": 0.28744280318152504,
      "grad_norm": 0.2344079613685608,
      "learning_rate": 2.124963437300069e-05,
      "loss": 0.9341983795166016,
      "step": 3510
    },
    {
      "epoch": 0.287524695718044,
      "grad_norm": 0.15803226828575134,
      "learning_rate": 2.1248040362588635e-05,
      "loss": 0.7519998550415039,
      "step": 3511
    },
    {
      "epoch": 0.2876065882545629,
      "grad_norm": 0.21238510310649872,
      "learning_rate": 2.1246445950218654e-05,
      "loss": 0.7486221790313721,
      "step": 3512
    },
    {
      "epoch": 0.2876884807910819,
      "grad_norm": 0.16170275211334229,
      "learning_rate": 2.1244851135960047e-05,
      "loss": 0.7832815647125244,
      "step": 3513
    },
    {
      "epoch": 0.28777037332760086,
      "grad_norm": 0.2384539246559143,
      "learning_rate": 2.1243255919882132e-05,
      "loss": 0.7580384016036987,
      "step": 3514
    },
    {
      "epoch": 0.2878522658641198,
      "grad_norm": 0.22442620992660522,
      "learning_rate": 2.1241660302054247e-05,
      "loss": 0.6989566087722778,
      "step": 3515
    },
    {
      "epoch": 0.28793415840063874,
      "grad_norm": 0.1991148591041565,
      "learning_rate": 2.124006428254574e-05,
      "loss": 0.9883272051811218,
      "step": 3516
    },
    {
      "epoch": 0.2880160509371577,
      "grad_norm": 0.20200927555561066,
      "learning_rate": 2.1238467861425985e-05,
      "loss": 0.7939639687538147,
      "step": 3517
    },
    {
      "epoch": 0.2880979434736767,
      "grad_norm": 0.22574682533740997,
      "learning_rate": 2.1236871038764362e-05,
      "loss": 0.850476861000061,
      "step": 3518
    },
    {
      "epoch": 0.28817983601019564,
      "grad_norm": 0.17249317467212677,
      "learning_rate": 2.123527381463028e-05,
      "loss": 0.5548695921897888,
      "step": 3519
    },
    {
      "epoch": 0.28826172854671456,
      "grad_norm": 0.1980728805065155,
      "learning_rate": 2.1233676189093164e-05,
      "loss": 0.8167330026626587,
      "step": 3520
    },
    {
      "epoch": 0.2883436210832335,
      "grad_norm": 0.1657007336616516,
      "learning_rate": 2.123207816222245e-05,
      "loss": 0.7155598402023315,
      "step": 3521
    },
    {
      "epoch": 0.2884255136197525,
      "grad_norm": 0.18101030588150024,
      "learning_rate": 2.123047973408759e-05,
      "loss": 0.6549408435821533,
      "step": 3522
    },
    {
      "epoch": 0.2885074061562714,
      "grad_norm": 0.20963455736637115,
      "learning_rate": 2.1228880904758068e-05,
      "loss": 0.7759702801704407,
      "step": 3523
    },
    {
      "epoch": 0.2885892986927904,
      "grad_norm": 0.39317113161087036,
      "learning_rate": 2.122728167430337e-05,
      "loss": 0.9718130826950073,
      "step": 3524
    },
    {
      "epoch": 0.28867119122930934,
      "grad_norm": 0.1974186897277832,
      "learning_rate": 2.1225682042793006e-05,
      "loss": 0.9817551374435425,
      "step": 3525
    },
    {
      "epoch": 0.2887530837658283,
      "grad_norm": 0.20485323667526245,
      "learning_rate": 2.1224082010296502e-05,
      "loss": 0.8142287135124207,
      "step": 3526
    },
    {
      "epoch": 0.2888349763023472,
      "grad_norm": 0.1927838921546936,
      "learning_rate": 2.1222481576883407e-05,
      "loss": 0.6998031139373779,
      "step": 3527
    },
    {
      "epoch": 0.2889168688388662,
      "grad_norm": 0.1966256946325302,
      "learning_rate": 2.1220880742623272e-05,
      "loss": 1.10746431350708,
      "step": 3528
    },
    {
      "epoch": 0.28899876137538516,
      "grad_norm": 0.2113838940858841,
      "learning_rate": 2.1219279507585684e-05,
      "loss": 0.7397045493125916,
      "step": 3529
    },
    {
      "epoch": 0.28908065391190413,
      "grad_norm": 0.20919625461101532,
      "learning_rate": 2.1217677871840237e-05,
      "loss": 1.1148093938827515,
      "step": 3530
    },
    {
      "epoch": 0.28916254644842304,
      "grad_norm": 0.19676482677459717,
      "learning_rate": 2.1216075835456545e-05,
      "loss": 0.7693652510643005,
      "step": 3531
    },
    {
      "epoch": 0.289244438984942,
      "grad_norm": 0.17656555771827698,
      "learning_rate": 2.121447339850424e-05,
      "loss": 0.5824464559555054,
      "step": 3532
    },
    {
      "epoch": 0.289326331521461,
      "grad_norm": 0.18711617588996887,
      "learning_rate": 2.1212870561052967e-05,
      "loss": 0.7225980758666992,
      "step": 3533
    },
    {
      "epoch": 0.2894082240579799,
      "grad_norm": 0.20248383283615112,
      "learning_rate": 2.1211267323172395e-05,
      "loss": 0.7111272215843201,
      "step": 3534
    },
    {
      "epoch": 0.28949011659449886,
      "grad_norm": 0.22039610147476196,
      "learning_rate": 2.120966368493221e-05,
      "loss": 0.7027662992477417,
      "step": 3535
    },
    {
      "epoch": 0.2895720091310178,
      "grad_norm": 0.22662951052188873,
      "learning_rate": 2.1208059646402103e-05,
      "loss": 0.7621035575866699,
      "step": 3536
    },
    {
      "epoch": 0.2896539016675368,
      "grad_norm": 0.17215535044670105,
      "learning_rate": 2.1206455207651805e-05,
      "loss": 0.866097629070282,
      "step": 3537
    },
    {
      "epoch": 0.2897357942040557,
      "grad_norm": 0.19153009355068207,
      "learning_rate": 2.120485036875104e-05,
      "loss": 0.8311400413513184,
      "step": 3538
    },
    {
      "epoch": 0.2898176867405747,
      "grad_norm": 0.2478572428226471,
      "learning_rate": 2.120324512976957e-05,
      "loss": 0.7503623366355896,
      "step": 3539
    },
    {
      "epoch": 0.28989957927709364,
      "grad_norm": 0.20078802108764648,
      "learning_rate": 2.120163949077716e-05,
      "loss": 0.5034589171409607,
      "step": 3540
    },
    {
      "epoch": 0.2899814718136126,
      "grad_norm": 0.22539538145065308,
      "learning_rate": 2.12000334518436e-05,
      "loss": 0.8971729874610901,
      "step": 3541
    },
    {
      "epoch": 0.2900633643501315,
      "grad_norm": 0.22430646419525146,
      "learning_rate": 2.119842701303869e-05,
      "loss": 1.1851565837860107,
      "step": 3542
    },
    {
      "epoch": 0.2901452568866505,
      "grad_norm": 0.20909461379051208,
      "learning_rate": 2.1196820174432264e-05,
      "loss": 0.5989973545074463,
      "step": 3543
    },
    {
      "epoch": 0.29022714942316946,
      "grad_norm": 0.1593896448612213,
      "learning_rate": 2.1195212936094148e-05,
      "loss": 0.6213828921318054,
      "step": 3544
    },
    {
      "epoch": 0.2903090419596884,
      "grad_norm": 0.18795660138130188,
      "learning_rate": 2.1193605298094208e-05,
      "loss": 0.515428900718689,
      "step": 3545
    },
    {
      "epoch": 0.29039093449620734,
      "grad_norm": 0.1951572149991989,
      "learning_rate": 2.1191997260502315e-05,
      "loss": 0.7568216919898987,
      "step": 3546
    },
    {
      "epoch": 0.2904728270327263,
      "grad_norm": 0.15674081444740295,
      "learning_rate": 2.119038882338836e-05,
      "loss": 0.6737725734710693,
      "step": 3547
    },
    {
      "epoch": 0.2905547195692453,
      "grad_norm": 0.18807759881019592,
      "learning_rate": 2.1188779986822256e-05,
      "loss": 0.9392871260643005,
      "step": 3548
    },
    {
      "epoch": 0.2906366121057642,
      "grad_norm": 0.2081177532672882,
      "learning_rate": 2.1187170750873932e-05,
      "loss": 0.9934238195419312,
      "step": 3549
    },
    {
      "epoch": 0.29071850464228316,
      "grad_norm": 0.21088671684265137,
      "learning_rate": 2.1185561115613322e-05,
      "loss": 0.9218080639839172,
      "step": 3550
    },
    {
      "epoch": 0.29080039717880213,
      "grad_norm": 0.208150252699852,
      "learning_rate": 2.1183951081110396e-05,
      "loss": 1.020330548286438,
      "step": 3551
    },
    {
      "epoch": 0.2908822897153211,
      "grad_norm": 0.19615116715431213,
      "learning_rate": 2.118234064743513e-05,
      "loss": 0.8098911046981812,
      "step": 3552
    },
    {
      "epoch": 0.29096418225184,
      "grad_norm": 0.16913655400276184,
      "learning_rate": 2.118072981465752e-05,
      "loss": 0.701073408126831,
      "step": 3553
    },
    {
      "epoch": 0.291046074788359,
      "grad_norm": 0.20514576137065887,
      "learning_rate": 2.117911858284758e-05,
      "loss": 0.9694563150405884,
      "step": 3554
    },
    {
      "epoch": 0.29112796732487795,
      "grad_norm": 0.1729450523853302,
      "learning_rate": 2.117750695207534e-05,
      "loss": 0.8312361836433411,
      "step": 3555
    },
    {
      "epoch": 0.29120985986139686,
      "grad_norm": 0.18179979920387268,
      "learning_rate": 2.117589492241085e-05,
      "loss": 0.7960910797119141,
      "step": 3556
    },
    {
      "epoch": 0.2912917523979158,
      "grad_norm": 0.2027096152305603,
      "learning_rate": 2.1174282493924172e-05,
      "loss": 0.5393216013908386,
      "step": 3557
    },
    {
      "epoch": 0.2913736449344348,
      "grad_norm": 0.45329728722572327,
      "learning_rate": 2.1172669666685392e-05,
      "loss": 0.8141299486160278,
      "step": 3558
    },
    {
      "epoch": 0.29145553747095376,
      "grad_norm": 0.1998831331729889,
      "learning_rate": 2.1171056440764606e-05,
      "loss": 1.109323263168335,
      "step": 3559
    },
    {
      "epoch": 0.2915374300074727,
      "grad_norm": 0.18283510208129883,
      "learning_rate": 2.116944281623194e-05,
      "loss": 0.7146959900856018,
      "step": 3560
    },
    {
      "epoch": 0.29161932254399164,
      "grad_norm": 0.17302998900413513,
      "learning_rate": 2.116782879315752e-05,
      "loss": 0.7244794368743896,
      "step": 3561
    },
    {
      "epoch": 0.2917012150805106,
      "grad_norm": 0.2004489153623581,
      "learning_rate": 2.1166214371611496e-05,
      "loss": 0.9209729433059692,
      "step": 3562
    },
    {
      "epoch": 0.2917831076170296,
      "grad_norm": 0.18157488107681274,
      "learning_rate": 2.116459955166405e-05,
      "loss": 0.8496153950691223,
      "step": 3563
    },
    {
      "epoch": 0.2918650001535485,
      "grad_norm": 0.1911879926919937,
      "learning_rate": 2.1162984333385358e-05,
      "loss": 1.1315703392028809,
      "step": 3564
    },
    {
      "epoch": 0.29194689269006746,
      "grad_norm": 0.1688978374004364,
      "learning_rate": 2.1161368716845626e-05,
      "loss": 0.8458208441734314,
      "step": 3565
    },
    {
      "epoch": 0.29202878522658643,
      "grad_norm": 0.20819826424121857,
      "learning_rate": 2.1159752702115075e-05,
      "loss": 0.7576515078544617,
      "step": 3566
    },
    {
      "epoch": 0.29211067776310534,
      "grad_norm": 0.23401206731796265,
      "learning_rate": 2.1158136289263952e-05,
      "loss": 1.089598298072815,
      "step": 3567
    },
    {
      "epoch": 0.2921925702996243,
      "grad_norm": 0.2130027562379837,
      "learning_rate": 2.1156519478362506e-05,
      "loss": 1.0247031450271606,
      "step": 3568
    },
    {
      "epoch": 0.2922744628361433,
      "grad_norm": 0.25619781017303467,
      "learning_rate": 2.1154902269481003e-05,
      "loss": 0.7694712281227112,
      "step": 3569
    },
    {
      "epoch": 0.29235635537266225,
      "grad_norm": 0.1922881156206131,
      "learning_rate": 2.1153284662689747e-05,
      "loss": 0.9210580587387085,
      "step": 3570
    },
    {
      "epoch": 0.29243824790918116,
      "grad_norm": 0.1944931447505951,
      "learning_rate": 2.115166665805904e-05,
      "loss": 0.9005422592163086,
      "step": 3571
    },
    {
      "epoch": 0.2925201404457001,
      "grad_norm": 0.23080092668533325,
      "learning_rate": 2.1150048255659206e-05,
      "loss": 0.7594001293182373,
      "step": 3572
    },
    {
      "epoch": 0.2926020329822191,
      "grad_norm": 0.19769082963466644,
      "learning_rate": 2.1148429455560586e-05,
      "loss": 0.9875636696815491,
      "step": 3573
    },
    {
      "epoch": 0.292683925518738,
      "grad_norm": 0.2345762848854065,
      "learning_rate": 2.1146810257833546e-05,
      "loss": 0.856637179851532,
      "step": 3574
    },
    {
      "epoch": 0.292765818055257,
      "grad_norm": 0.19932471215724945,
      "learning_rate": 2.1145190662548454e-05,
      "loss": 0.8109740614891052,
      "step": 3575
    },
    {
      "epoch": 0.29284771059177594,
      "grad_norm": 0.2150142788887024,
      "learning_rate": 2.1143570669775718e-05,
      "loss": 0.8369321823120117,
      "step": 3576
    },
    {
      "epoch": 0.2929296031282949,
      "grad_norm": 0.17657695710659027,
      "learning_rate": 2.1141950279585733e-05,
      "loss": 0.723195493221283,
      "step": 3577
    },
    {
      "epoch": 0.2930114956648138,
      "grad_norm": 0.1696932315826416,
      "learning_rate": 2.114032949204894e-05,
      "loss": 1.216187596321106,
      "step": 3578
    },
    {
      "epoch": 0.2930933882013328,
      "grad_norm": 0.199996218085289,
      "learning_rate": 2.113870830723578e-05,
      "loss": 0.9340181350708008,
      "step": 3579
    },
    {
      "epoch": 0.29317528073785176,
      "grad_norm": 0.15180617570877075,
      "learning_rate": 2.1137086725216717e-05,
      "loss": 0.624434769153595,
      "step": 3580
    },
    {
      "epoch": 0.29325717327437073,
      "grad_norm": 0.23232433199882507,
      "learning_rate": 2.113546474606223e-05,
      "loss": 1.0785369873046875,
      "step": 3581
    },
    {
      "epoch": 0.29333906581088964,
      "grad_norm": 0.20001237094402313,
      "learning_rate": 2.1133842369842827e-05,
      "loss": 0.8305016160011292,
      "step": 3582
    },
    {
      "epoch": 0.2934209583474086,
      "grad_norm": 0.19694243371486664,
      "learning_rate": 2.1132219596629004e-05,
      "loss": 0.7204241752624512,
      "step": 3583
    },
    {
      "epoch": 0.2935028508839276,
      "grad_norm": 0.18327634036540985,
      "learning_rate": 2.1130596426491314e-05,
      "loss": 0.6896508932113647,
      "step": 3584
    },
    {
      "epoch": 0.2935847434204465,
      "grad_norm": 0.19185766577720642,
      "learning_rate": 2.1128972859500294e-05,
      "loss": 0.9345558285713196,
      "step": 3585
    },
    {
      "epoch": 0.29366663595696546,
      "grad_norm": 0.19742132723331451,
      "learning_rate": 2.1127348895726515e-05,
      "loss": 0.6110600829124451,
      "step": 3586
    },
    {
      "epoch": 0.29374852849348443,
      "grad_norm": 0.1884407103061676,
      "learning_rate": 2.1125724535240555e-05,
      "loss": 0.8066259026527405,
      "step": 3587
    },
    {
      "epoch": 0.2938304210300034,
      "grad_norm": 0.17098784446716309,
      "learning_rate": 2.1124099778113024e-05,
      "loss": 0.6190654635429382,
      "step": 3588
    },
    {
      "epoch": 0.2939123135665223,
      "grad_norm": 0.2020869255065918,
      "learning_rate": 2.1122474624414537e-05,
      "loss": 0.826740562915802,
      "step": 3589
    },
    {
      "epoch": 0.2939942061030413,
      "grad_norm": 0.183569997549057,
      "learning_rate": 2.112084907421573e-05,
      "loss": 0.6702662110328674,
      "step": 3590
    },
    {
      "epoch": 0.29407609863956025,
      "grad_norm": 0.18634085357189178,
      "learning_rate": 2.1119223127587257e-05,
      "loss": 0.7662432193756104,
      "step": 3591
    },
    {
      "epoch": 0.2941579911760792,
      "grad_norm": 0.19652844965457916,
      "learning_rate": 2.1117596784599787e-05,
      "loss": 1.0405168533325195,
      "step": 3592
    },
    {
      "epoch": 0.2942398837125981,
      "grad_norm": 0.1693311482667923,
      "learning_rate": 2.111597004532401e-05,
      "loss": 1.054413080215454,
      "step": 3593
    },
    {
      "epoch": 0.2943217762491171,
      "grad_norm": 0.21019431948661804,
      "learning_rate": 2.111434290983063e-05,
      "loss": 0.9166066646575928,
      "step": 3594
    },
    {
      "epoch": 0.29440366878563606,
      "grad_norm": 0.1755557507276535,
      "learning_rate": 2.1112715378190358e-05,
      "loss": 1.0017973184585571,
      "step": 3595
    },
    {
      "epoch": 0.294485561322155,
      "grad_norm": 0.18864481151103973,
      "learning_rate": 2.1111087450473954e-05,
      "loss": 0.947195291519165,
      "step": 3596
    },
    {
      "epoch": 0.29456745385867394,
      "grad_norm": 0.20033425092697144,
      "learning_rate": 2.1109459126752158e-05,
      "loss": 0.9460005760192871,
      "step": 3597
    },
    {
      "epoch": 0.2946493463951929,
      "grad_norm": 0.18150371313095093,
      "learning_rate": 2.110783040709575e-05,
      "loss": 1.0964206457138062,
      "step": 3598
    },
    {
      "epoch": 0.2947312389317119,
      "grad_norm": 0.1885889321565628,
      "learning_rate": 2.110620129157552e-05,
      "loss": 0.8664576411247253,
      "step": 3599
    },
    {
      "epoch": 0.2948131314682308,
      "grad_norm": 0.1925138235092163,
      "learning_rate": 2.1104571780262277e-05,
      "loss": 1.0306211709976196,
      "step": 3600
    },
    {
      "epoch": 0.29489502400474976,
      "grad_norm": 0.22584065794944763,
      "learning_rate": 2.1102941873226848e-05,
      "loss": 0.9187113046646118,
      "step": 3601
    },
    {
      "epoch": 0.29497691654126873,
      "grad_norm": 0.1997673362493515,
      "learning_rate": 2.1101311570540073e-05,
      "loss": 0.8119727373123169,
      "step": 3602
    },
    {
      "epoch": 0.2950588090777877,
      "grad_norm": 0.20866994559764862,
      "learning_rate": 2.1099680872272807e-05,
      "loss": 0.817253828048706,
      "step": 3603
    },
    {
      "epoch": 0.2951407016143066,
      "grad_norm": 0.19118531048297882,
      "learning_rate": 2.1098049778495935e-05,
      "loss": 0.9325932264328003,
      "step": 3604
    },
    {
      "epoch": 0.2952225941508256,
      "grad_norm": 0.18405868113040924,
      "learning_rate": 2.109641828928035e-05,
      "loss": 0.5527405738830566,
      "step": 3605
    },
    {
      "epoch": 0.29530448668734455,
      "grad_norm": 0.18512536585330963,
      "learning_rate": 2.1094786404696955e-05,
      "loss": 0.5500252842903137,
      "step": 3606
    },
    {
      "epoch": 0.29538637922386346,
      "grad_norm": 0.1829407811164856,
      "learning_rate": 2.1093154124816682e-05,
      "loss": 0.7503373622894287,
      "step": 3607
    },
    {
      "epoch": 0.29546827176038243,
      "grad_norm": 0.3849031925201416,
      "learning_rate": 2.1091521449710487e-05,
      "loss": 0.8927099704742432,
      "step": 3608
    },
    {
      "epoch": 0.2955501642969014,
      "grad_norm": 0.19822444021701813,
      "learning_rate": 2.108988837944932e-05,
      "loss": 0.9564392566680908,
      "step": 3609
    },
    {
      "epoch": 0.29563205683342036,
      "grad_norm": 0.1835123598575592,
      "learning_rate": 2.1088254914104164e-05,
      "loss": 0.8965294361114502,
      "step": 3610
    },
    {
      "epoch": 0.2957139493699393,
      "grad_norm": 0.17707829177379608,
      "learning_rate": 2.1086621053746022e-05,
      "loss": 0.772913932800293,
      "step": 3611
    },
    {
      "epoch": 0.29579584190645825,
      "grad_norm": 0.23680436611175537,
      "learning_rate": 2.10849867984459e-05,
      "loss": 1.0605396032333374,
      "step": 3612
    },
    {
      "epoch": 0.2958777344429772,
      "grad_norm": 0.1721189171075821,
      "learning_rate": 2.1083352148274836e-05,
      "loss": 1.0507632493972778,
      "step": 3613
    },
    {
      "epoch": 0.2959596269794962,
      "grad_norm": 0.19446878135204315,
      "learning_rate": 2.1081717103303876e-05,
      "loss": 0.9577060341835022,
      "step": 3614
    },
    {
      "epoch": 0.2960415195160151,
      "grad_norm": 0.18839064240455627,
      "learning_rate": 2.1080081663604083e-05,
      "loss": 0.657987117767334,
      "step": 3615
    },
    {
      "epoch": 0.29612341205253406,
      "grad_norm": 0.17192865908145905,
      "learning_rate": 2.1078445829246544e-05,
      "loss": 0.7006998062133789,
      "step": 3616
    },
    {
      "epoch": 0.29620530458905303,
      "grad_norm": 0.20551976561546326,
      "learning_rate": 2.107680960030236e-05,
      "loss": 0.7701177597045898,
      "step": 3617
    },
    {
      "epoch": 0.29628719712557194,
      "grad_norm": 0.21339356899261475,
      "learning_rate": 2.1075172976842645e-05,
      "loss": 1.0517058372497559,
      "step": 3618
    },
    {
      "epoch": 0.2963690896620909,
      "grad_norm": 0.18560940027236938,
      "learning_rate": 2.1073535958938535e-05,
      "loss": 0.5469334721565247,
      "step": 3619
    },
    {
      "epoch": 0.2964509821986099,
      "grad_norm": 0.1744692623615265,
      "learning_rate": 2.107189854666118e-05,
      "loss": 0.8858669996261597,
      "step": 3620
    },
    {
      "epoch": 0.29653287473512885,
      "grad_norm": 0.21723338961601257,
      "learning_rate": 2.107026074008175e-05,
      "loss": 0.8496415019035339,
      "step": 3621
    },
    {
      "epoch": 0.29661476727164776,
      "grad_norm": 0.18160609900951385,
      "learning_rate": 2.106862253927143e-05,
      "loss": 0.8248187899589539,
      "step": 3622
    },
    {
      "epoch": 0.29669665980816673,
      "grad_norm": 0.19062523543834686,
      "learning_rate": 2.1066983944301424e-05,
      "loss": 1.081024408340454,
      "step": 3623
    },
    {
      "epoch": 0.2967785523446857,
      "grad_norm": 0.1949036568403244,
      "learning_rate": 2.1065344955242955e-05,
      "loss": 0.8662595748901367,
      "step": 3624
    },
    {
      "epoch": 0.29686044488120467,
      "grad_norm": 0.20386208593845367,
      "learning_rate": 2.106370557216725e-05,
      "loss": 0.7789850234985352,
      "step": 3625
    },
    {
      "epoch": 0.2969423374177236,
      "grad_norm": 0.19800038635730743,
      "learning_rate": 2.1062065795145576e-05,
      "loss": 0.7887215614318848,
      "step": 3626
    },
    {
      "epoch": 0.29702422995424255,
      "grad_norm": 0.18346603214740753,
      "learning_rate": 2.1060425624249195e-05,
      "loss": 0.8028406500816345,
      "step": 3627
    },
    {
      "epoch": 0.2971061224907615,
      "grad_norm": 0.21444019675254822,
      "learning_rate": 2.1058785059549402e-05,
      "loss": 0.9322720766067505,
      "step": 3628
    },
    {
      "epoch": 0.2971880150272804,
      "grad_norm": 0.20921628177165985,
      "learning_rate": 2.1057144101117498e-05,
      "loss": 0.8555570840835571,
      "step": 3629
    },
    {
      "epoch": 0.2972699075637994,
      "grad_norm": 0.21037790179252625,
      "learning_rate": 2.105550274902481e-05,
      "loss": 0.8249996900558472,
      "step": 3630
    },
    {
      "epoch": 0.29735180010031836,
      "grad_norm": 0.2035447657108307,
      "learning_rate": 2.105386100334267e-05,
      "loss": 0.8506997227668762,
      "step": 3631
    },
    {
      "epoch": 0.29743369263683733,
      "grad_norm": 0.1981995403766632,
      "learning_rate": 2.1052218864142448e-05,
      "loss": 1.1441645622253418,
      "step": 3632
    },
    {
      "epoch": 0.29751558517335625,
      "grad_norm": 0.18912583589553833,
      "learning_rate": 2.1050576331495508e-05,
      "loss": 0.7843077182769775,
      "step": 3633
    },
    {
      "epoch": 0.2975974777098752,
      "grad_norm": 0.20738671720027924,
      "learning_rate": 2.1048933405473243e-05,
      "loss": 1.1583847999572754,
      "step": 3634
    },
    {
      "epoch": 0.2976793702463942,
      "grad_norm": 0.2356564700603485,
      "learning_rate": 2.1047290086147066e-05,
      "loss": 0.7506566047668457,
      "step": 3635
    },
    {
      "epoch": 0.29776126278291315,
      "grad_norm": 0.2145264446735382,
      "learning_rate": 2.1045646373588394e-05,
      "loss": 0.8833611011505127,
      "step": 3636
    },
    {
      "epoch": 0.29784315531943206,
      "grad_norm": 0.2108660191297531,
      "learning_rate": 2.1044002267868677e-05,
      "loss": 0.9278150200843811,
      "step": 3637
    },
    {
      "epoch": 0.29792504785595103,
      "grad_norm": 0.21484282612800598,
      "learning_rate": 2.104235776905937e-05,
      "loss": 1.1076394319534302,
      "step": 3638
    },
    {
      "epoch": 0.29800694039247,
      "grad_norm": 0.20101384818553925,
      "learning_rate": 2.104071287723195e-05,
      "loss": 1.1551094055175781,
      "step": 3639
    },
    {
      "epoch": 0.2980888329289889,
      "grad_norm": 0.1831672042608261,
      "learning_rate": 2.1039067592457916e-05,
      "loss": 0.7470964789390564,
      "step": 3640
    },
    {
      "epoch": 0.2981707254655079,
      "grad_norm": 0.1978040337562561,
      "learning_rate": 2.1037421914808773e-05,
      "loss": 0.7266411781311035,
      "step": 3641
    },
    {
      "epoch": 0.29825261800202685,
      "grad_norm": 0.20666398108005524,
      "learning_rate": 2.1035775844356053e-05,
      "loss": 0.9648134708404541,
      "step": 3642
    },
    {
      "epoch": 0.2983345105385458,
      "grad_norm": 0.2071513831615448,
      "learning_rate": 2.1034129381171303e-05,
      "loss": 0.6932756304740906,
      "step": 3643
    },
    {
      "epoch": 0.29841640307506473,
      "grad_norm": 0.1790638267993927,
      "learning_rate": 2.1032482525326073e-05,
      "loss": 0.8512274026870728,
      "step": 3644
    },
    {
      "epoch": 0.2984982956115837,
      "grad_norm": 0.18330150842666626,
      "learning_rate": 2.1030835276891958e-05,
      "loss": 0.6236366033554077,
      "step": 3645
    },
    {
      "epoch": 0.29858018814810267,
      "grad_norm": 0.25310254096984863,
      "learning_rate": 2.102918763594054e-05,
      "loss": 0.9596943855285645,
      "step": 3646
    },
    {
      "epoch": 0.2986620806846216,
      "grad_norm": 0.16413013637065887,
      "learning_rate": 2.102753960254344e-05,
      "loss": 0.5448846220970154,
      "step": 3647
    },
    {
      "epoch": 0.29874397322114055,
      "grad_norm": 0.23257330060005188,
      "learning_rate": 2.102589117677229e-05,
      "loss": 0.9468790292739868,
      "step": 3648
    },
    {
      "epoch": 0.2988258657576595,
      "grad_norm": 0.20402218401432037,
      "learning_rate": 2.1024242358698734e-05,
      "loss": 0.9778248071670532,
      "step": 3649
    },
    {
      "epoch": 0.2989077582941785,
      "grad_norm": 0.19620724022388458,
      "learning_rate": 2.1022593148394438e-05,
      "loss": 0.7206214666366577,
      "step": 3650
    },
    {
      "epoch": 0.2989896508306974,
      "grad_norm": 0.1872878074645996,
      "learning_rate": 2.1020943545931083e-05,
      "loss": 1.1113299131393433,
      "step": 3651
    },
    {
      "epoch": 0.29907154336721636,
      "grad_norm": 0.1915750503540039,
      "learning_rate": 2.101929355138036e-05,
      "loss": 1.2483373880386353,
      "step": 3652
    },
    {
      "epoch": 0.29915343590373533,
      "grad_norm": 0.20338261127471924,
      "learning_rate": 2.1017643164813998e-05,
      "loss": 0.9897397756576538,
      "step": 3653
    },
    {
      "epoch": 0.2992353284402543,
      "grad_norm": 0.19959871470928192,
      "learning_rate": 2.101599238630372e-05,
      "loss": 0.6907796263694763,
      "step": 3654
    },
    {
      "epoch": 0.2993172209767732,
      "grad_norm": 0.22270812094211578,
      "learning_rate": 2.101434121592128e-05,
      "loss": 0.6984301209449768,
      "step": 3655
    },
    {
      "epoch": 0.2993991135132922,
      "grad_norm": 0.17777365446090698,
      "learning_rate": 2.1012689653738446e-05,
      "loss": 0.6999461054801941,
      "step": 3656
    },
    {
      "epoch": 0.29948100604981115,
      "grad_norm": 0.21583420038223267,
      "learning_rate": 2.1011037699826997e-05,
      "loss": 0.8415200710296631,
      "step": 3657
    },
    {
      "epoch": 0.29956289858633006,
      "grad_norm": 0.2374778836965561,
      "learning_rate": 2.1009385354258735e-05,
      "loss": 1.0429660081863403,
      "step": 3658
    },
    {
      "epoch": 0.29964479112284903,
      "grad_norm": 0.22632156312465668,
      "learning_rate": 2.1007732617105474e-05,
      "loss": 1.031368374824524,
      "step": 3659
    },
    {
      "epoch": 0.299726683659368,
      "grad_norm": 0.2669917643070221,
      "learning_rate": 2.100607948843906e-05,
      "loss": 0.9242194294929504,
      "step": 3660
    },
    {
      "epoch": 0.29980857619588697,
      "grad_norm": 0.24219609797000885,
      "learning_rate": 2.100442596833134e-05,
      "loss": 1.129044532775879,
      "step": 3661
    },
    {
      "epoch": 0.2998904687324059,
      "grad_norm": 0.2114255726337433,
      "learning_rate": 2.1002772056854174e-05,
      "loss": 1.0678277015686035,
      "step": 3662
    },
    {
      "epoch": 0.29997236126892485,
      "grad_norm": 0.1788359135389328,
      "learning_rate": 2.100111775407946e-05,
      "loss": 0.947293221950531,
      "step": 3663
    },
    {
      "epoch": 0.3000542538054438,
      "grad_norm": 0.20146743953227997,
      "learning_rate": 2.0999463060079095e-05,
      "loss": 1.0194687843322754,
      "step": 3664
    },
    {
      "epoch": 0.3001361463419628,
      "grad_norm": 0.1982983499765396,
      "learning_rate": 2.0997807974925e-05,
      "loss": 1.2301114797592163,
      "step": 3665
    },
    {
      "epoch": 0.3002180388784817,
      "grad_norm": 0.17931246757507324,
      "learning_rate": 2.0996152498689105e-05,
      "loss": 0.6765968203544617,
      "step": 3666
    },
    {
      "epoch": 0.30029993141500066,
      "grad_norm": 0.17129762470722198,
      "learning_rate": 2.0994496631443377e-05,
      "loss": 0.8984649181365967,
      "step": 3667
    },
    {
      "epoch": 0.30038182395151963,
      "grad_norm": 0.21193966269493103,
      "learning_rate": 2.0992840373259775e-05,
      "loss": 0.9020853042602539,
      "step": 3668
    },
    {
      "epoch": 0.30046371648803855,
      "grad_norm": 0.19254326820373535,
      "learning_rate": 2.0991183724210293e-05,
      "loss": 0.8749078512191772,
      "step": 3669
    },
    {
      "epoch": 0.3005456090245575,
      "grad_norm": 0.20751403272151947,
      "learning_rate": 2.098952668436693e-05,
      "loss": 1.054942011833191,
      "step": 3670
    },
    {
      "epoch": 0.3006275015610765,
      "grad_norm": 0.18410690128803253,
      "learning_rate": 2.0987869253801725e-05,
      "loss": 0.5626079440116882,
      "step": 3671
    },
    {
      "epoch": 0.30070939409759545,
      "grad_norm": 0.1679917722940445,
      "learning_rate": 2.0986211432586695e-05,
      "loss": 0.6299710273742676,
      "step": 3672
    },
    {
      "epoch": 0.30079128663411436,
      "grad_norm": 0.20177562534809113,
      "learning_rate": 2.0984553220793904e-05,
      "loss": 0.8676398992538452,
      "step": 3673
    },
    {
      "epoch": 0.30087317917063333,
      "grad_norm": 0.20765089988708496,
      "learning_rate": 2.0982894618495427e-05,
      "loss": 1.1479179859161377,
      "step": 3674
    },
    {
      "epoch": 0.3009550717071523,
      "grad_norm": 0.22143487632274628,
      "learning_rate": 2.098123562576335e-05,
      "loss": 0.6360337138175964,
      "step": 3675
    },
    {
      "epoch": 0.30103696424367127,
      "grad_norm": 0.21387112140655518,
      "learning_rate": 2.0979576242669785e-05,
      "loss": 0.9845086336135864,
      "step": 3676
    },
    {
      "epoch": 0.3011188567801902,
      "grad_norm": 0.20906053483486176,
      "learning_rate": 2.0977916469286853e-05,
      "loss": 0.8634927272796631,
      "step": 3677
    },
    {
      "epoch": 0.30120074931670915,
      "grad_norm": 0.19571611285209656,
      "learning_rate": 2.097625630568669e-05,
      "loss": 0.9353699684143066,
      "step": 3678
    },
    {
      "epoch": 0.3012826418532281,
      "grad_norm": 0.20345059037208557,
      "learning_rate": 2.097459575194146e-05,
      "loss": 0.9322031140327454,
      "step": 3679
    },
    {
      "epoch": 0.30136453438974703,
      "grad_norm": 0.20729608833789825,
      "learning_rate": 2.0972934808123334e-05,
      "loss": 0.9988180994987488,
      "step": 3680
    },
    {
      "epoch": 0.301446426926266,
      "grad_norm": 0.20294074714183807,
      "learning_rate": 2.0971273474304505e-05,
      "loss": 0.9455978870391846,
      "step": 3681
    },
    {
      "epoch": 0.30152831946278497,
      "grad_norm": 0.23743915557861328,
      "learning_rate": 2.0969611750557182e-05,
      "loss": 0.7566089034080505,
      "step": 3682
    },
    {
      "epoch": 0.30161021199930393,
      "grad_norm": 0.19040416181087494,
      "learning_rate": 2.0967949636953588e-05,
      "loss": 0.9789379835128784,
      "step": 3683
    },
    {
      "epoch": 0.30169210453582285,
      "grad_norm": 0.2011021375656128,
      "learning_rate": 2.0966287133565966e-05,
      "loss": 0.8621591925621033,
      "step": 3684
    },
    {
      "epoch": 0.3017739970723418,
      "grad_norm": 0.19810828566551208,
      "learning_rate": 2.0964624240466577e-05,
      "loss": 1.105165719985962,
      "step": 3685
    },
    {
      "epoch": 0.3018558896088608,
      "grad_norm": 0.21400846540927887,
      "learning_rate": 2.096296095772769e-05,
      "loss": 1.1357966661453247,
      "step": 3686
    },
    {
      "epoch": 0.30193778214537975,
      "grad_norm": 0.1794648915529251,
      "learning_rate": 2.0961297285421613e-05,
      "loss": 0.5741147994995117,
      "step": 3687
    },
    {
      "epoch": 0.30201967468189866,
      "grad_norm": 0.2044583112001419,
      "learning_rate": 2.095963322362064e-05,
      "loss": 0.6140956282615662,
      "step": 3688
    },
    {
      "epoch": 0.30210156721841763,
      "grad_norm": 0.19455432891845703,
      "learning_rate": 2.095796877239711e-05,
      "loss": 0.8320263624191284,
      "step": 3689
    },
    {
      "epoch": 0.3021834597549366,
      "grad_norm": 0.19657950103282928,
      "learning_rate": 2.0956303931823364e-05,
      "loss": 0.7934436798095703,
      "step": 3690
    },
    {
      "epoch": 0.3022653522914555,
      "grad_norm": 0.20718331634998322,
      "learning_rate": 2.0954638701971753e-05,
      "loss": 0.994743824005127,
      "step": 3691
    },
    {
      "epoch": 0.3023472448279745,
      "grad_norm": 0.18333777785301208,
      "learning_rate": 2.0952973082914666e-05,
      "loss": 0.963508129119873,
      "step": 3692
    },
    {
      "epoch": 0.30242913736449345,
      "grad_norm": 0.2397436797618866,
      "learning_rate": 2.0951307074724494e-05,
      "loss": 1.0450435876846313,
      "step": 3693
    },
    {
      "epoch": 0.3025110299010124,
      "grad_norm": 0.2019968330860138,
      "learning_rate": 2.094964067747365e-05,
      "loss": 0.9910508990287781,
      "step": 3694
    },
    {
      "epoch": 0.30259292243753133,
      "grad_norm": 0.2190040796995163,
      "learning_rate": 2.0947973891234565e-05,
      "loss": 0.8946284055709839,
      "step": 3695
    },
    {
      "epoch": 0.3026748149740503,
      "grad_norm": 0.17628934979438782,
      "learning_rate": 2.0946306716079674e-05,
      "loss": 0.9257820844650269,
      "step": 3696
    },
    {
      "epoch": 0.30275670751056927,
      "grad_norm": 0.1866961419582367,
      "learning_rate": 2.0944639152081448e-05,
      "loss": 0.6957502961158752,
      "step": 3697
    },
    {
      "epoch": 0.30283860004708824,
      "grad_norm": 0.20171265304088593,
      "learning_rate": 2.0942971199312368e-05,
      "loss": 0.9175857901573181,
      "step": 3698
    },
    {
      "epoch": 0.30292049258360715,
      "grad_norm": 0.213101327419281,
      "learning_rate": 2.0941302857844924e-05,
      "loss": 0.8888790011405945,
      "step": 3699
    },
    {
      "epoch": 0.3030023851201261,
      "grad_norm": 0.1841278076171875,
      "learning_rate": 2.0939634127751625e-05,
      "loss": 0.6397982835769653,
      "step": 3700
    },
    {
      "epoch": 0.3030842776566451,
      "grad_norm": 0.1682266741991043,
      "learning_rate": 2.0937965009105014e-05,
      "loss": 0.8855237364768982,
      "step": 3701
    },
    {
      "epoch": 0.303166170193164,
      "grad_norm": 0.2096906453371048,
      "learning_rate": 2.093629550197763e-05,
      "loss": 0.663582444190979,
      "step": 3702
    },
    {
      "epoch": 0.30324806272968297,
      "grad_norm": 0.16171486675739288,
      "learning_rate": 2.0934625606442037e-05,
      "loss": 0.4826012849807739,
      "step": 3703
    },
    {
      "epoch": 0.30332995526620193,
      "grad_norm": 0.20908123254776,
      "learning_rate": 2.093295532257082e-05,
      "loss": 1.1438027620315552,
      "step": 3704
    },
    {
      "epoch": 0.3034118478027209,
      "grad_norm": 0.18108218908309937,
      "learning_rate": 2.0931284650436567e-05,
      "loss": 0.6727376580238342,
      "step": 3705
    },
    {
      "epoch": 0.3034937403392398,
      "grad_norm": 0.24184975028038025,
      "learning_rate": 2.09296135901119e-05,
      "loss": 1.028395414352417,
      "step": 3706
    },
    {
      "epoch": 0.3035756328757588,
      "grad_norm": 0.23232777416706085,
      "learning_rate": 2.0927942141669443e-05,
      "loss": 1.176581859588623,
      "step": 3707
    },
    {
      "epoch": 0.30365752541227775,
      "grad_norm": 0.19576068222522736,
      "learning_rate": 2.092627030518186e-05,
      "loss": 0.8960968852043152,
      "step": 3708
    },
    {
      "epoch": 0.30373941794879666,
      "grad_norm": 0.1940532773733139,
      "learning_rate": 2.0924598080721797e-05,
      "loss": 0.7640482187271118,
      "step": 3709
    },
    {
      "epoch": 0.30382131048531563,
      "grad_norm": 0.2615938186645508,
      "learning_rate": 2.0922925468361946e-05,
      "loss": 0.803968071937561,
      "step": 3710
    },
    {
      "epoch": 0.3039032030218346,
      "grad_norm": 0.2292097806930542,
      "learning_rate": 2.0921252468175e-05,
      "loss": 0.8943720459938049,
      "step": 3711
    },
    {
      "epoch": 0.30398509555835357,
      "grad_norm": 0.1992519348859787,
      "learning_rate": 2.0919579080233683e-05,
      "loss": 1.2030646800994873,
      "step": 3712
    },
    {
      "epoch": 0.3040669880948725,
      "grad_norm": 0.19640801846981049,
      "learning_rate": 2.0917905304610716e-05,
      "loss": 0.9638792276382446,
      "step": 3713
    },
    {
      "epoch": 0.30414888063139145,
      "grad_norm": 0.18813134729862213,
      "learning_rate": 2.091623114137886e-05,
      "loss": 0.7159693241119385,
      "step": 3714
    },
    {
      "epoch": 0.3042307731679104,
      "grad_norm": 0.22801652550697327,
      "learning_rate": 2.091455659061087e-05,
      "loss": 0.8400739431381226,
      "step": 3715
    },
    {
      "epoch": 0.3043126657044294,
      "grad_norm": 0.20077286660671234,
      "learning_rate": 2.091288165237954e-05,
      "loss": 0.6092285513877869,
      "step": 3716
    },
    {
      "epoch": 0.3043945582409483,
      "grad_norm": 0.18133825063705444,
      "learning_rate": 2.091120632675766e-05,
      "loss": 0.5924459099769592,
      "step": 3717
    },
    {
      "epoch": 0.30447645077746727,
      "grad_norm": 0.21254850924015045,
      "learning_rate": 2.0909530613818054e-05,
      "loss": 1.1440659761428833,
      "step": 3718
    },
    {
      "epoch": 0.30455834331398624,
      "grad_norm": 0.18105019629001617,
      "learning_rate": 2.0907854513633547e-05,
      "loss": 0.7164005637168884,
      "step": 3719
    },
    {
      "epoch": 0.30464023585050515,
      "grad_norm": 0.21055519580841064,
      "learning_rate": 2.0906178026276998e-05,
      "loss": 0.9109532237052917,
      "step": 3720
    },
    {
      "epoch": 0.3047221283870241,
      "grad_norm": 0.20325630903244019,
      "learning_rate": 2.090450115182127e-05,
      "loss": 0.9352898597717285,
      "step": 3721
    },
    {
      "epoch": 0.3048040209235431,
      "grad_norm": 0.15850035846233368,
      "learning_rate": 2.0902823890339247e-05,
      "loss": 0.8019141554832458,
      "step": 3722
    },
    {
      "epoch": 0.30488591346006205,
      "grad_norm": 0.203294575214386,
      "learning_rate": 2.0901146241903826e-05,
      "loss": 0.9413224458694458,
      "step": 3723
    },
    {
      "epoch": 0.30496780599658097,
      "grad_norm": 0.19006268680095673,
      "learning_rate": 2.0899468206587932e-05,
      "loss": 0.7488620281219482,
      "step": 3724
    },
    {
      "epoch": 0.30504969853309993,
      "grad_norm": 0.20841634273529053,
      "learning_rate": 2.0897789784464492e-05,
      "loss": 0.8511260151863098,
      "step": 3725
    },
    {
      "epoch": 0.3051315910696189,
      "grad_norm": 0.21967197954654694,
      "learning_rate": 2.0896110975606462e-05,
      "loss": 1.0204687118530273,
      "step": 3726
    },
    {
      "epoch": 0.30521348360613787,
      "grad_norm": 0.1968212127685547,
      "learning_rate": 2.0894431780086813e-05,
      "loss": 0.8045575618743896,
      "step": 3727
    },
    {
      "epoch": 0.3052953761426568,
      "grad_norm": 0.1779811680316925,
      "learning_rate": 2.0892752197978518e-05,
      "loss": 0.6743385791778564,
      "step": 3728
    },
    {
      "epoch": 0.30537726867917575,
      "grad_norm": 0.20954355597496033,
      "learning_rate": 2.089107222935459e-05,
      "loss": 0.7055544257164001,
      "step": 3729
    },
    {
      "epoch": 0.3054591612156947,
      "grad_norm": 0.1805024892091751,
      "learning_rate": 2.0889391874288044e-05,
      "loss": 0.74539715051651,
      "step": 3730
    },
    {
      "epoch": 0.30554105375221363,
      "grad_norm": 0.196786031126976,
      "learning_rate": 2.0887711132851914e-05,
      "loss": 0.8997848033905029,
      "step": 3731
    },
    {
      "epoch": 0.3056229462887326,
      "grad_norm": 0.217910036444664,
      "learning_rate": 2.0886030005119252e-05,
      "loss": 0.81126469373703,
      "step": 3732
    },
    {
      "epoch": 0.30570483882525157,
      "grad_norm": 0.2062082588672638,
      "learning_rate": 2.0884348491163127e-05,
      "loss": 0.7422431707382202,
      "step": 3733
    },
    {
      "epoch": 0.30578673136177054,
      "grad_norm": 0.21547353267669678,
      "learning_rate": 2.0882666591056625e-05,
      "loss": 1.1990283727645874,
      "step": 3734
    },
    {
      "epoch": 0.30586862389828945,
      "grad_norm": 0.19358514249324799,
      "learning_rate": 2.0880984304872848e-05,
      "loss": 1.0718698501586914,
      "step": 3735
    },
    {
      "epoch": 0.3059505164348084,
      "grad_norm": 0.1743157058954239,
      "learning_rate": 2.0879301632684915e-05,
      "loss": 0.5294317007064819,
      "step": 3736
    },
    {
      "epoch": 0.3060324089713274,
      "grad_norm": 0.20120902359485626,
      "learning_rate": 2.087761857456596e-05,
      "loss": 1.0308455228805542,
      "step": 3737
    },
    {
      "epoch": 0.30611430150784635,
      "grad_norm": 0.23224692046642303,
      "learning_rate": 2.0875935130589136e-05,
      "loss": 0.7668212056159973,
      "step": 3738
    },
    {
      "epoch": 0.30619619404436527,
      "grad_norm": 0.21504060924053192,
      "learning_rate": 2.087425130082762e-05,
      "loss": 0.7705087065696716,
      "step": 3739
    },
    {
      "epoch": 0.30627808658088423,
      "grad_norm": 0.19296769797801971,
      "learning_rate": 2.087256708535459e-05,
      "loss": 0.999333381652832,
      "step": 3740
    },
    {
      "epoch": 0.3063599791174032,
      "grad_norm": 0.19944867491722107,
      "learning_rate": 2.087088248424325e-05,
      "loss": 0.952528715133667,
      "step": 3741
    },
    {
      "epoch": 0.3064418716539221,
      "grad_norm": 0.17591442167758942,
      "learning_rate": 2.086919749756682e-05,
      "loss": 0.48611778020858765,
      "step": 3742
    },
    {
      "epoch": 0.3065237641904411,
      "grad_norm": 0.1993882805109024,
      "learning_rate": 2.0867512125398538e-05,
      "loss": 0.6830252408981323,
      "step": 3743
    },
    {
      "epoch": 0.30660565672696005,
      "grad_norm": 0.19135212898254395,
      "learning_rate": 2.0865826367811654e-05,
      "loss": 0.7023112773895264,
      "step": 3744
    },
    {
      "epoch": 0.306687549263479,
      "grad_norm": 0.22780735790729523,
      "learning_rate": 2.0864140224879446e-05,
      "loss": 0.8729267120361328,
      "step": 3745
    },
    {
      "epoch": 0.30676944179999793,
      "grad_norm": 0.17214713990688324,
      "learning_rate": 2.086245369667519e-05,
      "loss": 0.7706936001777649,
      "step": 3746
    },
    {
      "epoch": 0.3068513343365169,
      "grad_norm": 0.1965707242488861,
      "learning_rate": 2.0860766783272195e-05,
      "loss": 0.8064629435539246,
      "step": 3747
    },
    {
      "epoch": 0.30693322687303587,
      "grad_norm": 0.20206986367702484,
      "learning_rate": 2.0859079484743783e-05,
      "loss": 0.8335772752761841,
      "step": 3748
    },
    {
      "epoch": 0.30701511940955484,
      "grad_norm": 0.18004991114139557,
      "learning_rate": 2.085739180116329e-05,
      "loss": 0.6776199340820312,
      "step": 3749
    },
    {
      "epoch": 0.30709701194607375,
      "grad_norm": 0.18720576167106628,
      "learning_rate": 2.0855703732604067e-05,
      "loss": 0.9954450726509094,
      "step": 3750
    },
    {
      "epoch": 0.3071789044825927,
      "grad_norm": 0.2131120264530182,
      "learning_rate": 2.0854015279139482e-05,
      "loss": 1.1569010019302368,
      "step": 3751
    },
    {
      "epoch": 0.3072607970191117,
      "grad_norm": 0.18281470239162445,
      "learning_rate": 2.0852326440842927e-05,
      "loss": 0.8434798121452332,
      "step": 3752
    },
    {
      "epoch": 0.3073426895556306,
      "grad_norm": 0.2585003972053528,
      "learning_rate": 2.0850637217787808e-05,
      "loss": 0.7501276731491089,
      "step": 3753
    },
    {
      "epoch": 0.30742458209214957,
      "grad_norm": 0.20072004199028015,
      "learning_rate": 2.0848947610047543e-05,
      "loss": 1.2636387348175049,
      "step": 3754
    },
    {
      "epoch": 0.30750647462866854,
      "grad_norm": 0.19847123324871063,
      "learning_rate": 2.0847257617695567e-05,
      "loss": 0.8386142253875732,
      "step": 3755
    },
    {
      "epoch": 0.3075883671651875,
      "grad_norm": 0.23981152474880219,
      "learning_rate": 2.0845567240805337e-05,
      "loss": 0.8591972589492798,
      "step": 3756
    },
    {
      "epoch": 0.3076702597017064,
      "grad_norm": 0.17332816123962402,
      "learning_rate": 2.084387647945032e-05,
      "loss": 0.805083155632019,
      "step": 3757
    },
    {
      "epoch": 0.3077521522382254,
      "grad_norm": 0.19180920720100403,
      "learning_rate": 2.084218533370401e-05,
      "loss": 0.8333313465118408,
      "step": 3758
    },
    {
      "epoch": 0.30783404477474435,
      "grad_norm": 0.17884868383407593,
      "learning_rate": 2.0840493803639902e-05,
      "loss": 1.0697728395462036,
      "step": 3759
    },
    {
      "epoch": 0.3079159373112633,
      "grad_norm": 0.1998935341835022,
      "learning_rate": 2.0838801889331527e-05,
      "loss": 0.9401984810829163,
      "step": 3760
    },
    {
      "epoch": 0.30799782984778223,
      "grad_norm": 0.17782531678676605,
      "learning_rate": 2.0837109590852412e-05,
      "loss": 0.8681778907775879,
      "step": 3761
    },
    {
      "epoch": 0.3080797223843012,
      "grad_norm": 0.18978998064994812,
      "learning_rate": 2.083541690827612e-05,
      "loss": 0.691510021686554,
      "step": 3762
    },
    {
      "epoch": 0.30816161492082017,
      "grad_norm": 0.19456076622009277,
      "learning_rate": 2.083372384167622e-05,
      "loss": 0.9550209045410156,
      "step": 3763
    },
    {
      "epoch": 0.3082435074573391,
      "grad_norm": 0.1908067762851715,
      "learning_rate": 2.0832030391126295e-05,
      "loss": 0.8028213977813721,
      "step": 3764
    },
    {
      "epoch": 0.30832539999385805,
      "grad_norm": 0.18947221338748932,
      "learning_rate": 2.0830336556699954e-05,
      "loss": 1.0443196296691895,
      "step": 3765
    },
    {
      "epoch": 0.308407292530377,
      "grad_norm": 0.20464542508125305,
      "learning_rate": 2.082864233847082e-05,
      "loss": 0.6871306896209717,
      "step": 3766
    },
    {
      "epoch": 0.308489185066896,
      "grad_norm": 0.21797701716423035,
      "learning_rate": 2.0826947736512523e-05,
      "loss": 0.6797203421592712,
      "step": 3767
    },
    {
      "epoch": 0.3085710776034149,
      "grad_norm": 0.2328663021326065,
      "learning_rate": 2.0825252750898723e-05,
      "loss": 0.907656192779541,
      "step": 3768
    },
    {
      "epoch": 0.30865297013993387,
      "grad_norm": 0.1911153495311737,
      "learning_rate": 2.0823557381703093e-05,
      "loss": 1.2358806133270264,
      "step": 3769
    },
    {
      "epoch": 0.30873486267645284,
      "grad_norm": 0.19534823298454285,
      "learning_rate": 2.0821861628999316e-05,
      "loss": 0.8565807342529297,
      "step": 3770
    },
    {
      "epoch": 0.3088167552129718,
      "grad_norm": 0.20858164131641388,
      "learning_rate": 2.0820165492861094e-05,
      "loss": 1.083613395690918,
      "step": 3771
    },
    {
      "epoch": 0.3088986477494907,
      "grad_norm": 0.18433508276939392,
      "learning_rate": 2.0818468973362158e-05,
      "loss": 0.7031400799751282,
      "step": 3772
    },
    {
      "epoch": 0.3089805402860097,
      "grad_norm": 0.21164171397686005,
      "learning_rate": 2.0816772070576236e-05,
      "loss": 0.7776554226875305,
      "step": 3773
    },
    {
      "epoch": 0.30906243282252865,
      "grad_norm": 0.19280529022216797,
      "learning_rate": 2.081507478457709e-05,
      "loss": 1.0052731037139893,
      "step": 3774
    },
    {
      "epoch": 0.30914432535904757,
      "grad_norm": 0.17771291732788086,
      "learning_rate": 2.0813377115438485e-05,
      "loss": 0.8894796371459961,
      "step": 3775
    },
    {
      "epoch": 0.30922621789556654,
      "grad_norm": 0.18012332916259766,
      "learning_rate": 2.0811679063234206e-05,
      "loss": 0.6829866766929626,
      "step": 3776
    },
    {
      "epoch": 0.3093081104320855,
      "grad_norm": 0.22205588221549988,
      "learning_rate": 2.080998062803807e-05,
      "loss": 0.8511006832122803,
      "step": 3777
    },
    {
      "epoch": 0.30939000296860447,
      "grad_norm": 0.187313973903656,
      "learning_rate": 2.0808281809923885e-05,
      "loss": 0.7881535291671753,
      "step": 3778
    },
    {
      "epoch": 0.3094718955051234,
      "grad_norm": 0.18435177206993103,
      "learning_rate": 2.0806582608965497e-05,
      "loss": 0.5894652009010315,
      "step": 3779
    },
    {
      "epoch": 0.30955378804164235,
      "grad_norm": 0.20513075590133667,
      "learning_rate": 2.0804883025236755e-05,
      "loss": 0.6846891641616821,
      "step": 3780
    },
    {
      "epoch": 0.3096356805781613,
      "grad_norm": 0.17659910023212433,
      "learning_rate": 2.080318305881153e-05,
      "loss": 0.6493246555328369,
      "step": 3781
    },
    {
      "epoch": 0.30971757311468023,
      "grad_norm": 0.171564519405365,
      "learning_rate": 2.080148270976372e-05,
      "loss": 0.939162015914917,
      "step": 3782
    },
    {
      "epoch": 0.3097994656511992,
      "grad_norm": 0.19607193768024445,
      "learning_rate": 2.0799781978167215e-05,
      "loss": 0.9123477935791016,
      "step": 3783
    },
    {
      "epoch": 0.30988135818771817,
      "grad_norm": 0.23267166316509247,
      "learning_rate": 2.079808086409594e-05,
      "loss": 0.977313756942749,
      "step": 3784
    },
    {
      "epoch": 0.30996325072423714,
      "grad_norm": 0.20389822125434875,
      "learning_rate": 2.0796379367623838e-05,
      "loss": 0.9569587707519531,
      "step": 3785
    },
    {
      "epoch": 0.31004514326075605,
      "grad_norm": 0.24494490027427673,
      "learning_rate": 2.0794677488824858e-05,
      "loss": 0.9752430319786072,
      "step": 3786
    },
    {
      "epoch": 0.310127035797275,
      "grad_norm": 0.21150928735733032,
      "learning_rate": 2.079297522777297e-05,
      "loss": 1.0910792350769043,
      "step": 3787
    },
    {
      "epoch": 0.310208928333794,
      "grad_norm": 0.17977339029312134,
      "learning_rate": 2.0791272584542162e-05,
      "loss": 0.7511414289474487,
      "step": 3788
    },
    {
      "epoch": 0.31029082087031296,
      "grad_norm": 0.17239928245544434,
      "learning_rate": 2.0789569559206442e-05,
      "loss": 0.7230886816978455,
      "step": 3789
    },
    {
      "epoch": 0.31037271340683187,
      "grad_norm": 0.1916889101266861,
      "learning_rate": 2.0787866151839826e-05,
      "loss": 0.8523291945457458,
      "step": 3790
    },
    {
      "epoch": 0.31045460594335084,
      "grad_norm": 0.19170795381069183,
      "learning_rate": 2.0786162362516353e-05,
      "loss": 0.6724187135696411,
      "step": 3791
    },
    {
      "epoch": 0.3105364984798698,
      "grad_norm": 0.19682304561138153,
      "learning_rate": 2.0784458191310074e-05,
      "loss": 1.1722692251205444,
      "step": 3792
    },
    {
      "epoch": 0.3106183910163887,
      "grad_norm": 0.16867810487747192,
      "learning_rate": 2.0782753638295066e-05,
      "loss": 0.7759914398193359,
      "step": 3793
    },
    {
      "epoch": 0.3107002835529077,
      "grad_norm": 0.1667218655347824,
      "learning_rate": 2.0781048703545408e-05,
      "loss": 0.884178638458252,
      "step": 3794
    },
    {
      "epoch": 0.31078217608942665,
      "grad_norm": 0.22587884962558746,
      "learning_rate": 2.077934338713521e-05,
      "loss": 0.9652723073959351,
      "step": 3795
    },
    {
      "epoch": 0.3108640686259456,
      "grad_norm": 0.24535419046878815,
      "learning_rate": 2.0777637689138583e-05,
      "loss": 0.8766284584999084,
      "step": 3796
    },
    {
      "epoch": 0.31094596116246453,
      "grad_norm": 0.1690375804901123,
      "learning_rate": 2.0775931609629673e-05,
      "loss": 0.848909318447113,
      "step": 3797
    },
    {
      "epoch": 0.3110278536989835,
      "grad_norm": 0.23776361346244812,
      "learning_rate": 2.077422514868263e-05,
      "loss": 0.9482001662254333,
      "step": 3798
    },
    {
      "epoch": 0.31110974623550247,
      "grad_norm": 0.18520626425743103,
      "learning_rate": 2.0772518306371625e-05,
      "loss": 1.0880141258239746,
      "step": 3799
    },
    {
      "epoch": 0.31119163877202144,
      "grad_norm": 0.17754656076431274,
      "learning_rate": 2.077081108277084e-05,
      "loss": 0.5987615585327148,
      "step": 3800
    },
    {
      "epoch": 0.31127353130854035,
      "grad_norm": 0.22625593841075897,
      "learning_rate": 2.0769103477954484e-05,
      "loss": 0.8761833310127258,
      "step": 3801
    },
    {
      "epoch": 0.3113554238450593,
      "grad_norm": 0.20020852982997894,
      "learning_rate": 2.076739549199677e-05,
      "loss": 0.9232507348060608,
      "step": 3802
    },
    {
      "epoch": 0.3114373163815783,
      "grad_norm": 0.21406102180480957,
      "learning_rate": 2.076568712497194e-05,
      "loss": 0.9071295857429504,
      "step": 3803
    },
    {
      "epoch": 0.3115192089180972,
      "grad_norm": 0.19963616132736206,
      "learning_rate": 2.0763978376954245e-05,
      "loss": 1.0720038414001465,
      "step": 3804
    },
    {
      "epoch": 0.31160110145461617,
      "grad_norm": 0.1983698308467865,
      "learning_rate": 2.0762269248017953e-05,
      "loss": 0.8587870001792908,
      "step": 3805
    },
    {
      "epoch": 0.31168299399113514,
      "grad_norm": 0.18734951317310333,
      "learning_rate": 2.076055973823735e-05,
      "loss": 1.0613579750061035,
      "step": 3806
    },
    {
      "epoch": 0.3117648865276541,
      "grad_norm": 0.20103240013122559,
      "learning_rate": 2.0758849847686737e-05,
      "loss": 0.6078776717185974,
      "step": 3807
    },
    {
      "epoch": 0.311846779064173,
      "grad_norm": 0.2544565796852112,
      "learning_rate": 2.075713957644044e-05,
      "loss": 0.8925833702087402,
      "step": 3808
    },
    {
      "epoch": 0.311928671600692,
      "grad_norm": 0.1899135559797287,
      "learning_rate": 2.0755428924572784e-05,
      "loss": 0.8773936629295349,
      "step": 3809
    },
    {
      "epoch": 0.31201056413721096,
      "grad_norm": 0.1791934221982956,
      "learning_rate": 2.0753717892158127e-05,
      "loss": 0.9249658584594727,
      "step": 3810
    },
    {
      "epoch": 0.3120924566737299,
      "grad_norm": 0.19495287537574768,
      "learning_rate": 2.075200647927084e-05,
      "loss": 0.6954169869422913,
      "step": 3811
    },
    {
      "epoch": 0.31217434921024884,
      "grad_norm": 0.18937784433364868,
      "learning_rate": 2.0750294685985296e-05,
      "loss": 0.9945257902145386,
      "step": 3812
    },
    {
      "epoch": 0.3122562417467678,
      "grad_norm": 0.20374411344528198,
      "learning_rate": 2.0748582512375916e-05,
      "loss": 0.9466389417648315,
      "step": 3813
    },
    {
      "epoch": 0.3123381342832868,
      "grad_norm": 0.21982556581497192,
      "learning_rate": 2.07468699585171e-05,
      "loss": 0.8534730076789856,
      "step": 3814
    },
    {
      "epoch": 0.3124200268198057,
      "grad_norm": 0.16985943913459778,
      "learning_rate": 2.0745157024483292e-05,
      "loss": 0.8691702485084534,
      "step": 3815
    },
    {
      "epoch": 0.31250191935632465,
      "grad_norm": 0.19603653252124786,
      "learning_rate": 2.0743443710348938e-05,
      "loss": 0.8510909080505371,
      "step": 3816
    },
    {
      "epoch": 0.3125838118928436,
      "grad_norm": 0.20499542355537415,
      "learning_rate": 2.0741730016188514e-05,
      "loss": 1.2135144472122192,
      "step": 3817
    },
    {
      "epoch": 0.3126657044293626,
      "grad_norm": 0.18448813259601593,
      "learning_rate": 2.0740015942076497e-05,
      "loss": 0.9232240319252014,
      "step": 3818
    },
    {
      "epoch": 0.3127475969658815,
      "grad_norm": 0.2162674516439438,
      "learning_rate": 2.0738301488087385e-05,
      "loss": 0.7483606934547424,
      "step": 3819
    },
    {
      "epoch": 0.31282948950240047,
      "grad_norm": 0.19455885887145996,
      "learning_rate": 2.0736586654295705e-05,
      "loss": 0.8170048594474792,
      "step": 3820
    },
    {
      "epoch": 0.31291138203891944,
      "grad_norm": 0.18474586308002472,
      "learning_rate": 2.0734871440775987e-05,
      "loss": 0.9258907437324524,
      "step": 3821
    },
    {
      "epoch": 0.3129932745754384,
      "grad_norm": 0.18167629837989807,
      "learning_rate": 2.0733155847602777e-05,
      "loss": 0.8377065658569336,
      "step": 3822
    },
    {
      "epoch": 0.3130751671119573,
      "grad_norm": 0.15868346393108368,
      "learning_rate": 2.0731439874850644e-05,
      "loss": 0.9797303080558777,
      "step": 3823
    },
    {
      "epoch": 0.3131570596484763,
      "grad_norm": 0.1682647466659546,
      "learning_rate": 2.0729723522594174e-05,
      "loss": 0.9187793731689453,
      "step": 3824
    },
    {
      "epoch": 0.31323895218499526,
      "grad_norm": 0.20399266481399536,
      "learning_rate": 2.0728006790907963e-05,
      "loss": 0.8983795642852783,
      "step": 3825
    },
    {
      "epoch": 0.31332084472151417,
      "grad_norm": 0.1870231032371521,
      "learning_rate": 2.072628967986663e-05,
      "loss": 0.8341652154922485,
      "step": 3826
    },
    {
      "epoch": 0.31340273725803314,
      "grad_norm": 0.21393652260303497,
      "learning_rate": 2.0724572189544806e-05,
      "loss": 0.8763335943222046,
      "step": 3827
    },
    {
      "epoch": 0.3134846297945521,
      "grad_norm": 0.20966021716594696,
      "learning_rate": 2.0722854320017144e-05,
      "loss": 0.9868144989013672,
      "step": 3828
    },
    {
      "epoch": 0.3135665223310711,
      "grad_norm": 0.20558182895183563,
      "learning_rate": 2.07211360713583e-05,
      "loss": 0.6942949295043945,
      "step": 3829
    },
    {
      "epoch": 0.31364841486759,
      "grad_norm": 0.21020591259002686,
      "learning_rate": 2.0719417443642966e-05,
      "loss": 0.9508772492408752,
      "step": 3830
    },
    {
      "epoch": 0.31373030740410895,
      "grad_norm": 0.20934821665287018,
      "learning_rate": 2.0717698436945836e-05,
      "loss": 0.877099335193634,
      "step": 3831
    },
    {
      "epoch": 0.3138121999406279,
      "grad_norm": 0.17882463335990906,
      "learning_rate": 2.071597905134163e-05,
      "loss": 0.8389298915863037,
      "step": 3832
    },
    {
      "epoch": 0.3138940924771469,
      "grad_norm": 0.16195642948150635,
      "learning_rate": 2.0714259286905072e-05,
      "loss": 0.5474529266357422,
      "step": 3833
    },
    {
      "epoch": 0.3139759850136658,
      "grad_norm": 0.18354681134223938,
      "learning_rate": 2.0712539143710912e-05,
      "loss": 0.7336528897285461,
      "step": 3834
    },
    {
      "epoch": 0.31405787755018477,
      "grad_norm": 0.19657662510871887,
      "learning_rate": 2.0710818621833925e-05,
      "loss": 0.8177025318145752,
      "step": 3835
    },
    {
      "epoch": 0.31413977008670374,
      "grad_norm": 0.1851317286491394,
      "learning_rate": 2.0709097721348877e-05,
      "loss": 0.8647388219833374,
      "step": 3836
    },
    {
      "epoch": 0.31422166262322265,
      "grad_norm": 0.20581874251365662,
      "learning_rate": 2.070737644233057e-05,
      "loss": 0.701156735420227,
      "step": 3837
    },
    {
      "epoch": 0.3143035551597416,
      "grad_norm": 0.18253971636295319,
      "learning_rate": 2.070565478485383e-05,
      "loss": 0.9010382294654846,
      "step": 3838
    },
    {
      "epoch": 0.3143854476962606,
      "grad_norm": 0.1823616474866867,
      "learning_rate": 2.0703932748993465e-05,
      "loss": 0.6611784100532532,
      "step": 3839
    },
    {
      "epoch": 0.31446734023277956,
      "grad_norm": 0.14296896755695343,
      "learning_rate": 2.0702210334824342e-05,
      "loss": 0.5418720245361328,
      "step": 3840
    },
    {
      "epoch": 0.31454923276929847,
      "grad_norm": 0.23105500638484955,
      "learning_rate": 2.070048754242131e-05,
      "loss": 1.025282382965088,
      "step": 3841
    },
    {
      "epoch": 0.31463112530581744,
      "grad_norm": 0.20904365181922913,
      "learning_rate": 2.069876437185926e-05,
      "loss": 0.7960243821144104,
      "step": 3842
    },
    {
      "epoch": 0.3147130178423364,
      "grad_norm": 0.18670852482318878,
      "learning_rate": 2.069704082321308e-05,
      "loss": 0.826521635055542,
      "step": 3843
    },
    {
      "epoch": 0.3147949103788553,
      "grad_norm": 0.17215362191200256,
      "learning_rate": 2.0695316896557685e-05,
      "loss": 0.7121895551681519,
      "step": 3844
    },
    {
      "epoch": 0.3148768029153743,
      "grad_norm": 0.22568632662296295,
      "learning_rate": 2.0693592591968006e-05,
      "loss": 0.8222243785858154,
      "step": 3845
    },
    {
      "epoch": 0.31495869545189326,
      "grad_norm": 0.2369709610939026,
      "learning_rate": 2.0691867909518985e-05,
      "loss": 1.3335106372833252,
      "step": 3846
    },
    {
      "epoch": 0.3150405879884122,
      "grad_norm": 0.2030957043170929,
      "learning_rate": 2.069014284928559e-05,
      "loss": 0.7807875871658325,
      "step": 3847
    },
    {
      "epoch": 0.31512248052493114,
      "grad_norm": 0.20485574007034302,
      "learning_rate": 2.068841741134279e-05,
      "loss": 0.9713330864906311,
      "step": 3848
    },
    {
      "epoch": 0.3152043730614501,
      "grad_norm": 0.20400424301624298,
      "learning_rate": 2.0686691595765585e-05,
      "loss": 1.037530779838562,
      "step": 3849
    },
    {
      "epoch": 0.3152862655979691,
      "grad_norm": 0.2002319097518921,
      "learning_rate": 2.0684965402628987e-05,
      "loss": 0.7449715733528137,
      "step": 3850
    },
    {
      "epoch": 0.31536815813448804,
      "grad_norm": 0.16794659197330475,
      "learning_rate": 2.068323883200802e-05,
      "loss": 0.9423976540565491,
      "step": 3851
    },
    {
      "epoch": 0.31545005067100695,
      "grad_norm": 0.18332268297672272,
      "learning_rate": 2.0681511883977728e-05,
      "loss": 0.615775465965271,
      "step": 3852
    },
    {
      "epoch": 0.3155319432075259,
      "grad_norm": 0.2267131358385086,
      "learning_rate": 2.0679784558613178e-05,
      "loss": 0.6362888813018799,
      "step": 3853
    },
    {
      "epoch": 0.3156138357440449,
      "grad_norm": 0.2197464555501938,
      "learning_rate": 2.0678056855989437e-05,
      "loss": 0.938831090927124,
      "step": 3854
    },
    {
      "epoch": 0.3156957282805638,
      "grad_norm": 0.21812167763710022,
      "learning_rate": 2.0676328776181606e-05,
      "loss": 0.7093120217323303,
      "step": 3855
    },
    {
      "epoch": 0.31577762081708277,
      "grad_norm": 0.15155361592769623,
      "learning_rate": 2.067460031926479e-05,
      "loss": 0.9182919263839722,
      "step": 3856
    },
    {
      "epoch": 0.31585951335360174,
      "grad_norm": 0.21244792640209198,
      "learning_rate": 2.0672871485314114e-05,
      "loss": 1.2241127490997314,
      "step": 3857
    },
    {
      "epoch": 0.3159414058901207,
      "grad_norm": 0.21129639446735382,
      "learning_rate": 2.0671142274404723e-05,
      "loss": 0.9119269251823425,
      "step": 3858
    },
    {
      "epoch": 0.3160232984266396,
      "grad_norm": 0.18024909496307373,
      "learning_rate": 2.066941268661178e-05,
      "loss": 0.7217423915863037,
      "step": 3859
    },
    {
      "epoch": 0.3161051909631586,
      "grad_norm": 0.2295406460762024,
      "learning_rate": 2.066768272201045e-05,
      "loss": 0.8601570725440979,
      "step": 3860
    },
    {
      "epoch": 0.31618708349967756,
      "grad_norm": 0.21038605272769928,
      "learning_rate": 2.0665952380675933e-05,
      "loss": 0.7318704724311829,
      "step": 3861
    },
    {
      "epoch": 0.3162689760361965,
      "grad_norm": 0.1876082569360733,
      "learning_rate": 2.066422166268343e-05,
      "loss": 0.858354389667511,
      "step": 3862
    },
    {
      "epoch": 0.31635086857271544,
      "grad_norm": 0.19325366616249084,
      "learning_rate": 2.066249056810817e-05,
      "loss": 0.6785359382629395,
      "step": 3863
    },
    {
      "epoch": 0.3164327611092344,
      "grad_norm": 0.19640369713306427,
      "learning_rate": 2.0660759097025386e-05,
      "loss": 0.927278459072113,
      "step": 3864
    },
    {
      "epoch": 0.3165146536457534,
      "grad_norm": 0.1848151832818985,
      "learning_rate": 2.065902724951035e-05,
      "loss": 0.558994710445404,
      "step": 3865
    },
    {
      "epoch": 0.3165965461822723,
      "grad_norm": 0.25601357221603394,
      "learning_rate": 2.065729502563832e-05,
      "loss": 1.0158015489578247,
      "step": 3866
    },
    {
      "epoch": 0.31667843871879126,
      "grad_norm": 0.17892087996006012,
      "learning_rate": 2.0655562425484594e-05,
      "loss": 0.7735381722450256,
      "step": 3867
    },
    {
      "epoch": 0.3167603312553102,
      "grad_norm": 0.2392985224723816,
      "learning_rate": 2.0653829449124473e-05,
      "loss": 1.0902955532073975,
      "step": 3868
    },
    {
      "epoch": 0.3168422237918292,
      "grad_norm": 0.1831737458705902,
      "learning_rate": 2.065209609663328e-05,
      "loss": 0.8795191645622253,
      "step": 3869
    },
    {
      "epoch": 0.3169241163283481,
      "grad_norm": 0.2295103520154953,
      "learning_rate": 2.065036236808636e-05,
      "loss": 0.6277364492416382,
      "step": 3870
    },
    {
      "epoch": 0.3170060088648671,
      "grad_norm": 0.19100867211818695,
      "learning_rate": 2.0648628263559063e-05,
      "loss": 0.8783537149429321,
      "step": 3871
    },
    {
      "epoch": 0.31708790140138604,
      "grad_norm": 0.19775338470935822,
      "learning_rate": 2.0646893783126756e-05,
      "loss": 0.7873702049255371,
      "step": 3872
    },
    {
      "epoch": 0.317169793937905,
      "grad_norm": 0.21302008628845215,
      "learning_rate": 2.0645158926864835e-05,
      "loss": 0.7487426996231079,
      "step": 3873
    },
    {
      "epoch": 0.3172516864744239,
      "grad_norm": 0.17428874969482422,
      "learning_rate": 2.06434236948487e-05,
      "loss": 0.8748592138290405,
      "step": 3874
    },
    {
      "epoch": 0.3173335790109429,
      "grad_norm": 0.1516123265028,
      "learning_rate": 2.0641688087153768e-05,
      "loss": 0.7990954518318176,
      "step": 3875
    },
    {
      "epoch": 0.31741547154746186,
      "grad_norm": 0.18156829476356506,
      "learning_rate": 2.0639952103855483e-05,
      "loss": 0.8907844424247742,
      "step": 3876
    },
    {
      "epoch": 0.31749736408398077,
      "grad_norm": 0.18587413430213928,
      "learning_rate": 2.0638215745029292e-05,
      "loss": 1.0132131576538086,
      "step": 3877
    },
    {
      "epoch": 0.31757925662049974,
      "grad_norm": 0.19086430966854095,
      "learning_rate": 2.0636479010750668e-05,
      "loss": 0.737686038017273,
      "step": 3878
    },
    {
      "epoch": 0.3176611491570187,
      "grad_norm": 0.1720389574766159,
      "learning_rate": 2.0634741901095096e-05,
      "loss": 0.8174468278884888,
      "step": 3879
    },
    {
      "epoch": 0.3177430416935377,
      "grad_norm": 0.17553453147411346,
      "learning_rate": 2.0633004416138076e-05,
      "loss": 0.9217116832733154,
      "step": 3880
    },
    {
      "epoch": 0.3178249342300566,
      "grad_norm": 0.1806309074163437,
      "learning_rate": 2.0631266555955127e-05,
      "loss": 0.7567261457443237,
      "step": 3881
    },
    {
      "epoch": 0.31790682676657556,
      "grad_norm": 0.21313153207302094,
      "learning_rate": 2.0629528320621784e-05,
      "loss": 1.0394160747528076,
      "step": 3882
    },
    {
      "epoch": 0.3179887193030945,
      "grad_norm": 0.18041832745075226,
      "learning_rate": 2.0627789710213598e-05,
      "loss": 0.8996294736862183,
      "step": 3883
    },
    {
      "epoch": 0.3180706118396135,
      "grad_norm": 0.18491987884044647,
      "learning_rate": 2.0626050724806136e-05,
      "loss": 0.6309583783149719,
      "step": 3884
    },
    {
      "epoch": 0.3181525043761324,
      "grad_norm": 0.21424968540668488,
      "learning_rate": 2.0624311364474982e-05,
      "loss": 0.8386402726173401,
      "step": 3885
    },
    {
      "epoch": 0.3182343969126514,
      "grad_norm": 0.20529955625534058,
      "learning_rate": 2.062257162929573e-05,
      "loss": 0.8438444137573242,
      "step": 3886
    },
    {
      "epoch": 0.31831628944917034,
      "grad_norm": 0.16787421703338623,
      "learning_rate": 2.062083151934401e-05,
      "loss": 0.8958845734596252,
      "step": 3887
    },
    {
      "epoch": 0.31839818198568925,
      "grad_norm": 0.18092255294322968,
      "learning_rate": 2.0619091034695438e-05,
      "loss": 1.0243592262268066,
      "step": 3888
    },
    {
      "epoch": 0.3184800745222082,
      "grad_norm": 0.18010175228118896,
      "learning_rate": 2.0617350175425677e-05,
      "loss": 0.8616993427276611,
      "step": 3889
    },
    {
      "epoch": 0.3185619670587272,
      "grad_norm": 0.20390157401561737,
      "learning_rate": 2.061560894161038e-05,
      "loss": 0.8038257360458374,
      "step": 3890
    },
    {
      "epoch": 0.31864385959524616,
      "grad_norm": 0.1707228720188141,
      "learning_rate": 2.0613867333325238e-05,
      "loss": 0.7912411093711853,
      "step": 3891
    },
    {
      "epoch": 0.31872575213176507,
      "grad_norm": 0.17128688097000122,
      "learning_rate": 2.0612125350645938e-05,
      "loss": 0.9279229044914246,
      "step": 3892
    },
    {
      "epoch": 0.31880764466828404,
      "grad_norm": 0.22608759999275208,
      "learning_rate": 2.0610382993648203e-05,
      "loss": 0.7735222578048706,
      "step": 3893
    },
    {
      "epoch": 0.318889537204803,
      "grad_norm": 0.19106890261173248,
      "learning_rate": 2.060864026240776e-05,
      "loss": 0.9058697819709778,
      "step": 3894
    },
    {
      "epoch": 0.318971429741322,
      "grad_norm": 0.18119873106479645,
      "learning_rate": 2.060689715700035e-05,
      "loss": 0.7403808832168579,
      "step": 3895
    },
    {
      "epoch": 0.3190533222778409,
      "grad_norm": 0.21368330717086792,
      "learning_rate": 2.0605153677501744e-05,
      "loss": 1.1984975337982178,
      "step": 3896
    },
    {
      "epoch": 0.31913521481435986,
      "grad_norm": 0.19462525844573975,
      "learning_rate": 2.0603409823987716e-05,
      "loss": 0.8307121992111206,
      "step": 3897
    },
    {
      "epoch": 0.3192171073508788,
      "grad_norm": 0.21561211347579956,
      "learning_rate": 2.0601665596534063e-05,
      "loss": 0.6790074110031128,
      "step": 3898
    },
    {
      "epoch": 0.31929899988739774,
      "grad_norm": 0.2133370041847229,
      "learning_rate": 2.0599920995216598e-05,
      "loss": 0.6817212104797363,
      "step": 3899
    },
    {
      "epoch": 0.3193808924239167,
      "grad_norm": 0.19220006465911865,
      "learning_rate": 2.0598176020111144e-05,
      "loss": 0.9359793663024902,
      "step": 3900
    },
    {
      "epoch": 0.3194627849604357,
      "grad_norm": 0.21502521634101868,
      "learning_rate": 2.0596430671293547e-05,
      "loss": 0.9597976803779602,
      "step": 3901
    },
    {
      "epoch": 0.31954467749695464,
      "grad_norm": 0.1860586553812027,
      "learning_rate": 2.0594684948839668e-05,
      "loss": 0.8201647996902466,
      "step": 3902
    },
    {
      "epoch": 0.31962657003347356,
      "grad_norm": 0.18704071640968323,
      "learning_rate": 2.059293885282538e-05,
      "loss": 0.7935498952865601,
      "step": 3903
    },
    {
      "epoch": 0.3197084625699925,
      "grad_norm": 0.19323976337909698,
      "learning_rate": 2.0591192383326578e-05,
      "loss": 0.8398987054824829,
      "step": 3904
    },
    {
      "epoch": 0.3197903551065115,
      "grad_norm": 0.20410950481891632,
      "learning_rate": 2.058944554041917e-05,
      "loss": 0.9792617559432983,
      "step": 3905
    },
    {
      "epoch": 0.31987224764303046,
      "grad_norm": 0.15536758303642273,
      "learning_rate": 2.0587698324179083e-05,
      "loss": 0.8749412894248962,
      "step": 3906
    },
    {
      "epoch": 0.3199541401795494,
      "grad_norm": 0.21461927890777588,
      "learning_rate": 2.0585950734682256e-05,
      "loss": 0.7866345643997192,
      "step": 3907
    },
    {
      "epoch": 0.32003603271606834,
      "grad_norm": 0.1821107566356659,
      "learning_rate": 2.0584202772004647e-05,
      "loss": 0.9668945670127869,
      "step": 3908
    },
    {
      "epoch": 0.3201179252525873,
      "grad_norm": 0.18682432174682617,
      "learning_rate": 2.058245443622223e-05,
      "loss": 0.9067835807800293,
      "step": 3909
    },
    {
      "epoch": 0.3201998177891062,
      "grad_norm": 0.18764114379882812,
      "learning_rate": 2.0580705727410995e-05,
      "loss": 0.723134458065033,
      "step": 3910
    },
    {
      "epoch": 0.3202817103256252,
      "grad_norm": 0.17350256443023682,
      "learning_rate": 2.0578956645646948e-05,
      "loss": 0.6418245434761047,
      "step": 3911
    },
    {
      "epoch": 0.32036360286214416,
      "grad_norm": 0.2288198322057724,
      "learning_rate": 2.057720719100611e-05,
      "loss": 0.856052577495575,
      "step": 3912
    },
    {
      "epoch": 0.3204454953986631,
      "grad_norm": 0.20863474905490875,
      "learning_rate": 2.0575457363564524e-05,
      "loss": 1.129311203956604,
      "step": 3913
    },
    {
      "epoch": 0.32052738793518204,
      "grad_norm": 0.1890038698911667,
      "learning_rate": 2.0573707163398238e-05,
      "loss": 0.9141160249710083,
      "step": 3914
    },
    {
      "epoch": 0.320609280471701,
      "grad_norm": 0.17243625223636627,
      "learning_rate": 2.0571956590583326e-05,
      "loss": 0.8704214692115784,
      "step": 3915
    },
    {
      "epoch": 0.32069117300822,
      "grad_norm": 0.17833374440670013,
      "learning_rate": 2.0570205645195874e-05,
      "loss": 0.4970870316028595,
      "step": 3916
    },
    {
      "epoch": 0.3207730655447389,
      "grad_norm": 0.1937279850244522,
      "learning_rate": 2.056845432731199e-05,
      "loss": 0.7491456270217896,
      "step": 3917
    },
    {
      "epoch": 0.32085495808125786,
      "grad_norm": 0.15782898664474487,
      "learning_rate": 2.0566702637007787e-05,
      "loss": 0.6254806518554688,
      "step": 3918
    },
    {
      "epoch": 0.3209368506177768,
      "grad_norm": 0.1783123016357422,
      "learning_rate": 2.0564950574359405e-05,
      "loss": 0.8593987226486206,
      "step": 3919
    },
    {
      "epoch": 0.3210187431542958,
      "grad_norm": 0.18524810671806335,
      "learning_rate": 2.0563198139442997e-05,
      "loss": 0.755302369594574,
      "step": 3920
    },
    {
      "epoch": 0.3211006356908147,
      "grad_norm": 0.16901125013828278,
      "learning_rate": 2.0561445332334724e-05,
      "loss": 0.7434175610542297,
      "step": 3921
    },
    {
      "epoch": 0.3211825282273337,
      "grad_norm": 0.18124042451381683,
      "learning_rate": 2.0559692153110777e-05,
      "loss": 0.6324748992919922,
      "step": 3922
    },
    {
      "epoch": 0.32126442076385264,
      "grad_norm": 0.21092143654823303,
      "learning_rate": 2.0557938601847357e-05,
      "loss": 0.7077627182006836,
      "step": 3923
    },
    {
      "epoch": 0.3213463133003716,
      "grad_norm": 0.18164531886577606,
      "learning_rate": 2.0556184678620674e-05,
      "loss": 0.9406149983406067,
      "step": 3924
    },
    {
      "epoch": 0.3214282058368905,
      "grad_norm": 0.20724418759346008,
      "learning_rate": 2.0554430383506967e-05,
      "loss": 0.947371780872345,
      "step": 3925
    },
    {
      "epoch": 0.3215100983734095,
      "grad_norm": 0.2119489163160324,
      "learning_rate": 2.0552675716582483e-05,
      "loss": 0.6114541292190552,
      "step": 3926
    },
    {
      "epoch": 0.32159199090992846,
      "grad_norm": 0.20484659075737,
      "learning_rate": 2.055092067792348e-05,
      "loss": 0.8839597702026367,
      "step": 3927
    },
    {
      "epoch": 0.3216738834464474,
      "grad_norm": 0.19731709361076355,
      "learning_rate": 2.0549165267606254e-05,
      "loss": 0.6500949859619141,
      "step": 3928
    },
    {
      "epoch": 0.32175577598296634,
      "grad_norm": 0.1871248185634613,
      "learning_rate": 2.054740948570709e-05,
      "loss": 0.7536279559135437,
      "step": 3929
    },
    {
      "epoch": 0.3218376685194853,
      "grad_norm": 0.21729592978954315,
      "learning_rate": 2.0545653332302305e-05,
      "loss": 0.8361296057701111,
      "step": 3930
    },
    {
      "epoch": 0.3219195610560043,
      "grad_norm": 0.17965973913669586,
      "learning_rate": 2.054389680746823e-05,
      "loss": 0.9434899091720581,
      "step": 3931
    },
    {
      "epoch": 0.3220014535925232,
      "grad_norm": 0.1846047192811966,
      "learning_rate": 2.054213991128121e-05,
      "loss": 0.8550084233283997,
      "step": 3932
    },
    {
      "epoch": 0.32208334612904216,
      "grad_norm": 0.22459964454174042,
      "learning_rate": 2.0540382643817612e-05,
      "loss": 1.0315570831298828,
      "step": 3933
    },
    {
      "epoch": 0.3221652386655611,
      "grad_norm": 0.1752089112997055,
      "learning_rate": 2.0538625005153804e-05,
      "loss": 0.9271630048751831,
      "step": 3934
    },
    {
      "epoch": 0.3222471312020801,
      "grad_norm": 0.24773702025413513,
      "learning_rate": 2.0536866995366184e-05,
      "loss": 0.8134306073188782,
      "step": 3935
    },
    {
      "epoch": 0.322329023738599,
      "grad_norm": 0.1779778003692627,
      "learning_rate": 2.0535108614531163e-05,
      "loss": 0.567272961139679,
      "step": 3936
    },
    {
      "epoch": 0.322410916275118,
      "grad_norm": 0.19251899421215057,
      "learning_rate": 2.053334986272517e-05,
      "loss": 0.641463577747345,
      "step": 3937
    },
    {
      "epoch": 0.32249280881163694,
      "grad_norm": 0.17036749422550201,
      "learning_rate": 2.0531590740024647e-05,
      "loss": 0.9544930458068848,
      "step": 3938
    },
    {
      "epoch": 0.32257470134815586,
      "grad_norm": 0.2248890995979309,
      "learning_rate": 2.052983124650605e-05,
      "loss": 0.8219738006591797,
      "step": 3939
    },
    {
      "epoch": 0.3226565938846748,
      "grad_norm": 0.2041727751493454,
      "learning_rate": 2.052807138224586e-05,
      "loss": 0.8931393623352051,
      "step": 3940
    },
    {
      "epoch": 0.3227384864211938,
      "grad_norm": 0.17000088095664978,
      "learning_rate": 2.0526311147320555e-05,
      "loss": 0.855679452419281,
      "step": 3941
    },
    {
      "epoch": 0.32282037895771276,
      "grad_norm": 0.2026071399450302,
      "learning_rate": 2.0524550541806654e-05,
      "loss": 0.7965473532676697,
      "step": 3942
    },
    {
      "epoch": 0.3229022714942317,
      "grad_norm": 0.18399696052074432,
      "learning_rate": 2.0522789565780678e-05,
      "loss": 0.803577721118927,
      "step": 3943
    },
    {
      "epoch": 0.32298416403075064,
      "grad_norm": 0.2059074491262436,
      "learning_rate": 2.0521028219319163e-05,
      "loss": 1.0486431121826172,
      "step": 3944
    },
    {
      "epoch": 0.3230660565672696,
      "grad_norm": 0.18338526785373688,
      "learning_rate": 2.0519266502498666e-05,
      "loss": 1.0016177892684937,
      "step": 3945
    },
    {
      "epoch": 0.3231479491037886,
      "grad_norm": 0.1801069974899292,
      "learning_rate": 2.0517504415395758e-05,
      "loss": 0.5547131896018982,
      "step": 3946
    },
    {
      "epoch": 0.3232298416403075,
      "grad_norm": 0.1768687665462494,
      "learning_rate": 2.0515741958087027e-05,
      "loss": 1.0042048692703247,
      "step": 3947
    },
    {
      "epoch": 0.32331173417682646,
      "grad_norm": 0.17749008536338806,
      "learning_rate": 2.051397913064908e-05,
      "loss": 0.6840186715126038,
      "step": 3948
    },
    {
      "epoch": 0.32339362671334543,
      "grad_norm": 0.1998671144247055,
      "learning_rate": 2.051221593315853e-05,
      "loss": 0.9369029998779297,
      "step": 3949
    },
    {
      "epoch": 0.32347551924986434,
      "grad_norm": 0.19199851155281067,
      "learning_rate": 2.051045236569202e-05,
      "loss": 0.8841372728347778,
      "step": 3950
    },
    {
      "epoch": 0.3235574117863833,
      "grad_norm": 0.22909732162952423,
      "learning_rate": 2.0508688428326194e-05,
      "loss": 0.6738160848617554,
      "step": 3951
    },
    {
      "epoch": 0.3236393043229023,
      "grad_norm": 0.1914035528898239,
      "learning_rate": 2.0506924121137723e-05,
      "loss": 0.7841907143592834,
      "step": 3952
    },
    {
      "epoch": 0.32372119685942125,
      "grad_norm": 0.18163597583770752,
      "learning_rate": 2.0505159444203294e-05,
      "loss": 1.0898945331573486,
      "step": 3953
    },
    {
      "epoch": 0.32380308939594016,
      "grad_norm": 0.21326129138469696,
      "learning_rate": 2.0503394397599608e-05,
      "loss": 0.7765082716941833,
      "step": 3954
    },
    {
      "epoch": 0.3238849819324591,
      "grad_norm": 0.1583007276058197,
      "learning_rate": 2.0501628981403374e-05,
      "loss": 0.7911266684532166,
      "step": 3955
    },
    {
      "epoch": 0.3239668744689781,
      "grad_norm": 0.17539963126182556,
      "learning_rate": 2.0499863195691335e-05,
      "loss": 0.9357081651687622,
      "step": 3956
    },
    {
      "epoch": 0.32404876700549706,
      "grad_norm": 0.18082186579704285,
      "learning_rate": 2.049809704054023e-05,
      "loss": 0.7464720606803894,
      "step": 3957
    },
    {
      "epoch": 0.324130659542016,
      "grad_norm": 0.1655406951904297,
      "learning_rate": 2.0496330516026828e-05,
      "loss": 0.8147081136703491,
      "step": 3958
    },
    {
      "epoch": 0.32421255207853494,
      "grad_norm": 0.2123180329799652,
      "learning_rate": 2.0494563622227903e-05,
      "loss": 0.8541117310523987,
      "step": 3959
    },
    {
      "epoch": 0.3242944446150539,
      "grad_norm": 0.22505545616149902,
      "learning_rate": 2.0492796359220263e-05,
      "loss": 0.960929274559021,
      "step": 3960
    },
    {
      "epoch": 0.3243763371515728,
      "grad_norm": 0.1861141473054886,
      "learning_rate": 2.049102872708071e-05,
      "loss": 0.8769974112510681,
      "step": 3961
    },
    {
      "epoch": 0.3244582296880918,
      "grad_norm": 0.18319188058376312,
      "learning_rate": 2.0489260725886075e-05,
      "loss": 0.9222878217697144,
      "step": 3962
    },
    {
      "epoch": 0.32454012222461076,
      "grad_norm": 0.16979198157787323,
      "learning_rate": 2.048749235571321e-05,
      "loss": 0.8676437139511108,
      "step": 3963
    },
    {
      "epoch": 0.32462201476112973,
      "grad_norm": 0.16637015342712402,
      "learning_rate": 2.0485723616638965e-05,
      "loss": 1.018117904663086,
      "step": 3964
    },
    {
      "epoch": 0.32470390729764864,
      "grad_norm": 0.1901896893978119,
      "learning_rate": 2.0483954508740223e-05,
      "loss": 1.0465023517608643,
      "step": 3965
    },
    {
      "epoch": 0.3247857998341676,
      "grad_norm": 0.19633741676807404,
      "learning_rate": 2.0482185032093873e-05,
      "loss": 0.6522077918052673,
      "step": 3966
    },
    {
      "epoch": 0.3248676923706866,
      "grad_norm": 0.2304004728794098,
      "learning_rate": 2.048041518677683e-05,
      "loss": 0.8279807567596436,
      "step": 3967
    },
    {
      "epoch": 0.32494958490720555,
      "grad_norm": 0.17468871176242828,
      "learning_rate": 2.0478644972866008e-05,
      "loss": 1.110076904296875,
      "step": 3968
    },
    {
      "epoch": 0.32503147744372446,
      "grad_norm": 0.1968289017677307,
      "learning_rate": 2.0476874390438362e-05,
      "loss": 0.671642541885376,
      "step": 3969
    },
    {
      "epoch": 0.3251133699802434,
      "grad_norm": 0.18486829102039337,
      "learning_rate": 2.047510343957084e-05,
      "loss": 0.7551819086074829,
      "step": 3970
    },
    {
      "epoch": 0.3251952625167624,
      "grad_norm": 0.26064807176589966,
      "learning_rate": 2.047333212034041e-05,
      "loss": 0.9243712425231934,
      "step": 3971
    },
    {
      "epoch": 0.3252771550532813,
      "grad_norm": 0.21018044650554657,
      "learning_rate": 2.047156043282407e-05,
      "loss": 0.6543718576431274,
      "step": 3972
    },
    {
      "epoch": 0.3253590475898003,
      "grad_norm": 0.18876391649246216,
      "learning_rate": 2.0469788377098826e-05,
      "loss": 0.82879638671875,
      "step": 3973
    },
    {
      "epoch": 0.32544094012631924,
      "grad_norm": 0.22995233535766602,
      "learning_rate": 2.0468015953241692e-05,
      "loss": 0.7684871554374695,
      "step": 3974
    },
    {
      "epoch": 0.3255228326628382,
      "grad_norm": 0.18712438642978668,
      "learning_rate": 2.046624316132971e-05,
      "loss": 0.9632517099380493,
      "step": 3975
    },
    {
      "epoch": 0.3256047251993571,
      "grad_norm": 0.18045209348201752,
      "learning_rate": 2.0464470001439927e-05,
      "loss": 0.7628248929977417,
      "step": 3976
    },
    {
      "epoch": 0.3256866177358761,
      "grad_norm": 0.18567217886447906,
      "learning_rate": 2.046269647364941e-05,
      "loss": 0.9612525105476379,
      "step": 3977
    },
    {
      "epoch": 0.32576851027239506,
      "grad_norm": 0.19326569139957428,
      "learning_rate": 2.046092257803526e-05,
      "loss": 0.9389368295669556,
      "step": 3978
    },
    {
      "epoch": 0.32585040280891403,
      "grad_norm": 0.17253535985946655,
      "learning_rate": 2.0459148314674563e-05,
      "loss": 0.7407574653625488,
      "step": 3979
    },
    {
      "epoch": 0.32593229534543294,
      "grad_norm": 0.18855491280555725,
      "learning_rate": 2.0457373683644437e-05,
      "loss": 1.1804454326629639,
      "step": 3980
    },
    {
      "epoch": 0.3260141878819519,
      "grad_norm": 0.2047351449728012,
      "learning_rate": 2.0455598685022018e-05,
      "loss": 0.9871070384979248,
      "step": 3981
    },
    {
      "epoch": 0.3260960804184709,
      "grad_norm": 0.15462957322597504,
      "learning_rate": 2.0453823318884452e-05,
      "loss": 0.6662690043449402,
      "step": 3982
    },
    {
      "epoch": 0.3261779729549898,
      "grad_norm": 0.14857029914855957,
      "learning_rate": 2.0452047585308913e-05,
      "loss": 0.3944564163684845,
      "step": 3983
    },
    {
      "epoch": 0.32625986549150876,
      "grad_norm": 0.20832253992557526,
      "learning_rate": 2.045027148437257e-05,
      "loss": 0.8589080572128296,
      "step": 3984
    },
    {
      "epoch": 0.32634175802802773,
      "grad_norm": 0.17122115194797516,
      "learning_rate": 2.044849501615262e-05,
      "loss": 0.9156578183174133,
      "step": 3985
    },
    {
      "epoch": 0.3264236505645467,
      "grad_norm": 0.21610112488269806,
      "learning_rate": 2.0446718180726288e-05,
      "loss": 0.9434451460838318,
      "step": 3986
    },
    {
      "epoch": 0.3265055431010656,
      "grad_norm": 0.22846412658691406,
      "learning_rate": 2.044494097817079e-05,
      "loss": 0.9660305380821228,
      "step": 3987
    },
    {
      "epoch": 0.3265874356375846,
      "grad_norm": 0.19854377210140228,
      "learning_rate": 2.0443163408563374e-05,
      "loss": 0.7821274995803833,
      "step": 3988
    },
    {
      "epoch": 0.32666932817410355,
      "grad_norm": 0.20556685328483582,
      "learning_rate": 2.0441385471981303e-05,
      "loss": 0.8567473888397217,
      "step": 3989
    },
    {
      "epoch": 0.32675122071062246,
      "grad_norm": 0.18805284798145294,
      "learning_rate": 2.0439607168501852e-05,
      "loss": 0.7281758785247803,
      "step": 3990
    },
    {
      "epoch": 0.3268331132471414,
      "grad_norm": 0.19593830406665802,
      "learning_rate": 2.043782849820231e-05,
      "loss": 0.6428591012954712,
      "step": 3991
    },
    {
      "epoch": 0.3269150057836604,
      "grad_norm": 0.22520758211612701,
      "learning_rate": 2.0436049461159993e-05,
      "loss": 0.7370777726173401,
      "step": 3992
    },
    {
      "epoch": 0.32699689832017936,
      "grad_norm": 0.17537952959537506,
      "learning_rate": 2.0434270057452222e-05,
      "loss": 0.5172303318977356,
      "step": 3993
    },
    {
      "epoch": 0.3270787908566983,
      "grad_norm": 0.19850336015224457,
      "learning_rate": 2.043249028715633e-05,
      "loss": 1.1505286693572998,
      "step": 3994
    },
    {
      "epoch": 0.32716068339321724,
      "grad_norm": 0.22050577402114868,
      "learning_rate": 2.0430710150349685e-05,
      "loss": 0.7051447033882141,
      "step": 3995
    },
    {
      "epoch": 0.3272425759297362,
      "grad_norm": 0.2753869593143463,
      "learning_rate": 2.042892964710965e-05,
      "loss": 0.7393182516098022,
      "step": 3996
    },
    {
      "epoch": 0.3273244684662552,
      "grad_norm": 0.22312431037425995,
      "learning_rate": 2.0427148777513617e-05,
      "loss": 1.0518414974212646,
      "step": 3997
    },
    {
      "epoch": 0.3274063610027741,
      "grad_norm": 0.19419540464878082,
      "learning_rate": 2.0425367541638988e-05,
      "loss": 0.9165040850639343,
      "step": 3998
    },
    {
      "epoch": 0.32748825353929306,
      "grad_norm": 0.20299820601940155,
      "learning_rate": 2.0423585939563187e-05,
      "loss": 0.8882592916488647,
      "step": 3999
    },
    {
      "epoch": 0.32757014607581203,
      "grad_norm": 0.16102394461631775,
      "learning_rate": 2.0421803971363645e-05,
      "loss": 1.0121883153915405,
      "step": 4000
    },
    {
      "epoch": 0.32765203861233094,
      "grad_norm": 0.17446939647197723,
      "learning_rate": 2.0420021637117815e-05,
      "loss": 0.6966046690940857,
      "step": 4001
    },
    {
      "epoch": 0.3277339311488499,
      "grad_norm": 0.21143308281898499,
      "learning_rate": 2.0418238936903168e-05,
      "loss": 1.127868413925171,
      "step": 4002
    },
    {
      "epoch": 0.3278158236853689,
      "grad_norm": 0.22681520879268646,
      "learning_rate": 2.0416455870797182e-05,
      "loss": 0.73637855052948,
      "step": 4003
    },
    {
      "epoch": 0.32789771622188785,
      "grad_norm": 0.189267098903656,
      "learning_rate": 2.041467243887736e-05,
      "loss": 0.8378512263298035,
      "step": 4004
    },
    {
      "epoch": 0.32797960875840676,
      "grad_norm": 0.16721610724925995,
      "learning_rate": 2.0412888641221217e-05,
      "loss": 1.018246054649353,
      "step": 4005
    },
    {
      "epoch": 0.32806150129492573,
      "grad_norm": 0.19538401067256927,
      "learning_rate": 2.041110447790628e-05,
      "loss": 1.0527546405792236,
      "step": 4006
    },
    {
      "epoch": 0.3281433938314447,
      "grad_norm": 0.18923735618591309,
      "learning_rate": 2.0409319949010103e-05,
      "loss": 0.7907548546791077,
      "step": 4007
    },
    {
      "epoch": 0.32822528636796366,
      "grad_norm": 0.1968170702457428,
      "learning_rate": 2.0407535054610243e-05,
      "loss": 1.136070728302002,
      "step": 4008
    },
    {
      "epoch": 0.3283071789044826,
      "grad_norm": 0.186052143573761,
      "learning_rate": 2.0405749794784287e-05,
      "loss": 0.8291460275650024,
      "step": 4009
    },
    {
      "epoch": 0.32838907144100155,
      "grad_norm": 0.195217564702034,
      "learning_rate": 2.0403964169609818e-05,
      "loss": 0.8916608095169067,
      "step": 4010
    },
    {
      "epoch": 0.3284709639775205,
      "grad_norm": 0.22029122710227966,
      "learning_rate": 2.040217817916446e-05,
      "loss": 0.8192560076713562,
      "step": 4011
    },
    {
      "epoch": 0.3285528565140394,
      "grad_norm": 0.17189323902130127,
      "learning_rate": 2.0400391823525825e-05,
      "loss": 0.9450331330299377,
      "step": 4012
    },
    {
      "epoch": 0.3286347490505584,
      "grad_norm": 0.1893395483493805,
      "learning_rate": 2.0398605102771563e-05,
      "loss": 0.7952911853790283,
      "step": 4013
    },
    {
      "epoch": 0.32871664158707736,
      "grad_norm": 0.18401770293712616,
      "learning_rate": 2.0396818016979335e-05,
      "loss": 0.881236732006073,
      "step": 4014
    },
    {
      "epoch": 0.32879853412359633,
      "grad_norm": 0.20696619153022766,
      "learning_rate": 2.039503056622681e-05,
      "loss": 1.0850868225097656,
      "step": 4015
    },
    {
      "epoch": 0.32888042666011524,
      "grad_norm": 0.18049083650112152,
      "learning_rate": 2.0393242750591685e-05,
      "loss": 0.9143643975257874,
      "step": 4016
    },
    {
      "epoch": 0.3289623191966342,
      "grad_norm": 0.19269223511219025,
      "learning_rate": 2.0391454570151657e-05,
      "loss": 0.9244289994239807,
      "step": 4017
    },
    {
      "epoch": 0.3290442117331532,
      "grad_norm": 0.2084731161594391,
      "learning_rate": 2.038966602498445e-05,
      "loss": 1.1961641311645508,
      "step": 4018
    },
    {
      "epoch": 0.32912610426967215,
      "grad_norm": 0.15834586322307587,
      "learning_rate": 2.0387877115167802e-05,
      "loss": 0.6730955839157104,
      "step": 4019
    },
    {
      "epoch": 0.32920799680619106,
      "grad_norm": 0.20531298220157623,
      "learning_rate": 2.038608784077947e-05,
      "loss": 0.70001220703125,
      "step": 4020
    },
    {
      "epoch": 0.32928988934271003,
      "grad_norm": 0.19275188446044922,
      "learning_rate": 2.0384298201897223e-05,
      "loss": 1.0028375387191772,
      "step": 4021
    },
    {
      "epoch": 0.329371781879229,
      "grad_norm": 0.20015354454517365,
      "learning_rate": 2.038250819859884e-05,
      "loss": 1.0834403038024902,
      "step": 4022
    },
    {
      "epoch": 0.3294536744157479,
      "grad_norm": 0.22932399809360504,
      "learning_rate": 2.0380717830962126e-05,
      "loss": 0.983782172203064,
      "step": 4023
    },
    {
      "epoch": 0.3295355669522669,
      "grad_norm": 0.16976185142993927,
      "learning_rate": 2.03789270990649e-05,
      "loss": 0.7869149446487427,
      "step": 4024
    },
    {
      "epoch": 0.32961745948878585,
      "grad_norm": 0.22935444116592407,
      "learning_rate": 2.037713600298499e-05,
      "loss": 0.7244135737419128,
      "step": 4025
    },
    {
      "epoch": 0.3296993520253048,
      "grad_norm": 0.2144642025232315,
      "learning_rate": 2.0375344542800246e-05,
      "loss": 1.1511712074279785,
      "step": 4026
    },
    {
      "epoch": 0.3297812445618237,
      "grad_norm": 0.19879969954490662,
      "learning_rate": 2.0373552718588535e-05,
      "loss": 1.047278881072998,
      "step": 4027
    },
    {
      "epoch": 0.3298631370983427,
      "grad_norm": 0.19602613151073456,
      "learning_rate": 2.0371760530427733e-05,
      "loss": 0.8940179347991943,
      "step": 4028
    },
    {
      "epoch": 0.32994502963486166,
      "grad_norm": 0.16749311983585358,
      "learning_rate": 2.036996797839574e-05,
      "loss": 0.3920304477214813,
      "step": 4029
    },
    {
      "epoch": 0.33002692217138063,
      "grad_norm": 0.2268643081188202,
      "learning_rate": 2.0368175062570464e-05,
      "loss": 0.7390793561935425,
      "step": 4030
    },
    {
      "epoch": 0.33010881470789954,
      "grad_norm": 0.2297021448612213,
      "learning_rate": 2.036638178302983e-05,
      "loss": 1.1392782926559448,
      "step": 4031
    },
    {
      "epoch": 0.3301907072444185,
      "grad_norm": 0.22649626433849335,
      "learning_rate": 2.0364588139851786e-05,
      "loss": 0.8550149202346802,
      "step": 4032
    },
    {
      "epoch": 0.3302725997809375,
      "grad_norm": 0.23461101949214935,
      "learning_rate": 2.0362794133114295e-05,
      "loss": 1.0357513427734375,
      "step": 4033
    },
    {
      "epoch": 0.3303544923174564,
      "grad_norm": 0.22288455069065094,
      "learning_rate": 2.0360999762895326e-05,
      "loss": 0.748895525932312,
      "step": 4034
    },
    {
      "epoch": 0.33043638485397536,
      "grad_norm": 0.17852593958377838,
      "learning_rate": 2.0359205029272866e-05,
      "loss": 0.6420503258705139,
      "step": 4035
    },
    {
      "epoch": 0.33051827739049433,
      "grad_norm": 0.18965943157672882,
      "learning_rate": 2.035740993232493e-05,
      "loss": 0.8011335730552673,
      "step": 4036
    },
    {
      "epoch": 0.3306001699270133,
      "grad_norm": 0.19641093909740448,
      "learning_rate": 2.0355614472129538e-05,
      "loss": 0.9267378449440002,
      "step": 4037
    },
    {
      "epoch": 0.3306820624635322,
      "grad_norm": 0.21265588700771332,
      "learning_rate": 2.0353818648764724e-05,
      "loss": 0.7502137422561646,
      "step": 4038
    },
    {
      "epoch": 0.3307639550000512,
      "grad_norm": 0.18382368981838226,
      "learning_rate": 2.0352022462308544e-05,
      "loss": 0.6006070375442505,
      "step": 4039
    },
    {
      "epoch": 0.33084584753657015,
      "grad_norm": 0.170362651348114,
      "learning_rate": 2.0350225912839075e-05,
      "loss": 0.8745612502098083,
      "step": 4040
    },
    {
      "epoch": 0.3309277400730891,
      "grad_norm": 0.1651085466146469,
      "learning_rate": 2.034842900043439e-05,
      "loss": 0.7015158534049988,
      "step": 4041
    },
    {
      "epoch": 0.33100963260960803,
      "grad_norm": 0.19304487109184265,
      "learning_rate": 2.0346631725172595e-05,
      "loss": 0.7266128659248352,
      "step": 4042
    },
    {
      "epoch": 0.331091525146127,
      "grad_norm": 0.1942589432001114,
      "learning_rate": 2.034483408713181e-05,
      "loss": 0.830193042755127,
      "step": 4043
    },
    {
      "epoch": 0.33117341768264597,
      "grad_norm": 0.23720845580101013,
      "learning_rate": 2.0343036086390168e-05,
      "loss": 1.2950538396835327,
      "step": 4044
    },
    {
      "epoch": 0.3312553102191649,
      "grad_norm": 0.20238645374774933,
      "learning_rate": 2.034123772302581e-05,
      "loss": 0.7765851616859436,
      "step": 4045
    },
    {
      "epoch": 0.33133720275568385,
      "grad_norm": 0.17884604632854462,
      "learning_rate": 2.033943899711691e-05,
      "loss": 0.44498249888420105,
      "step": 4046
    },
    {
      "epoch": 0.3314190952922028,
      "grad_norm": 0.18991456925868988,
      "learning_rate": 2.0337639908741642e-05,
      "loss": 0.8519623875617981,
      "step": 4047
    },
    {
      "epoch": 0.3315009878287218,
      "grad_norm": 0.19754613935947418,
      "learning_rate": 2.0335840457978204e-05,
      "loss": 0.9353293180465698,
      "step": 4048
    },
    {
      "epoch": 0.3315828803652407,
      "grad_norm": 0.18952597677707672,
      "learning_rate": 2.0334040644904802e-05,
      "loss": 1.0402045249938965,
      "step": 4049
    },
    {
      "epoch": 0.33166477290175966,
      "grad_norm": 0.23784281313419342,
      "learning_rate": 2.0332240469599678e-05,
      "loss": 0.7042875289916992,
      "step": 4050
    },
    {
      "epoch": 0.33174666543827863,
      "grad_norm": 0.19683462381362915,
      "learning_rate": 2.0330439932141057e-05,
      "loss": 0.8289764523506165,
      "step": 4051
    },
    {
      "epoch": 0.33182855797479754,
      "grad_norm": 0.18794338405132294,
      "learning_rate": 2.0328639032607208e-05,
      "loss": 0.8717424273490906,
      "step": 4052
    },
    {
      "epoch": 0.3319104505113165,
      "grad_norm": 0.4129798710346222,
      "learning_rate": 2.0326837771076407e-05,
      "loss": 0.9264035224914551,
      "step": 4053
    },
    {
      "epoch": 0.3319923430478355,
      "grad_norm": 0.17741082608699799,
      "learning_rate": 2.0325036147626938e-05,
      "loss": 0.8676615953445435,
      "step": 4054
    },
    {
      "epoch": 0.33207423558435445,
      "grad_norm": 0.17749322950839996,
      "learning_rate": 2.032323416233711e-05,
      "loss": 0.8169325590133667,
      "step": 4055
    },
    {
      "epoch": 0.33215612812087336,
      "grad_norm": 0.17200404405593872,
      "learning_rate": 2.0321431815285244e-05,
      "loss": 0.8413490056991577,
      "step": 4056
    },
    {
      "epoch": 0.33223802065739233,
      "grad_norm": 0.2325853407382965,
      "learning_rate": 2.0319629106549674e-05,
      "loss": 0.6615083813667297,
      "step": 4057
    },
    {
      "epoch": 0.3323199131939113,
      "grad_norm": 0.20090986788272858,
      "learning_rate": 2.0317826036208766e-05,
      "loss": 0.7788219451904297,
      "step": 4058
    },
    {
      "epoch": 0.33240180573043027,
      "grad_norm": 0.20628434419631958,
      "learning_rate": 2.031602260434087e-05,
      "loss": 0.8360165357589722,
      "step": 4059
    },
    {
      "epoch": 0.3324836982669492,
      "grad_norm": 0.1663960963487625,
      "learning_rate": 2.0314218811024386e-05,
      "loss": 0.861330509185791,
      "step": 4060
    },
    {
      "epoch": 0.33256559080346815,
      "grad_norm": 0.215473011136055,
      "learning_rate": 2.031241465633771e-05,
      "loss": 0.8846909999847412,
      "step": 4061
    },
    {
      "epoch": 0.3326474833399871,
      "grad_norm": 0.19832509756088257,
      "learning_rate": 2.0310610140359254e-05,
      "loss": 0.8285926580429077,
      "step": 4062
    },
    {
      "epoch": 0.33272937587650603,
      "grad_norm": 0.19998326897621155,
      "learning_rate": 2.030880526316745e-05,
      "loss": 0.7034660577774048,
      "step": 4063
    },
    {
      "epoch": 0.332811268413025,
      "grad_norm": 0.19860784709453583,
      "learning_rate": 2.0307000024840753e-05,
      "loss": 0.8173413872718811,
      "step": 4064
    },
    {
      "epoch": 0.33289316094954396,
      "grad_norm": 0.1582118570804596,
      "learning_rate": 2.0305194425457616e-05,
      "loss": 0.9139061570167542,
      "step": 4065
    },
    {
      "epoch": 0.33297505348606293,
      "grad_norm": 0.19425052404403687,
      "learning_rate": 2.0303388465096528e-05,
      "loss": 0.8305050134658813,
      "step": 4066
    },
    {
      "epoch": 0.33305694602258185,
      "grad_norm": 0.14408493041992188,
      "learning_rate": 2.030158214383598e-05,
      "loss": 0.5420559644699097,
      "step": 4067
    },
    {
      "epoch": 0.3331388385591008,
      "grad_norm": 0.19711323082447052,
      "learning_rate": 2.029977546175447e-05,
      "loss": 0.9916682839393616,
      "step": 4068
    },
    {
      "epoch": 0.3332207310956198,
      "grad_norm": 0.19299691915512085,
      "learning_rate": 2.029796841893054e-05,
      "loss": 0.8007975816726685,
      "step": 4069
    },
    {
      "epoch": 0.33330262363213875,
      "grad_norm": 0.20652566850185394,
      "learning_rate": 2.0296161015442726e-05,
      "loss": 0.8380779027938843,
      "step": 4070
    },
    {
      "epoch": 0.33338451616865766,
      "grad_norm": 0.18891799449920654,
      "learning_rate": 2.0294353251369585e-05,
      "loss": 0.5312878489494324,
      "step": 4071
    },
    {
      "epoch": 0.33346640870517663,
      "grad_norm": 0.20972566306591034,
      "learning_rate": 2.029254512678969e-05,
      "loss": 0.9175806045532227,
      "step": 4072
    },
    {
      "epoch": 0.3335483012416956,
      "grad_norm": 0.2145683616399765,
      "learning_rate": 2.0290736641781628e-05,
      "loss": 0.9316257238388062,
      "step": 4073
    },
    {
      "epoch": 0.3336301937782145,
      "grad_norm": 0.1590670794248581,
      "learning_rate": 2.0288927796424008e-05,
      "loss": 0.8267413973808289,
      "step": 4074
    },
    {
      "epoch": 0.3337120863147335,
      "grad_norm": 0.1610410511493683,
      "learning_rate": 2.0287118590795442e-05,
      "loss": 0.8455234169960022,
      "step": 4075
    },
    {
      "epoch": 0.33379397885125245,
      "grad_norm": 0.16455750167369843,
      "learning_rate": 2.028530902497457e-05,
      "loss": 0.7639331817626953,
      "step": 4076
    },
    {
      "epoch": 0.3338758713877714,
      "grad_norm": 0.19437019526958466,
      "learning_rate": 2.0283499099040046e-05,
      "loss": 0.6027877926826477,
      "step": 4077
    },
    {
      "epoch": 0.33395776392429033,
      "grad_norm": 0.1702795922756195,
      "learning_rate": 2.028168881307053e-05,
      "loss": 0.713958740234375,
      "step": 4078
    },
    {
      "epoch": 0.3340396564608093,
      "grad_norm": 0.19411027431488037,
      "learning_rate": 2.0279878167144712e-05,
      "loss": 0.7649436593055725,
      "step": 4079
    },
    {
      "epoch": 0.33412154899732827,
      "grad_norm": 0.17687830328941345,
      "learning_rate": 2.0278067161341285e-05,
      "loss": 1.0156811475753784,
      "step": 4080
    },
    {
      "epoch": 0.33420344153384723,
      "grad_norm": 0.20212875306606293,
      "learning_rate": 2.0276255795738968e-05,
      "loss": 0.8082864284515381,
      "step": 4081
    },
    {
      "epoch": 0.33428533407036615,
      "grad_norm": 0.20079933106899261,
      "learning_rate": 2.027444407041648e-05,
      "loss": 0.7874143123626709,
      "step": 4082
    },
    {
      "epoch": 0.3343672266068851,
      "grad_norm": 0.17478986084461212,
      "learning_rate": 2.0272631985452576e-05,
      "loss": 0.7269238829612732,
      "step": 4083
    },
    {
      "epoch": 0.3344491191434041,
      "grad_norm": 0.210065096616745,
      "learning_rate": 2.0270819540926015e-05,
      "loss": 0.6287314891815186,
      "step": 4084
    },
    {
      "epoch": 0.334531011679923,
      "grad_norm": 0.17534174025058746,
      "learning_rate": 2.0269006736915572e-05,
      "loss": 0.9051300287246704,
      "step": 4085
    },
    {
      "epoch": 0.33461290421644196,
      "grad_norm": 0.21370923519134521,
      "learning_rate": 2.0267193573500034e-05,
      "loss": 0.7044405937194824,
      "step": 4086
    },
    {
      "epoch": 0.33469479675296093,
      "grad_norm": 0.3106364607810974,
      "learning_rate": 2.026538005075822e-05,
      "loss": 0.7187976241111755,
      "step": 4087
    },
    {
      "epoch": 0.3347766892894799,
      "grad_norm": 0.1656503528356552,
      "learning_rate": 2.026356616876894e-05,
      "loss": 0.9378139972686768,
      "step": 4088
    },
    {
      "epoch": 0.3348585818259988,
      "grad_norm": 0.17929284274578094,
      "learning_rate": 2.026175192761104e-05,
      "loss": 1.0935980081558228,
      "step": 4089
    },
    {
      "epoch": 0.3349404743625178,
      "grad_norm": 0.16197103261947632,
      "learning_rate": 2.025993732736338e-05,
      "loss": 0.5480328798294067,
      "step": 4090
    },
    {
      "epoch": 0.33502236689903675,
      "grad_norm": 0.2683809697628021,
      "learning_rate": 2.0258122368104817e-05,
      "loss": 0.5010266900062561,
      "step": 4091
    },
    {
      "epoch": 0.3351042594355557,
      "grad_norm": 0.19227901101112366,
      "learning_rate": 2.0256307049914246e-05,
      "loss": 0.9019798040390015,
      "step": 4092
    },
    {
      "epoch": 0.33518615197207463,
      "grad_norm": 0.20224416255950928,
      "learning_rate": 2.0254491372870564e-05,
      "loss": 1.3310216665267944,
      "step": 4093
    },
    {
      "epoch": 0.3352680445085936,
      "grad_norm": 0.1752881556749344,
      "learning_rate": 2.0252675337052687e-05,
      "loss": 0.6946004629135132,
      "step": 4094
    },
    {
      "epoch": 0.33534993704511257,
      "grad_norm": 0.23102502524852753,
      "learning_rate": 2.0250858942539552e-05,
      "loss": 0.947023868560791,
      "step": 4095
    },
    {
      "epoch": 0.3354318295816315,
      "grad_norm": 0.19129011034965515,
      "learning_rate": 2.024904218941011e-05,
      "loss": 0.751158595085144,
      "step": 4096
    },
    {
      "epoch": 0.33551372211815045,
      "grad_norm": 0.2030688226222992,
      "learning_rate": 2.0247225077743313e-05,
      "loss": 0.8773525357246399,
      "step": 4097
    },
    {
      "epoch": 0.3355956146546694,
      "grad_norm": 0.1738300919532776,
      "learning_rate": 2.0245407607618146e-05,
      "loss": 0.8313312530517578,
      "step": 4098
    },
    {
      "epoch": 0.3356775071911884,
      "grad_norm": 0.1841152310371399,
      "learning_rate": 2.0243589779113604e-05,
      "loss": 0.7006422877311707,
      "step": 4099
    },
    {
      "epoch": 0.3357593997277073,
      "grad_norm": 0.19686107337474823,
      "learning_rate": 2.02417715923087e-05,
      "loss": 0.6052550077438354,
      "step": 4100
    },
    {
      "epoch": 0.33584129226422627,
      "grad_norm": 0.17831183969974518,
      "learning_rate": 2.0239953047282457e-05,
      "loss": 0.6575282216072083,
      "step": 4101
    },
    {
      "epoch": 0.33592318480074523,
      "grad_norm": 0.17297640442848206,
      "learning_rate": 2.0238134144113916e-05,
      "loss": 0.8445392847061157,
      "step": 4102
    },
    {
      "epoch": 0.3360050773372642,
      "grad_norm": 0.17780667543411255,
      "learning_rate": 2.0236314882882132e-05,
      "loss": 0.6708532571792603,
      "step": 4103
    },
    {
      "epoch": 0.3360869698737831,
      "grad_norm": 0.1817796379327774,
      "learning_rate": 2.0234495263666185e-05,
      "loss": 0.9366111159324646,
      "step": 4104
    },
    {
      "epoch": 0.3361688624103021,
      "grad_norm": 0.2038915604352951,
      "learning_rate": 2.0232675286545156e-05,
      "loss": 1.074502944946289,
      "step": 4105
    },
    {
      "epoch": 0.33625075494682105,
      "grad_norm": 0.21009889245033264,
      "learning_rate": 2.023085495159815e-05,
      "loss": 0.8993433713912964,
      "step": 4106
    },
    {
      "epoch": 0.33633264748333996,
      "grad_norm": 0.19389379024505615,
      "learning_rate": 2.0229034258904293e-05,
      "loss": 1.0217398405075073,
      "step": 4107
    },
    {
      "epoch": 0.33641454001985893,
      "grad_norm": 0.17575329542160034,
      "learning_rate": 2.0227213208542704e-05,
      "loss": 1.0826812982559204,
      "step": 4108
    },
    {
      "epoch": 0.3364964325563779,
      "grad_norm": 0.19492943584918976,
      "learning_rate": 2.0225391800592553e-05,
      "loss": 0.7999382019042969,
      "step": 4109
    },
    {
      "epoch": 0.33657832509289687,
      "grad_norm": 0.18974758684635162,
      "learning_rate": 2.022357003513299e-05,
      "loss": 0.6127067804336548,
      "step": 4110
    },
    {
      "epoch": 0.3366602176294158,
      "grad_norm": 0.17256681621074677,
      "learning_rate": 2.0221747912243207e-05,
      "loss": 0.6307836174964905,
      "step": 4111
    },
    {
      "epoch": 0.33674211016593475,
      "grad_norm": 0.2065284252166748,
      "learning_rate": 2.0219925432002396e-05,
      "loss": 1.0226571559906006,
      "step": 4112
    },
    {
      "epoch": 0.3368240027024537,
      "grad_norm": 0.16375331580638885,
      "learning_rate": 2.0218102594489773e-05,
      "loss": 0.6260685920715332,
      "step": 4113
    },
    {
      "epoch": 0.3369058952389727,
      "grad_norm": 0.16487957537174225,
      "learning_rate": 2.0216279399784556e-05,
      "loss": 0.9201445579528809,
      "step": 4114
    },
    {
      "epoch": 0.3369877877754916,
      "grad_norm": 0.18548017740249634,
      "learning_rate": 2.0214455847966003e-05,
      "loss": 0.8532046675682068,
      "step": 4115
    },
    {
      "epoch": 0.33706968031201057,
      "grad_norm": 0.2213117927312851,
      "learning_rate": 2.021263193911336e-05,
      "loss": 1.012744426727295,
      "step": 4116
    },
    {
      "epoch": 0.33715157284852953,
      "grad_norm": 0.2301885336637497,
      "learning_rate": 2.021080767330591e-05,
      "loss": 0.7633083462715149,
      "step": 4117
    },
    {
      "epoch": 0.33723346538504845,
      "grad_norm": 0.22310608625411987,
      "learning_rate": 2.0208983050622944e-05,
      "loss": 0.7867302894592285,
      "step": 4118
    },
    {
      "epoch": 0.3373153579215674,
      "grad_norm": 0.1906222403049469,
      "learning_rate": 2.0207158071143758e-05,
      "loss": 0.7465484142303467,
      "step": 4119
    },
    {
      "epoch": 0.3373972504580864,
      "grad_norm": 0.22309747338294983,
      "learning_rate": 2.0205332734947685e-05,
      "loss": 0.7260332703590393,
      "step": 4120
    },
    {
      "epoch": 0.33747914299460535,
      "grad_norm": 0.20883961021900177,
      "learning_rate": 2.0203507042114058e-05,
      "loss": 1.0104247331619263,
      "step": 4121
    },
    {
      "epoch": 0.33756103553112426,
      "grad_norm": 0.180904358625412,
      "learning_rate": 2.020168099272222e-05,
      "loss": 0.97743821144104,
      "step": 4122
    },
    {
      "epoch": 0.33764292806764323,
      "grad_norm": 0.19980546832084656,
      "learning_rate": 2.019985458685155e-05,
      "loss": 1.0802273750305176,
      "step": 4123
    },
    {
      "epoch": 0.3377248206041622,
      "grad_norm": 0.1654147505760193,
      "learning_rate": 2.0198027824581427e-05,
      "loss": 0.8144475221633911,
      "step": 4124
    },
    {
      "epoch": 0.3378067131406811,
      "grad_norm": 0.21910542249679565,
      "learning_rate": 2.0196200705991246e-05,
      "loss": 0.8999002575874329,
      "step": 4125
    },
    {
      "epoch": 0.3378886056772001,
      "grad_norm": 0.19500267505645752,
      "learning_rate": 2.0194373231160428e-05,
      "loss": 0.8315403461456299,
      "step": 4126
    },
    {
      "epoch": 0.33797049821371905,
      "grad_norm": 0.19605717062950134,
      "learning_rate": 2.0192545400168397e-05,
      "loss": 0.9278091192245483,
      "step": 4127
    },
    {
      "epoch": 0.338052390750238,
      "grad_norm": 0.17522062361240387,
      "learning_rate": 2.01907172130946e-05,
      "loss": 0.6893069744110107,
      "step": 4128
    },
    {
      "epoch": 0.33813428328675693,
      "grad_norm": 0.20571212470531464,
      "learning_rate": 2.0188888670018504e-05,
      "loss": 0.7394000291824341,
      "step": 4129
    },
    {
      "epoch": 0.3382161758232759,
      "grad_norm": 0.20247095823287964,
      "learning_rate": 2.0187059771019577e-05,
      "loss": 1.0518132448196411,
      "step": 4130
    },
    {
      "epoch": 0.33829806835979487,
      "grad_norm": 0.1958884447813034,
      "learning_rate": 2.0185230516177307e-05,
      "loss": 1.017464280128479,
      "step": 4131
    },
    {
      "epoch": 0.33837996089631384,
      "grad_norm": 0.1891353875398636,
      "learning_rate": 2.018340090557121e-05,
      "loss": 0.8115677833557129,
      "step": 4132
    },
    {
      "epoch": 0.33846185343283275,
      "grad_norm": 0.2231300175189972,
      "learning_rate": 2.0181570939280803e-05,
      "loss": 0.970905065536499,
      "step": 4133
    },
    {
      "epoch": 0.3385437459693517,
      "grad_norm": 0.19984278082847595,
      "learning_rate": 2.0179740617385627e-05,
      "loss": 0.7906361818313599,
      "step": 4134
    },
    {
      "epoch": 0.3386256385058707,
      "grad_norm": 0.2086818665266037,
      "learning_rate": 2.0177909939965235e-05,
      "loss": 0.7698701620101929,
      "step": 4135
    },
    {
      "epoch": 0.3387075310423896,
      "grad_norm": 0.2006273865699768,
      "learning_rate": 2.0176078907099192e-05,
      "loss": 1.41153085231781,
      "step": 4136
    },
    {
      "epoch": 0.33878942357890857,
      "grad_norm": 0.173396497964859,
      "learning_rate": 2.0174247518867087e-05,
      "loss": 0.6647032499313354,
      "step": 4137
    },
    {
      "epoch": 0.33887131611542753,
      "grad_norm": 0.1706957072019577,
      "learning_rate": 2.0172415775348513e-05,
      "loss": 0.5756266117095947,
      "step": 4138
    },
    {
      "epoch": 0.3389532086519465,
      "grad_norm": 0.2908274233341217,
      "learning_rate": 2.017058367662309e-05,
      "loss": 0.7330725193023682,
      "step": 4139
    },
    {
      "epoch": 0.3390351011884654,
      "grad_norm": 0.23696134984493256,
      "learning_rate": 2.0168751222770456e-05,
      "loss": 1.2401033639907837,
      "step": 4140
    },
    {
      "epoch": 0.3391169937249844,
      "grad_norm": 0.18230324983596802,
      "learning_rate": 2.0166918413870243e-05,
      "loss": 0.9775770902633667,
      "step": 4141
    },
    {
      "epoch": 0.33919888626150335,
      "grad_norm": 0.18045051395893097,
      "learning_rate": 2.0165085250002123e-05,
      "loss": 0.8250633478164673,
      "step": 4142
    },
    {
      "epoch": 0.3392807787980223,
      "grad_norm": 0.21234093606472015,
      "learning_rate": 2.0163251731245762e-05,
      "loss": 1.0050899982452393,
      "step": 4143
    },
    {
      "epoch": 0.33936267133454123,
      "grad_norm": 0.21815265715122223,
      "learning_rate": 2.0161417857680866e-05,
      "loss": 0.9787946939468384,
      "step": 4144
    },
    {
      "epoch": 0.3394445638710602,
      "grad_norm": 0.19218075275421143,
      "learning_rate": 2.015958362938713e-05,
      "loss": 0.8710106015205383,
      "step": 4145
    },
    {
      "epoch": 0.33952645640757917,
      "grad_norm": 0.1919543296098709,
      "learning_rate": 2.0157749046444286e-05,
      "loss": 0.8262546062469482,
      "step": 4146
    },
    {
      "epoch": 0.3396083489440981,
      "grad_norm": 0.20037543773651123,
      "learning_rate": 2.015591410893207e-05,
      "loss": 0.8349179625511169,
      "step": 4147
    },
    {
      "epoch": 0.33969024148061705,
      "grad_norm": 0.19892346858978271,
      "learning_rate": 2.015407881693023e-05,
      "loss": 0.5769320726394653,
      "step": 4148
    },
    {
      "epoch": 0.339772134017136,
      "grad_norm": 0.1886955201625824,
      "learning_rate": 2.0152243170518545e-05,
      "loss": 0.5961681008338928,
      "step": 4149
    },
    {
      "epoch": 0.339854026553655,
      "grad_norm": 0.21364907920360565,
      "learning_rate": 2.0150407169776792e-05,
      "loss": 0.9519136548042297,
      "step": 4150
    },
    {
      "epoch": 0.3399359190901739,
      "grad_norm": 0.21014325320720673,
      "learning_rate": 2.0148570814784777e-05,
      "loss": 0.8765990138053894,
      "step": 4151
    },
    {
      "epoch": 0.34001781162669287,
      "grad_norm": 0.1934913992881775,
      "learning_rate": 2.014673410562231e-05,
      "loss": 0.7026876211166382,
      "step": 4152
    },
    {
      "epoch": 0.34009970416321184,
      "grad_norm": 0.20816461741924286,
      "learning_rate": 2.0144897042369228e-05,
      "loss": 0.6527125239372253,
      "step": 4153
    },
    {
      "epoch": 0.3401815966997308,
      "grad_norm": 0.21145157516002655,
      "learning_rate": 2.014305962510537e-05,
      "loss": 0.9180698990821838,
      "step": 4154
    },
    {
      "epoch": 0.3402634892362497,
      "grad_norm": 0.21570849418640137,
      "learning_rate": 2.0141221853910608e-05,
      "loss": 1.093538761138916,
      "step": 4155
    },
    {
      "epoch": 0.3403453817727687,
      "grad_norm": 0.16500669717788696,
      "learning_rate": 2.013938372886481e-05,
      "loss": 1.1335633993148804,
      "step": 4156
    },
    {
      "epoch": 0.34042727430928765,
      "grad_norm": 0.18976552784442902,
      "learning_rate": 2.013754525004787e-05,
      "loss": 0.7708661556243896,
      "step": 4157
    },
    {
      "epoch": 0.34050916684580657,
      "grad_norm": 0.18272224068641663,
      "learning_rate": 2.0135706417539698e-05,
      "loss": 0.6988851428031921,
      "step": 4158
    },
    {
      "epoch": 0.34059105938232553,
      "grad_norm": 0.1677340567111969,
      "learning_rate": 2.0133867231420216e-05,
      "loss": 1.115917682647705,
      "step": 4159
    },
    {
      "epoch": 0.3406729519188445,
      "grad_norm": 0.20438788831233978,
      "learning_rate": 2.0132027691769365e-05,
      "loss": 0.7363069653511047,
      "step": 4160
    },
    {
      "epoch": 0.34075484445536347,
      "grad_norm": 0.20224568247795105,
      "learning_rate": 2.0130187798667095e-05,
      "loss": 0.9210011959075928,
      "step": 4161
    },
    {
      "epoch": 0.3408367369918824,
      "grad_norm": 0.17178243398666382,
      "learning_rate": 2.012834755219338e-05,
      "loss": 0.6549001932144165,
      "step": 4162
    },
    {
      "epoch": 0.34091862952840135,
      "grad_norm": 0.16897203028202057,
      "learning_rate": 2.01265069524282e-05,
      "loss": 1.1297110319137573,
      "step": 4163
    },
    {
      "epoch": 0.3410005220649203,
      "grad_norm": 0.19492314755916595,
      "learning_rate": 2.012466599945156e-05,
      "loss": 0.8482123017311096,
      "step": 4164
    },
    {
      "epoch": 0.3410824146014393,
      "grad_norm": 0.16053996980190277,
      "learning_rate": 2.012282469334347e-05,
      "loss": 0.8623709678649902,
      "step": 4165
    },
    {
      "epoch": 0.3411643071379582,
      "grad_norm": 0.21516868472099304,
      "learning_rate": 2.0120983034183966e-05,
      "loss": 0.9922193884849548,
      "step": 4166
    },
    {
      "epoch": 0.34124619967447717,
      "grad_norm": 0.2561231553554535,
      "learning_rate": 2.0119141022053093e-05,
      "loss": 0.7635040283203125,
      "step": 4167
    },
    {
      "epoch": 0.34132809221099614,
      "grad_norm": 0.1531168520450592,
      "learning_rate": 2.01172986570309e-05,
      "loss": 0.8474525213241577,
      "step": 4168
    },
    {
      "epoch": 0.34140998474751505,
      "grad_norm": 0.18329504132270813,
      "learning_rate": 2.0115455939197485e-05,
      "loss": 0.9704961180686951,
      "step": 4169
    },
    {
      "epoch": 0.341491877284034,
      "grad_norm": 0.416801393032074,
      "learning_rate": 2.0113612868632927e-05,
      "loss": 0.8646448254585266,
      "step": 4170
    },
    {
      "epoch": 0.341573769820553,
      "grad_norm": 0.16965526342391968,
      "learning_rate": 2.011176944541734e-05,
      "loss": 0.6681275367736816,
      "step": 4171
    },
    {
      "epoch": 0.34165566235707195,
      "grad_norm": 0.20634815096855164,
      "learning_rate": 2.0109925669630835e-05,
      "loss": 1.004655361175537,
      "step": 4172
    },
    {
      "epoch": 0.34173755489359087,
      "grad_norm": 0.17234119772911072,
      "learning_rate": 2.010808154135356e-05,
      "loss": 0.9319829940795898,
      "step": 4173
    },
    {
      "epoch": 0.34181944743010984,
      "grad_norm": 0.18674084544181824,
      "learning_rate": 2.010623706066567e-05,
      "loss": 0.9745858907699585,
      "step": 4174
    },
    {
      "epoch": 0.3419013399666288,
      "grad_norm": 0.1892077475786209,
      "learning_rate": 2.010439222764733e-05,
      "loss": 0.9484982490539551,
      "step": 4175
    },
    {
      "epoch": 0.34198323250314777,
      "grad_norm": 0.16572999954223633,
      "learning_rate": 2.0102547042378723e-05,
      "loss": 0.8517343401908875,
      "step": 4176
    },
    {
      "epoch": 0.3420651250396667,
      "grad_norm": 0.1920069009065628,
      "learning_rate": 2.010070150494005e-05,
      "loss": 0.9989929795265198,
      "step": 4177
    },
    {
      "epoch": 0.34214701757618565,
      "grad_norm": 0.19334693253040314,
      "learning_rate": 2.0098855615411525e-05,
      "loss": 0.9607858061790466,
      "step": 4178
    },
    {
      "epoch": 0.3422289101127046,
      "grad_norm": 0.18254682421684265,
      "learning_rate": 2.009700937387338e-05,
      "loss": 0.8767298460006714,
      "step": 4179
    },
    {
      "epoch": 0.34231080264922353,
      "grad_norm": 0.2217516452074051,
      "learning_rate": 2.0095162780405857e-05,
      "loss": 0.6976207494735718,
      "step": 4180
    },
    {
      "epoch": 0.3423926951857425,
      "grad_norm": 0.19698381423950195,
      "learning_rate": 2.0093315835089217e-05,
      "loss": 0.6450433731079102,
      "step": 4181
    },
    {
      "epoch": 0.34247458772226147,
      "grad_norm": 0.2059532254934311,
      "learning_rate": 2.009146853800374e-05,
      "loss": 0.5354965329170227,
      "step": 4182
    },
    {
      "epoch": 0.34255648025878044,
      "grad_norm": 0.1903630495071411,
      "learning_rate": 2.008962088922971e-05,
      "loss": 1.2999749183654785,
      "step": 4183
    },
    {
      "epoch": 0.34263837279529935,
      "grad_norm": 0.18949788808822632,
      "learning_rate": 2.0087772888847438e-05,
      "loss": 1.024554967880249,
      "step": 4184
    },
    {
      "epoch": 0.3427202653318183,
      "grad_norm": 0.177216038107872,
      "learning_rate": 2.0085924536937247e-05,
      "loss": 1.057919979095459,
      "step": 4185
    },
    {
      "epoch": 0.3428021578683373,
      "grad_norm": 0.18121379613876343,
      "learning_rate": 2.0084075833579473e-05,
      "loss": 0.9753561019897461,
      "step": 4186
    },
    {
      "epoch": 0.3428840504048562,
      "grad_norm": 0.2093733847141266,
      "learning_rate": 2.0082226778854466e-05,
      "loss": 0.8342843055725098,
      "step": 4187
    },
    {
      "epoch": 0.34296594294137517,
      "grad_norm": 0.1915922313928604,
      "learning_rate": 2.008037737284259e-05,
      "loss": 1.049103021621704,
      "step": 4188
    },
    {
      "epoch": 0.34304783547789414,
      "grad_norm": 0.16693562269210815,
      "learning_rate": 2.007852761562424e-05,
      "loss": 0.9955962896347046,
      "step": 4189
    },
    {
      "epoch": 0.3431297280144131,
      "grad_norm": 0.240909606218338,
      "learning_rate": 2.0076677507279803e-05,
      "loss": 1.1978267431259155,
      "step": 4190
    },
    {
      "epoch": 0.343211620550932,
      "grad_norm": 0.16977940499782562,
      "learning_rate": 2.0074827047889698e-05,
      "loss": 0.7878450751304626,
      "step": 4191
    },
    {
      "epoch": 0.343293513087451,
      "grad_norm": 0.18611086905002594,
      "learning_rate": 2.0072976237534354e-05,
      "loss": 0.9752344489097595,
      "step": 4192
    },
    {
      "epoch": 0.34337540562396995,
      "grad_norm": 0.22668877243995667,
      "learning_rate": 2.0071125076294206e-05,
      "loss": 0.7601072192192078,
      "step": 4193
    },
    {
      "epoch": 0.3434572981604889,
      "grad_norm": 0.17141306400299072,
      "learning_rate": 2.0069273564249724e-05,
      "loss": 0.6256403923034668,
      "step": 4194
    },
    {
      "epoch": 0.34353919069700783,
      "grad_norm": 0.1692526787519455,
      "learning_rate": 2.0067421701481382e-05,
      "loss": 0.5932313799858093,
      "step": 4195
    },
    {
      "epoch": 0.3436210832335268,
      "grad_norm": 0.17114360630512238,
      "learning_rate": 2.006556948806966e-05,
      "loss": 0.8519042134284973,
      "step": 4196
    },
    {
      "epoch": 0.34370297577004577,
      "grad_norm": 0.16610580682754517,
      "learning_rate": 2.0063716924095066e-05,
      "loss": 0.6596360206604004,
      "step": 4197
    },
    {
      "epoch": 0.3437848683065647,
      "grad_norm": 0.2002793848514557,
      "learning_rate": 2.0061864009638127e-05,
      "loss": 1.0652709007263184,
      "step": 4198
    },
    {
      "epoch": 0.34386676084308365,
      "grad_norm": 0.1663132905960083,
      "learning_rate": 2.0060010744779372e-05,
      "loss": 0.9694660305976868,
      "step": 4199
    },
    {
      "epoch": 0.3439486533796026,
      "grad_norm": 0.19559408724308014,
      "learning_rate": 2.0058157129599357e-05,
      "loss": 0.7791293263435364,
      "step": 4200
    },
    {
      "epoch": 0.3440305459161216,
      "grad_norm": 0.17432673275470734,
      "learning_rate": 2.005630316417864e-05,
      "loss": 0.8515185117721558,
      "step": 4201
    },
    {
      "epoch": 0.3441124384526405,
      "grad_norm": 0.16162970662117004,
      "learning_rate": 2.0054448848597807e-05,
      "loss": 0.542104959487915,
      "step": 4202
    },
    {
      "epoch": 0.34419433098915947,
      "grad_norm": 0.1894133985042572,
      "learning_rate": 2.0052594182937455e-05,
      "loss": 1.1690123081207275,
      "step": 4203
    },
    {
      "epoch": 0.34427622352567844,
      "grad_norm": 0.2007523626089096,
      "learning_rate": 2.0050739167278193e-05,
      "loss": 0.8925947546958923,
      "step": 4204
    },
    {
      "epoch": 0.3443581160621974,
      "grad_norm": 0.21960397064685822,
      "learning_rate": 2.0048883801700646e-05,
      "loss": 0.8110586404800415,
      "step": 4205
    },
    {
      "epoch": 0.3444400085987163,
      "grad_norm": 0.22228048741817474,
      "learning_rate": 2.0047028086285457e-05,
      "loss": 0.9946169853210449,
      "step": 4206
    },
    {
      "epoch": 0.3445219011352353,
      "grad_norm": 0.1810937523841858,
      "learning_rate": 2.0045172021113285e-05,
      "loss": 0.7024261355400085,
      "step": 4207
    },
    {
      "epoch": 0.34460379367175425,
      "grad_norm": 0.42427799105644226,
      "learning_rate": 2.0043315606264807e-05,
      "loss": 0.8590747714042664,
      "step": 4208
    },
    {
      "epoch": 0.34468568620827317,
      "grad_norm": 0.20287349820137024,
      "learning_rate": 2.0041458841820696e-05,
      "loss": 0.9089409708976746,
      "step": 4209
    },
    {
      "epoch": 0.34476757874479214,
      "grad_norm": 0.17215101420879364,
      "learning_rate": 2.003960172786167e-05,
      "loss": 0.555267333984375,
      "step": 4210
    },
    {
      "epoch": 0.3448494712813111,
      "grad_norm": 0.18266358971595764,
      "learning_rate": 2.0037744264468436e-05,
      "loss": 0.9355218410491943,
      "step": 4211
    },
    {
      "epoch": 0.3449313638178301,
      "grad_norm": 0.20984376966953278,
      "learning_rate": 2.0035886451721736e-05,
      "loss": 0.8429971933364868,
      "step": 4212
    },
    {
      "epoch": 0.345013256354349,
      "grad_norm": 0.18971949815750122,
      "learning_rate": 2.003402828970231e-05,
      "loss": 0.6979826092720032,
      "step": 4213
    },
    {
      "epoch": 0.34509514889086795,
      "grad_norm": 0.1872001439332962,
      "learning_rate": 2.0032169778490926e-05,
      "loss": 0.7802934050559998,
      "step": 4214
    },
    {
      "epoch": 0.3451770414273869,
      "grad_norm": 0.16957956552505493,
      "learning_rate": 2.003031091816836e-05,
      "loss": 0.6597304940223694,
      "step": 4215
    },
    {
      "epoch": 0.3452589339639059,
      "grad_norm": 0.16619089245796204,
      "learning_rate": 2.0028451708815408e-05,
      "loss": 0.5896395444869995,
      "step": 4216
    },
    {
      "epoch": 0.3453408265004248,
      "grad_norm": 0.18357901275157928,
      "learning_rate": 2.0026592150512876e-05,
      "loss": 0.9305190443992615,
      "step": 4217
    },
    {
      "epoch": 0.34542271903694377,
      "grad_norm": 0.19903860986232758,
      "learning_rate": 2.0024732243341594e-05,
      "loss": 0.7996610403060913,
      "step": 4218
    },
    {
      "epoch": 0.34550461157346274,
      "grad_norm": 0.23519916832447052,
      "learning_rate": 2.002287198738239e-05,
      "loss": 0.9352813959121704,
      "step": 4219
    },
    {
      "epoch": 0.34558650410998165,
      "grad_norm": 0.19062447547912598,
      "learning_rate": 2.0021011382716134e-05,
      "loss": 0.7166861891746521,
      "step": 4220
    },
    {
      "epoch": 0.3456683966465006,
      "grad_norm": 0.19423682987689972,
      "learning_rate": 2.0019150429423688e-05,
      "loss": 0.8326184153556824,
      "step": 4221
    },
    {
      "epoch": 0.3457502891830196,
      "grad_norm": 0.1745479851961136,
      "learning_rate": 2.0017289127585926e-05,
      "loss": 0.7536429762840271,
      "step": 4222
    },
    {
      "epoch": 0.34583218171953856,
      "grad_norm": 0.18591739237308502,
      "learning_rate": 2.0015427477283766e-05,
      "loss": 0.8680458068847656,
      "step": 4223
    },
    {
      "epoch": 0.34591407425605747,
      "grad_norm": 0.1790480762720108,
      "learning_rate": 2.001356547859811e-05,
      "loss": 0.7651387453079224,
      "step": 4224
    },
    {
      "epoch": 0.34599596679257644,
      "grad_norm": 0.20816820859909058,
      "learning_rate": 2.0011703131609897e-05,
      "loss": 0.9094093441963196,
      "step": 4225
    },
    {
      "epoch": 0.3460778593290954,
      "grad_norm": 0.15505242347717285,
      "learning_rate": 2.0009840436400067e-05,
      "loss": 0.9693399667739868,
      "step": 4226
    },
    {
      "epoch": 0.3461597518656144,
      "grad_norm": 0.2405937910079956,
      "learning_rate": 2.000797739304958e-05,
      "loss": 0.9961382150650024,
      "step": 4227
    },
    {
      "epoch": 0.3462416444021333,
      "grad_norm": 0.18022936582565308,
      "learning_rate": 2.000611400163941e-05,
      "loss": 0.822860598564148,
      "step": 4228
    },
    {
      "epoch": 0.34632353693865225,
      "grad_norm": 0.20053011178970337,
      "learning_rate": 2.000425026225056e-05,
      "loss": 1.0446475744247437,
      "step": 4229
    },
    {
      "epoch": 0.3464054294751712,
      "grad_norm": 0.27562493085861206,
      "learning_rate": 2.000238617496402e-05,
      "loss": 1.021561622619629,
      "step": 4230
    },
    {
      "epoch": 0.34648732201169014,
      "grad_norm": 0.19245405495166779,
      "learning_rate": 2.0000521739860816e-05,
      "loss": 0.8204619288444519,
      "step": 4231
    },
    {
      "epoch": 0.3465692145482091,
      "grad_norm": 0.16932040452957153,
      "learning_rate": 1.999865695702199e-05,
      "loss": 0.558012068271637,
      "step": 4232
    },
    {
      "epoch": 0.34665110708472807,
      "grad_norm": 0.2020881325006485,
      "learning_rate": 1.9996791826528584e-05,
      "loss": 1.2564321756362915,
      "step": 4233
    },
    {
      "epoch": 0.34673299962124704,
      "grad_norm": 0.19408632814884186,
      "learning_rate": 1.9994926348461675e-05,
      "loss": 0.592451810836792,
      "step": 4234
    },
    {
      "epoch": 0.34681489215776595,
      "grad_norm": 0.18360961973667145,
      "learning_rate": 1.9993060522902333e-05,
      "loss": 1.1483325958251953,
      "step": 4235
    },
    {
      "epoch": 0.3468967846942849,
      "grad_norm": 0.17517441511154175,
      "learning_rate": 1.999119434993166e-05,
      "loss": 0.7502737045288086,
      "step": 4236
    },
    {
      "epoch": 0.3469786772308039,
      "grad_norm": 0.18871162831783295,
      "learning_rate": 1.998932782963077e-05,
      "loss": 0.7700850367546082,
      "step": 4237
    },
    {
      "epoch": 0.34706056976732286,
      "grad_norm": 0.1738772690296173,
      "learning_rate": 1.9987460962080786e-05,
      "loss": 0.9592525362968445,
      "step": 4238
    },
    {
      "epoch": 0.34714246230384177,
      "grad_norm": 0.18718184530735016,
      "learning_rate": 1.998559374736285e-05,
      "loss": 0.575971245765686,
      "step": 4239
    },
    {
      "epoch": 0.34722435484036074,
      "grad_norm": 0.1919238418340683,
      "learning_rate": 1.9983726185558116e-05,
      "loss": 1.109427809715271,
      "step": 4240
    },
    {
      "epoch": 0.3473062473768797,
      "grad_norm": 0.15160447359085083,
      "learning_rate": 1.9981858276747766e-05,
      "loss": 0.6453830599784851,
      "step": 4241
    },
    {
      "epoch": 0.3473881399133986,
      "grad_norm": 0.22129864990711212,
      "learning_rate": 1.9979990021012978e-05,
      "loss": 0.8160969018936157,
      "step": 4242
    },
    {
      "epoch": 0.3474700324499176,
      "grad_norm": 0.1703474521636963,
      "learning_rate": 1.9978121418434952e-05,
      "loss": 1.0769004821777344,
      "step": 4243
    },
    {
      "epoch": 0.34755192498643656,
      "grad_norm": 0.19220326840877533,
      "learning_rate": 1.9976252469094916e-05,
      "loss": 0.6733195185661316,
      "step": 4244
    },
    {
      "epoch": 0.3476338175229555,
      "grad_norm": 0.15140116214752197,
      "learning_rate": 1.9974383173074093e-05,
      "loss": 0.8357107043266296,
      "step": 4245
    },
    {
      "epoch": 0.34771571005947444,
      "grad_norm": 0.21509481966495514,
      "learning_rate": 1.9972513530453734e-05,
      "loss": 0.8916155695915222,
      "step": 4246
    },
    {
      "epoch": 0.3477976025959934,
      "grad_norm": 0.2284659892320633,
      "learning_rate": 1.99706435413151e-05,
      "loss": 1.1279882192611694,
      "step": 4247
    },
    {
      "epoch": 0.3478794951325124,
      "grad_norm": 0.19866876304149628,
      "learning_rate": 1.996877320573947e-05,
      "loss": 0.9605868458747864,
      "step": 4248
    },
    {
      "epoch": 0.34796138766903134,
      "grad_norm": 0.19072696566581726,
      "learning_rate": 1.9966902523808135e-05,
      "loss": 0.9076762795448303,
      "step": 4249
    },
    {
      "epoch": 0.34804328020555025,
      "grad_norm": 0.20904573798179626,
      "learning_rate": 1.99650314956024e-05,
      "loss": 1.008750081062317,
      "step": 4250
    },
    {
      "epoch": 0.3481251727420692,
      "grad_norm": 0.20247584581375122,
      "learning_rate": 1.9963160121203594e-05,
      "loss": 0.8050978779792786,
      "step": 4251
    },
    {
      "epoch": 0.3482070652785882,
      "grad_norm": 0.19660364091396332,
      "learning_rate": 1.9961288400693054e-05,
      "loss": 0.8156879544258118,
      "step": 4252
    },
    {
      "epoch": 0.3482889578151071,
      "grad_norm": 0.1963547319173813,
      "learning_rate": 1.9959416334152127e-05,
      "loss": 0.7328289747238159,
      "step": 4253
    },
    {
      "epoch": 0.34837085035162607,
      "grad_norm": 0.1964053213596344,
      "learning_rate": 1.9957543921662182e-05,
      "loss": 0.6124880909919739,
      "step": 4254
    },
    {
      "epoch": 0.34845274288814504,
      "grad_norm": 0.19500236213207245,
      "learning_rate": 1.9955671163304605e-05,
      "loss": 1.0373597145080566,
      "step": 4255
    },
    {
      "epoch": 0.348534635424664,
      "grad_norm": 0.1970420926809311,
      "learning_rate": 1.9953798059160792e-05,
      "loss": 0.666071355342865,
      "step": 4256
    },
    {
      "epoch": 0.3486165279611829,
      "grad_norm": 0.1643259972333908,
      "learning_rate": 1.9951924609312158e-05,
      "loss": 0.6159888505935669,
      "step": 4257
    },
    {
      "epoch": 0.3486984204977019,
      "grad_norm": 0.15608172118663788,
      "learning_rate": 1.9950050813840125e-05,
      "loss": 0.8564125299453735,
      "step": 4258
    },
    {
      "epoch": 0.34878031303422086,
      "grad_norm": 0.17953112721443176,
      "learning_rate": 1.994817667282614e-05,
      "loss": 0.8252280950546265,
      "step": 4259
    },
    {
      "epoch": 0.34886220557073977,
      "grad_norm": 0.17098568379878998,
      "learning_rate": 1.9946302186351668e-05,
      "loss": 0.72845059633255,
      "step": 4260
    },
    {
      "epoch": 0.34894409810725874,
      "grad_norm": 0.1770675927400589,
      "learning_rate": 1.9944427354498168e-05,
      "loss": 0.7689481973648071,
      "step": 4261
    },
    {
      "epoch": 0.3490259906437777,
      "grad_norm": 0.1830819845199585,
      "learning_rate": 1.9942552177347134e-05,
      "loss": 0.869820237159729,
      "step": 4262
    },
    {
      "epoch": 0.3491078831802967,
      "grad_norm": 0.16418322920799255,
      "learning_rate": 1.9940676654980073e-05,
      "loss": 0.7928192019462585,
      "step": 4263
    },
    {
      "epoch": 0.3491897757168156,
      "grad_norm": 0.20262765884399414,
      "learning_rate": 1.9938800787478497e-05,
      "loss": 1.225271224975586,
      "step": 4264
    },
    {
      "epoch": 0.34927166825333456,
      "grad_norm": 0.2257627695798874,
      "learning_rate": 1.993692457492394e-05,
      "loss": 1.019595742225647,
      "step": 4265
    },
    {
      "epoch": 0.3493535607898535,
      "grad_norm": 0.18162813782691956,
      "learning_rate": 1.9935048017397954e-05,
      "loss": 0.9958438277244568,
      "step": 4266
    },
    {
      "epoch": 0.3494354533263725,
      "grad_norm": 0.16226407885551453,
      "learning_rate": 1.9933171114982095e-05,
      "loss": 0.7579657435417175,
      "step": 4267
    },
    {
      "epoch": 0.3495173458628914,
      "grad_norm": 0.21937301754951477,
      "learning_rate": 1.993129386775795e-05,
      "loss": 0.7065722346305847,
      "step": 4268
    },
    {
      "epoch": 0.3495992383994104,
      "grad_norm": 0.20830292999744415,
      "learning_rate": 1.9929416275807104e-05,
      "loss": 0.7484489679336548,
      "step": 4269
    },
    {
      "epoch": 0.34968113093592934,
      "grad_norm": 0.19231916964054108,
      "learning_rate": 1.9927538339211165e-05,
      "loss": 1.1127445697784424,
      "step": 4270
    },
    {
      "epoch": 0.34976302347244825,
      "grad_norm": 0.21205741167068481,
      "learning_rate": 1.9925660058051764e-05,
      "loss": 1.2230138778686523,
      "step": 4271
    },
    {
      "epoch": 0.3498449160089672,
      "grad_norm": 0.13818219304084778,
      "learning_rate": 1.9923781432410535e-05,
      "loss": 0.5718998908996582,
      "step": 4272
    },
    {
      "epoch": 0.3499268085454862,
      "grad_norm": 0.1705896109342575,
      "learning_rate": 1.9921902462369125e-05,
      "loss": 0.5857288837432861,
      "step": 4273
    },
    {
      "epoch": 0.35000870108200516,
      "grad_norm": 0.16927361488342285,
      "learning_rate": 1.9920023148009208e-05,
      "loss": 1.0685962438583374,
      "step": 4274
    },
    {
      "epoch": 0.35009059361852407,
      "grad_norm": 0.19389848411083221,
      "learning_rate": 1.9918143489412465e-05,
      "loss": 0.7640177607536316,
      "step": 4275
    },
    {
      "epoch": 0.35017248615504304,
      "grad_norm": 0.18173375725746155,
      "learning_rate": 1.9916263486660588e-05,
      "loss": 0.6967271566390991,
      "step": 4276
    },
    {
      "epoch": 0.350254378691562,
      "grad_norm": 0.18806962668895721,
      "learning_rate": 1.99143831398353e-05,
      "loss": 0.903894305229187,
      "step": 4277
    },
    {
      "epoch": 0.350336271228081,
      "grad_norm": 0.19347622990608215,
      "learning_rate": 1.9912502449018327e-05,
      "loss": 0.9440847635269165,
      "step": 4278
    },
    {
      "epoch": 0.3504181637645999,
      "grad_norm": 0.1341981738805771,
      "learning_rate": 1.99106214142914e-05,
      "loss": 0.7661923766136169,
      "step": 4279
    },
    {
      "epoch": 0.35050005630111886,
      "grad_norm": 0.20395943522453308,
      "learning_rate": 1.990874003573629e-05,
      "loss": 0.9385321140289307,
      "step": 4280
    },
    {
      "epoch": 0.3505819488376378,
      "grad_norm": 0.1743140071630478,
      "learning_rate": 1.9906858313434763e-05,
      "loss": 0.8520224094390869,
      "step": 4281
    },
    {
      "epoch": 0.35066384137415674,
      "grad_norm": 0.1934700608253479,
      "learning_rate": 1.990497624746861e-05,
      "loss": 1.0494107007980347,
      "step": 4282
    },
    {
      "epoch": 0.3507457339106757,
      "grad_norm": 0.19144435226917267,
      "learning_rate": 1.9903093837919632e-05,
      "loss": 0.962820291519165,
      "step": 4283
    },
    {
      "epoch": 0.3508276264471947,
      "grad_norm": 0.15662795305252075,
      "learning_rate": 1.9901211084869644e-05,
      "loss": 0.5281029939651489,
      "step": 4284
    },
    {
      "epoch": 0.35090951898371364,
      "grad_norm": 0.23295748233795166,
      "learning_rate": 1.9899327988400478e-05,
      "loss": 0.6705214381217957,
      "step": 4285
    },
    {
      "epoch": 0.35099141152023255,
      "grad_norm": 0.18341365456581116,
      "learning_rate": 1.9897444548593984e-05,
      "loss": 0.8179709911346436,
      "step": 4286
    },
    {
      "epoch": 0.3510733040567515,
      "grad_norm": 0.1692945808172226,
      "learning_rate": 1.9895560765532017e-05,
      "loss": 0.8284463286399841,
      "step": 4287
    },
    {
      "epoch": 0.3511551965932705,
      "grad_norm": 0.17492544651031494,
      "learning_rate": 1.9893676639296467e-05,
      "loss": 1.2347835302352905,
      "step": 4288
    },
    {
      "epoch": 0.35123708912978946,
      "grad_norm": 0.1503075808286667,
      "learning_rate": 1.989179216996922e-05,
      "loss": 0.7111825942993164,
      "step": 4289
    },
    {
      "epoch": 0.35131898166630837,
      "grad_norm": 0.16423186659812927,
      "learning_rate": 1.9889907357632174e-05,
      "loss": 0.8143547773361206,
      "step": 4290
    },
    {
      "epoch": 0.35140087420282734,
      "grad_norm": 0.16754703223705292,
      "learning_rate": 1.9888022202367264e-05,
      "loss": 0.9910504221916199,
      "step": 4291
    },
    {
      "epoch": 0.3514827667393463,
      "grad_norm": 0.17540523409843445,
      "learning_rate": 1.988613670425642e-05,
      "loss": 0.8069806098937988,
      "step": 4292
    },
    {
      "epoch": 0.3515646592758652,
      "grad_norm": 0.15714527666568756,
      "learning_rate": 1.988425086338159e-05,
      "loss": 0.7547283172607422,
      "step": 4293
    },
    {
      "epoch": 0.3516465518123842,
      "grad_norm": 0.21366292238235474,
      "learning_rate": 1.9882364679824747e-05,
      "loss": 0.6168365478515625,
      "step": 4294
    },
    {
      "epoch": 0.35172844434890316,
      "grad_norm": 0.2177937626838684,
      "learning_rate": 1.988047815366787e-05,
      "loss": 1.053797960281372,
      "step": 4295
    },
    {
      "epoch": 0.3518103368854221,
      "grad_norm": 0.1770176887512207,
      "learning_rate": 1.9878591284992955e-05,
      "loss": 0.7101513147354126,
      "step": 4296
    },
    {
      "epoch": 0.35189222942194104,
      "grad_norm": 0.1441406011581421,
      "learning_rate": 1.9876704073882015e-05,
      "loss": 0.7538225054740906,
      "step": 4297
    },
    {
      "epoch": 0.35197412195846,
      "grad_norm": 0.19140158593654633,
      "learning_rate": 1.9874816520417072e-05,
      "loss": 1.0232203006744385,
      "step": 4298
    },
    {
      "epoch": 0.352056014494979,
      "grad_norm": 0.16455036401748657,
      "learning_rate": 1.987292862468017e-05,
      "loss": 0.7590416073799133,
      "step": 4299
    },
    {
      "epoch": 0.35213790703149794,
      "grad_norm": 0.2422240972518921,
      "learning_rate": 1.9871040386753366e-05,
      "loss": 0.9051753282546997,
      "step": 4300
    },
    {
      "epoch": 0.35221979956801686,
      "grad_norm": 0.17255592346191406,
      "learning_rate": 1.9869151806718725e-05,
      "loss": 0.6966418623924255,
      "step": 4301
    },
    {
      "epoch": 0.3523016921045358,
      "grad_norm": 0.19271442294120789,
      "learning_rate": 1.9867262884658335e-05,
      "loss": 1.1196495294570923,
      "step": 4302
    },
    {
      "epoch": 0.3523835846410548,
      "grad_norm": 0.17691737413406372,
      "learning_rate": 1.9865373620654302e-05,
      "loss": 0.6582580208778381,
      "step": 4303
    },
    {
      "epoch": 0.3524654771775737,
      "grad_norm": 0.22815318405628204,
      "learning_rate": 1.986348401478873e-05,
      "loss": 0.7224811911582947,
      "step": 4304
    },
    {
      "epoch": 0.3525473697140927,
      "grad_norm": 0.19707545638084412,
      "learning_rate": 1.986159406714376e-05,
      "loss": 0.8755302429199219,
      "step": 4305
    },
    {
      "epoch": 0.35262926225061164,
      "grad_norm": 0.27530646324157715,
      "learning_rate": 1.9859703777801533e-05,
      "loss": 0.8659212589263916,
      "step": 4306
    },
    {
      "epoch": 0.3527111547871306,
      "grad_norm": 0.1753542423248291,
      "learning_rate": 1.9857813146844204e-05,
      "loss": 0.6943292021751404,
      "step": 4307
    },
    {
      "epoch": 0.3527930473236495,
      "grad_norm": 0.22514866292476654,
      "learning_rate": 1.9855922174353954e-05,
      "loss": 1.1258009672164917,
      "step": 4308
    },
    {
      "epoch": 0.3528749398601685,
      "grad_norm": 0.1806892454624176,
      "learning_rate": 1.9854030860412973e-05,
      "loss": 0.6063889861106873,
      "step": 4309
    },
    {
      "epoch": 0.35295683239668746,
      "grad_norm": 0.18263940513134003,
      "learning_rate": 1.9852139205103457e-05,
      "loss": 1.044600009918213,
      "step": 4310
    },
    {
      "epoch": 0.3530387249332064,
      "grad_norm": 0.22632959485054016,
      "learning_rate": 1.9850247208507633e-05,
      "loss": 1.0330491065979004,
      "step": 4311
    },
    {
      "epoch": 0.35312061746972534,
      "grad_norm": 0.2883365750312805,
      "learning_rate": 1.9848354870707734e-05,
      "loss": 0.7334637641906738,
      "step": 4312
    },
    {
      "epoch": 0.3532025100062443,
      "grad_norm": 0.1803656965494156,
      "learning_rate": 1.9846462191786003e-05,
      "loss": 0.9466259479522705,
      "step": 4313
    },
    {
      "epoch": 0.3532844025427633,
      "grad_norm": 0.18818654119968414,
      "learning_rate": 1.984456917182471e-05,
      "loss": 0.5185015797615051,
      "step": 4314
    },
    {
      "epoch": 0.3533662950792822,
      "grad_norm": 0.17948360741138458,
      "learning_rate": 1.9842675810906135e-05,
      "loss": 0.968687117099762,
      "step": 4315
    },
    {
      "epoch": 0.35344818761580116,
      "grad_norm": 0.15650972723960876,
      "learning_rate": 1.984078210911256e-05,
      "loss": 0.5491303205490112,
      "step": 4316
    },
    {
      "epoch": 0.3535300801523201,
      "grad_norm": 0.1698748767375946,
      "learning_rate": 1.9838888066526308e-05,
      "loss": 0.7467520236968994,
      "step": 4317
    },
    {
      "epoch": 0.3536119726888391,
      "grad_norm": 0.15565535426139832,
      "learning_rate": 1.983699368322969e-05,
      "loss": 0.8507865071296692,
      "step": 4318
    },
    {
      "epoch": 0.353693865225358,
      "grad_norm": 0.18987567722797394,
      "learning_rate": 1.9835098959305054e-05,
      "loss": 1.0501869916915894,
      "step": 4319
    },
    {
      "epoch": 0.353775757761877,
      "grad_norm": 0.18029314279556274,
      "learning_rate": 1.9833203894834743e-05,
      "loss": 0.7362135052680969,
      "step": 4320
    },
    {
      "epoch": 0.35385765029839594,
      "grad_norm": 0.17571952939033508,
      "learning_rate": 1.983130848990113e-05,
      "loss": 0.6698356866836548,
      "step": 4321
    },
    {
      "epoch": 0.35393954283491486,
      "grad_norm": 0.1656728982925415,
      "learning_rate": 1.9829412744586592e-05,
      "loss": 0.8587588667869568,
      "step": 4322
    },
    {
      "epoch": 0.3540214353714338,
      "grad_norm": 0.18742333352565765,
      "learning_rate": 1.982751665897353e-05,
      "loss": 0.9078229069709778,
      "step": 4323
    },
    {
      "epoch": 0.3541033279079528,
      "grad_norm": 0.17899560928344727,
      "learning_rate": 1.9825620233144355e-05,
      "loss": 0.7761216759681702,
      "step": 4324
    },
    {
      "epoch": 0.35418522044447176,
      "grad_norm": 0.18317992985248566,
      "learning_rate": 1.9823723467181494e-05,
      "loss": 1.0347492694854736,
      "step": 4325
    },
    {
      "epoch": 0.3542671129809907,
      "grad_norm": 0.16234563291072845,
      "learning_rate": 1.982182636116739e-05,
      "loss": 0.45347118377685547,
      "step": 4326
    },
    {
      "epoch": 0.35434900551750964,
      "grad_norm": 0.18607275187969208,
      "learning_rate": 1.9819928915184493e-05,
      "loss": 0.6735910773277283,
      "step": 4327
    },
    {
      "epoch": 0.3544308980540286,
      "grad_norm": 0.18664667010307312,
      "learning_rate": 1.981803112931528e-05,
      "loss": 0.9249265193939209,
      "step": 4328
    },
    {
      "epoch": 0.3545127905905476,
      "grad_norm": 0.17043016850948334,
      "learning_rate": 1.981613300364223e-05,
      "loss": 0.6687347888946533,
      "step": 4329
    },
    {
      "epoch": 0.3545946831270665,
      "grad_norm": 0.19587352871894836,
      "learning_rate": 1.981423453824785e-05,
      "loss": 1.0197314023971558,
      "step": 4330
    },
    {
      "epoch": 0.35467657566358546,
      "grad_norm": 0.17299440503120422,
      "learning_rate": 1.9812335733214652e-05,
      "loss": 0.6650671362876892,
      "step": 4331
    },
    {
      "epoch": 0.3547584682001044,
      "grad_norm": 0.18933191895484924,
      "learning_rate": 1.9810436588625168e-05,
      "loss": 1.0059159994125366,
      "step": 4332
    },
    {
      "epoch": 0.35484036073662334,
      "grad_norm": 0.18306343257427216,
      "learning_rate": 1.9808537104561938e-05,
      "loss": 0.7553528547286987,
      "step": 4333
    },
    {
      "epoch": 0.3549222532731423,
      "grad_norm": 0.1626724749803543,
      "learning_rate": 1.980663728110753e-05,
      "loss": 0.6667891144752502,
      "step": 4334
    },
    {
      "epoch": 0.3550041458096613,
      "grad_norm": 0.1766515076160431,
      "learning_rate": 1.9804737118344504e-05,
      "loss": 0.8149048686027527,
      "step": 4335
    },
    {
      "epoch": 0.35508603834618024,
      "grad_norm": 0.1662658303976059,
      "learning_rate": 1.9802836616355465e-05,
      "loss": 0.6821157336235046,
      "step": 4336
    },
    {
      "epoch": 0.35516793088269916,
      "grad_norm": 0.15897555649280548,
      "learning_rate": 1.9800935775223006e-05,
      "loss": 0.9281177520751953,
      "step": 4337
    },
    {
      "epoch": 0.3552498234192181,
      "grad_norm": 0.20516298711299896,
      "learning_rate": 1.9799034595029747e-05,
      "loss": 0.5188025236129761,
      "step": 4338
    },
    {
      "epoch": 0.3553317159557371,
      "grad_norm": 0.1998523324728012,
      "learning_rate": 1.9797133075858327e-05,
      "loss": 0.7048397660255432,
      "step": 4339
    },
    {
      "epoch": 0.35541360849225606,
      "grad_norm": 0.1746547520160675,
      "learning_rate": 1.9795231217791386e-05,
      "loss": 0.7627096772193909,
      "step": 4340
    },
    {
      "epoch": 0.355495501028775,
      "grad_norm": 0.21026314795017242,
      "learning_rate": 1.979332902091159e-05,
      "loss": 0.7111027836799622,
      "step": 4341
    },
    {
      "epoch": 0.35557739356529394,
      "grad_norm": 0.18885231018066406,
      "learning_rate": 1.9791426485301618e-05,
      "loss": 0.7639370560646057,
      "step": 4342
    },
    {
      "epoch": 0.3556592861018129,
      "grad_norm": 0.16572143137454987,
      "learning_rate": 1.978952361104416e-05,
      "loss": 0.914178192615509,
      "step": 4343
    },
    {
      "epoch": 0.3557411786383318,
      "grad_norm": 0.18069341778755188,
      "learning_rate": 1.9787620398221923e-05,
      "loss": 0.7422683238983154,
      "step": 4344
    },
    {
      "epoch": 0.3558230711748508,
      "grad_norm": 0.16477006673812866,
      "learning_rate": 1.9785716846917632e-05,
      "loss": 0.8486352562904358,
      "step": 4345
    },
    {
      "epoch": 0.35590496371136976,
      "grad_norm": 0.17174987494945526,
      "learning_rate": 1.9783812957214015e-05,
      "loss": 0.8172500133514404,
      "step": 4346
    },
    {
      "epoch": 0.3559868562478887,
      "grad_norm": 0.2293248474597931,
      "learning_rate": 1.978190872919383e-05,
      "loss": 0.970594048500061,
      "step": 4347
    },
    {
      "epoch": 0.35606874878440764,
      "grad_norm": 0.18764536082744598,
      "learning_rate": 1.978000416293984e-05,
      "loss": 0.905680239200592,
      "step": 4348
    },
    {
      "epoch": 0.3561506413209266,
      "grad_norm": 0.16522447764873505,
      "learning_rate": 1.9778099258534825e-05,
      "loss": 0.8441263437271118,
      "step": 4349
    },
    {
      "epoch": 0.3562325338574456,
      "grad_norm": 0.1701192557811737,
      "learning_rate": 1.977619401606158e-05,
      "loss": 0.6271524429321289,
      "step": 4350
    },
    {
      "epoch": 0.35631442639396455,
      "grad_norm": 0.19550827145576477,
      "learning_rate": 1.9774288435602917e-05,
      "loss": 0.8148967027664185,
      "step": 4351
    },
    {
      "epoch": 0.35639631893048346,
      "grad_norm": 0.2333151400089264,
      "learning_rate": 1.9772382517241653e-05,
      "loss": 0.8291774392127991,
      "step": 4352
    },
    {
      "epoch": 0.3564782114670024,
      "grad_norm": 0.20872347056865692,
      "learning_rate": 1.977047626106064e-05,
      "loss": 0.7301253080368042,
      "step": 4353
    },
    {
      "epoch": 0.3565601040035214,
      "grad_norm": 0.15973712503910065,
      "learning_rate": 1.9768569667142723e-05,
      "loss": 0.5338894128799438,
      "step": 4354
    },
    {
      "epoch": 0.3566419965400403,
      "grad_norm": 0.19325588643550873,
      "learning_rate": 1.9766662735570768e-05,
      "loss": 0.9150426983833313,
      "step": 4355
    },
    {
      "epoch": 0.3567238890765593,
      "grad_norm": 0.21037428081035614,
      "learning_rate": 1.9764755466427665e-05,
      "loss": 1.0235742330551147,
      "step": 4356
    },
    {
      "epoch": 0.35680578161307824,
      "grad_norm": 0.23164819180965424,
      "learning_rate": 1.976284785979631e-05,
      "loss": 0.7465975284576416,
      "step": 4357
    },
    {
      "epoch": 0.3568876741495972,
      "grad_norm": 0.19242756068706512,
      "learning_rate": 1.976093991575961e-05,
      "loss": 0.7502991557121277,
      "step": 4358
    },
    {
      "epoch": 0.3569695666861161,
      "grad_norm": 0.18773558735847473,
      "learning_rate": 1.9759031634400498e-05,
      "loss": 0.9565351009368896,
      "step": 4359
    },
    {
      "epoch": 0.3570514592226351,
      "grad_norm": 0.1774405837059021,
      "learning_rate": 1.975712301580192e-05,
      "loss": 0.9624481797218323,
      "step": 4360
    },
    {
      "epoch": 0.35713335175915406,
      "grad_norm": 0.18466518819332123,
      "learning_rate": 1.9755214060046818e-05,
      "loss": 0.6589597463607788,
      "step": 4361
    },
    {
      "epoch": 0.35721524429567303,
      "grad_norm": 0.19594885408878326,
      "learning_rate": 1.9753304767218178e-05,
      "loss": 0.8354603052139282,
      "step": 4362
    },
    {
      "epoch": 0.35729713683219194,
      "grad_norm": 0.19095203280448914,
      "learning_rate": 1.9751395137398974e-05,
      "loss": 0.655829131603241,
      "step": 4363
    },
    {
      "epoch": 0.3573790293687109,
      "grad_norm": 0.21325017511844635,
      "learning_rate": 1.9749485170672216e-05,
      "loss": 1.09444260597229,
      "step": 4364
    },
    {
      "epoch": 0.3574609219052299,
      "grad_norm": 0.1545410007238388,
      "learning_rate": 1.9747574867120912e-05,
      "loss": 0.6071938872337341,
      "step": 4365
    },
    {
      "epoch": 0.3575428144417488,
      "grad_norm": 0.1673203855752945,
      "learning_rate": 1.9745664226828096e-05,
      "loss": 0.7164135575294495,
      "step": 4366
    },
    {
      "epoch": 0.35762470697826776,
      "grad_norm": 0.1788136512041092,
      "learning_rate": 1.9743753249876807e-05,
      "loss": 0.7308321595191956,
      "step": 4367
    },
    {
      "epoch": 0.3577065995147867,
      "grad_norm": 0.189166858792305,
      "learning_rate": 1.9741841936350112e-05,
      "loss": 0.6530867218971252,
      "step": 4368
    },
    {
      "epoch": 0.3577884920513057,
      "grad_norm": 0.17598296701908112,
      "learning_rate": 1.9739930286331078e-05,
      "loss": 0.595264196395874,
      "step": 4369
    },
    {
      "epoch": 0.3578703845878246,
      "grad_norm": 0.1730663776397705,
      "learning_rate": 1.97380182999028e-05,
      "loss": 0.8073391914367676,
      "step": 4370
    },
    {
      "epoch": 0.3579522771243436,
      "grad_norm": 0.1832951158285141,
      "learning_rate": 1.9736105977148368e-05,
      "loss": 0.8449031710624695,
      "step": 4371
    },
    {
      "epoch": 0.35803416966086254,
      "grad_norm": 0.17041772603988647,
      "learning_rate": 1.9734193318150914e-05,
      "loss": 1.0381929874420166,
      "step": 4372
    },
    {
      "epoch": 0.3581160621973815,
      "grad_norm": 0.17107076942920685,
      "learning_rate": 1.9732280322993557e-05,
      "loss": 0.5129362940788269,
      "step": 4373
    },
    {
      "epoch": 0.3581979547339004,
      "grad_norm": 0.18035803735256195,
      "learning_rate": 1.9730366991759457e-05,
      "loss": 0.98234623670578,
      "step": 4374
    },
    {
      "epoch": 0.3582798472704194,
      "grad_norm": 0.19209909439086914,
      "learning_rate": 1.9728453324531764e-05,
      "loss": 1.0065691471099854,
      "step": 4375
    },
    {
      "epoch": 0.35836173980693836,
      "grad_norm": 0.21134227514266968,
      "learning_rate": 1.972653932139366e-05,
      "loss": 0.5568132996559143,
      "step": 4376
    },
    {
      "epoch": 0.3584436323434573,
      "grad_norm": 0.18635819852352142,
      "learning_rate": 1.9724624982428335e-05,
      "loss": 0.6490844488143921,
      "step": 4377
    },
    {
      "epoch": 0.35852552487997624,
      "grad_norm": 0.22073891758918762,
      "learning_rate": 1.972271030771899e-05,
      "loss": 1.0237884521484375,
      "step": 4378
    },
    {
      "epoch": 0.3586074174164952,
      "grad_norm": 0.22047628462314606,
      "learning_rate": 1.972079529734885e-05,
      "loss": 1.0606614351272583,
      "step": 4379
    },
    {
      "epoch": 0.3586893099530142,
      "grad_norm": 0.17419691383838654,
      "learning_rate": 1.9718879951401146e-05,
      "loss": 0.6295831799507141,
      "step": 4380
    },
    {
      "epoch": 0.3587712024895331,
      "grad_norm": 0.1864946484565735,
      "learning_rate": 1.9716964269959127e-05,
      "loss": 0.6515485644340515,
      "step": 4381
    },
    {
      "epoch": 0.35885309502605206,
      "grad_norm": 0.1874263733625412,
      "learning_rate": 1.9715048253106057e-05,
      "loss": 0.8942470550537109,
      "step": 4382
    },
    {
      "epoch": 0.35893498756257103,
      "grad_norm": 0.2038254737854004,
      "learning_rate": 1.9713131900925216e-05,
      "loss": 0.9094241261482239,
      "step": 4383
    },
    {
      "epoch": 0.35901688009909,
      "grad_norm": 0.23463955521583557,
      "learning_rate": 1.9711215213499892e-05,
      "loss": 1.0054821968078613,
      "step": 4384
    },
    {
      "epoch": 0.3590987726356089,
      "grad_norm": 0.19530893862247467,
      "learning_rate": 1.9709298190913395e-05,
      "loss": 0.6390146017074585,
      "step": 4385
    },
    {
      "epoch": 0.3591806651721279,
      "grad_norm": 0.2438758760690689,
      "learning_rate": 1.970738083324905e-05,
      "loss": 0.7766982316970825,
      "step": 4386
    },
    {
      "epoch": 0.35926255770864685,
      "grad_norm": 0.1736193299293518,
      "learning_rate": 1.9705463140590185e-05,
      "loss": 0.47715631127357483,
      "step": 4387
    },
    {
      "epoch": 0.35934445024516576,
      "grad_norm": 0.17889083921909332,
      "learning_rate": 1.9703545113020157e-05,
      "loss": 0.9706364274024963,
      "step": 4388
    },
    {
      "epoch": 0.3594263427816847,
      "grad_norm": 0.22794081270694733,
      "learning_rate": 1.970162675062233e-05,
      "loss": 0.8281793594360352,
      "step": 4389
    },
    {
      "epoch": 0.3595082353182037,
      "grad_norm": 0.18033763766288757,
      "learning_rate": 1.9699708053480087e-05,
      "loss": 0.8748568296432495,
      "step": 4390
    },
    {
      "epoch": 0.35959012785472266,
      "grad_norm": 0.19694048166275024,
      "learning_rate": 1.9697789021676816e-05,
      "loss": 0.6674535870552063,
      "step": 4391
    },
    {
      "epoch": 0.3596720203912416,
      "grad_norm": 0.19479872286319733,
      "learning_rate": 1.9695869655295927e-05,
      "loss": 0.9064968228340149,
      "step": 4392
    },
    {
      "epoch": 0.35975391292776054,
      "grad_norm": 0.18926383554935455,
      "learning_rate": 1.9693949954420847e-05,
      "loss": 0.7663837671279907,
      "step": 4393
    },
    {
      "epoch": 0.3598358054642795,
      "grad_norm": 0.18565207719802856,
      "learning_rate": 1.9692029919135014e-05,
      "loss": 0.8070425391197205,
      "step": 4394
    },
    {
      "epoch": 0.3599176980007984,
      "grad_norm": 0.1964593529701233,
      "learning_rate": 1.969010954952188e-05,
      "loss": 0.9323968887329102,
      "step": 4395
    },
    {
      "epoch": 0.3599995905373174,
      "grad_norm": 0.15867365896701813,
      "learning_rate": 1.968818884566491e-05,
      "loss": 0.7625018358230591,
      "step": 4396
    },
    {
      "epoch": 0.36008148307383636,
      "grad_norm": 0.17026787996292114,
      "learning_rate": 1.9686267807647594e-05,
      "loss": 0.9231011271476746,
      "step": 4397
    },
    {
      "epoch": 0.36016337561035533,
      "grad_norm": 0.2280600517988205,
      "learning_rate": 1.9684346435553415e-05,
      "loss": 0.7552457451820374,
      "step": 4398
    },
    {
      "epoch": 0.36024526814687424,
      "grad_norm": 0.18990188837051392,
      "learning_rate": 1.968242472946589e-05,
      "loss": 0.7399788498878479,
      "step": 4399
    },
    {
      "epoch": 0.3603271606833932,
      "grad_norm": 0.18415255844593048,
      "learning_rate": 1.9680502689468547e-05,
      "loss": 0.714137077331543,
      "step": 4400
    },
    {
      "epoch": 0.3604090532199122,
      "grad_norm": 0.1729525476694107,
      "learning_rate": 1.967858031564492e-05,
      "loss": 0.8887204527854919,
      "step": 4401
    },
    {
      "epoch": 0.36049094575643115,
      "grad_norm": 0.17705397307872772,
      "learning_rate": 1.9676657608078573e-05,
      "loss": 0.6605455279350281,
      "step": 4402
    },
    {
      "epoch": 0.36057283829295006,
      "grad_norm": 0.1925312727689743,
      "learning_rate": 1.9674734566853066e-05,
      "loss": 0.8454083204269409,
      "step": 4403
    },
    {
      "epoch": 0.36065473082946903,
      "grad_norm": 0.18488526344299316,
      "learning_rate": 1.967281119205198e-05,
      "loss": 1.1678919792175293,
      "step": 4404
    },
    {
      "epoch": 0.360736623365988,
      "grad_norm": 0.15734003484249115,
      "learning_rate": 1.9670887483758927e-05,
      "loss": 0.8140684366226196,
      "step": 4405
    },
    {
      "epoch": 0.3608185159025069,
      "grad_norm": 0.19680911302566528,
      "learning_rate": 1.9668963442057503e-05,
      "loss": 0.984260082244873,
      "step": 4406
    },
    {
      "epoch": 0.3609004084390259,
      "grad_norm": 0.19573622941970825,
      "learning_rate": 1.966703906703134e-05,
      "loss": 0.8262365460395813,
      "step": 4407
    },
    {
      "epoch": 0.36098230097554485,
      "grad_norm": 0.1551995426416397,
      "learning_rate": 1.9665114358764085e-05,
      "loss": 0.9407004714012146,
      "step": 4408
    },
    {
      "epoch": 0.3610641935120638,
      "grad_norm": 0.1732412725687027,
      "learning_rate": 1.966318931733939e-05,
      "loss": 0.9971442222595215,
      "step": 4409
    },
    {
      "epoch": 0.3611460860485827,
      "grad_norm": 0.18320736289024353,
      "learning_rate": 1.966126394284092e-05,
      "loss": 1.1332502365112305,
      "step": 4410
    },
    {
      "epoch": 0.3612279785851017,
      "grad_norm": 0.20795078575611115,
      "learning_rate": 1.965933823535237e-05,
      "loss": 0.8641313314437866,
      "step": 4411
    },
    {
      "epoch": 0.36130987112162066,
      "grad_norm": 0.22419123351573944,
      "learning_rate": 1.9657412194957432e-05,
      "loss": 0.7736263871192932,
      "step": 4412
    },
    {
      "epoch": 0.36139176365813963,
      "grad_norm": 0.16440878808498383,
      "learning_rate": 1.9655485821739823e-05,
      "loss": 0.8979436159133911,
      "step": 4413
    },
    {
      "epoch": 0.36147365619465854,
      "grad_norm": 0.21328333020210266,
      "learning_rate": 1.9653559115783267e-05,
      "loss": 0.7405194044113159,
      "step": 4414
    },
    {
      "epoch": 0.3615555487311775,
      "grad_norm": 0.21749716997146606,
      "learning_rate": 1.965163207717151e-05,
      "loss": 0.9242398738861084,
      "step": 4415
    },
    {
      "epoch": 0.3616374412676965,
      "grad_norm": 0.16800417006015778,
      "learning_rate": 1.9649704705988308e-05,
      "loss": 0.7356133460998535,
      "step": 4416
    },
    {
      "epoch": 0.3617193338042154,
      "grad_norm": 0.22920911014080048,
      "learning_rate": 1.9647777002317434e-05,
      "loss": 0.6302648782730103,
      "step": 4417
    },
    {
      "epoch": 0.36180122634073436,
      "grad_norm": 0.21627835929393768,
      "learning_rate": 1.9645848966242672e-05,
      "loss": 0.6307827830314636,
      "step": 4418
    },
    {
      "epoch": 0.36188311887725333,
      "grad_norm": 0.1857755482196808,
      "learning_rate": 1.9643920597847823e-05,
      "loss": 1.0630323886871338,
      "step": 4419
    },
    {
      "epoch": 0.3619650114137723,
      "grad_norm": 0.16455215215682983,
      "learning_rate": 1.96419918972167e-05,
      "loss": 0.6739237308502197,
      "step": 4420
    },
    {
      "epoch": 0.3620469039502912,
      "grad_norm": 0.17445345222949982,
      "learning_rate": 1.9640062864433136e-05,
      "loss": 0.5842790007591248,
      "step": 4421
    },
    {
      "epoch": 0.3621287964868102,
      "grad_norm": 0.20279544591903687,
      "learning_rate": 1.9638133499580972e-05,
      "loss": 0.8445159792900085,
      "step": 4422
    },
    {
      "epoch": 0.36221068902332915,
      "grad_norm": 0.18346132338047028,
      "learning_rate": 1.963620380274407e-05,
      "loss": 0.8005721569061279,
      "step": 4423
    },
    {
      "epoch": 0.3622925815598481,
      "grad_norm": 0.2000623345375061,
      "learning_rate": 1.9634273774006295e-05,
      "loss": 0.7876731157302856,
      "step": 4424
    },
    {
      "epoch": 0.362374474096367,
      "grad_norm": 0.17045186460018158,
      "learning_rate": 1.963234341345154e-05,
      "loss": 0.8150262832641602,
      "step": 4425
    },
    {
      "epoch": 0.362456366632886,
      "grad_norm": 0.20091471076011658,
      "learning_rate": 1.9630412721163706e-05,
      "loss": 0.7972959280014038,
      "step": 4426
    },
    {
      "epoch": 0.36253825916940496,
      "grad_norm": 0.1542266309261322,
      "learning_rate": 1.9628481697226705e-05,
      "loss": 0.7244962453842163,
      "step": 4427
    },
    {
      "epoch": 0.3626201517059239,
      "grad_norm": 0.16996188461780548,
      "learning_rate": 1.9626550341724474e-05,
      "loss": 0.5139568448066711,
      "step": 4428
    },
    {
      "epoch": 0.36270204424244284,
      "grad_norm": 0.1878669410943985,
      "learning_rate": 1.9624618654740955e-05,
      "loss": 0.7973442673683167,
      "step": 4429
    },
    {
      "epoch": 0.3627839367789618,
      "grad_norm": 0.18063956499099731,
      "learning_rate": 1.9622686636360098e-05,
      "loss": 1.0832916498184204,
      "step": 4430
    },
    {
      "epoch": 0.3628658293154808,
      "grad_norm": 0.16756555438041687,
      "learning_rate": 1.962075428666589e-05,
      "loss": 0.7238767743110657,
      "step": 4431
    },
    {
      "epoch": 0.3629477218519997,
      "grad_norm": 0.23344026505947113,
      "learning_rate": 1.961882160574231e-05,
      "loss": 0.8512411713600159,
      "step": 4432
    },
    {
      "epoch": 0.36302961438851866,
      "grad_norm": 0.2140486091375351,
      "learning_rate": 1.9616888593673367e-05,
      "loss": 0.9644138813018799,
      "step": 4433
    },
    {
      "epoch": 0.36311150692503763,
      "grad_norm": 0.1673780083656311,
      "learning_rate": 1.9614955250543074e-05,
      "loss": 0.7698075771331787,
      "step": 4434
    },
    {
      "epoch": 0.3631933994615566,
      "grad_norm": 0.1794469952583313,
      "learning_rate": 1.961302157643546e-05,
      "loss": 0.6335943341255188,
      "step": 4435
    },
    {
      "epoch": 0.3632752919980755,
      "grad_norm": 0.17696282267570496,
      "learning_rate": 1.961108757143457e-05,
      "loss": 0.7825096845626831,
      "step": 4436
    },
    {
      "epoch": 0.3633571845345945,
      "grad_norm": 0.1811041384935379,
      "learning_rate": 1.960915323562447e-05,
      "loss": 0.7318376302719116,
      "step": 4437
    },
    {
      "epoch": 0.36343907707111345,
      "grad_norm": 0.19609248638153076,
      "learning_rate": 1.960721856908923e-05,
      "loss": 0.7728772163391113,
      "step": 4438
    },
    {
      "epoch": 0.36352096960763236,
      "grad_norm": 0.19586695730686188,
      "learning_rate": 1.960528357191294e-05,
      "loss": 0.8522976636886597,
      "step": 4439
    },
    {
      "epoch": 0.36360286214415133,
      "grad_norm": 0.1879831999540329,
      "learning_rate": 1.96033482441797e-05,
      "loss": 0.9742370843887329,
      "step": 4440
    },
    {
      "epoch": 0.3636847546806703,
      "grad_norm": 0.2076287567615509,
      "learning_rate": 1.9601412585973633e-05,
      "loss": 0.8953176736831665,
      "step": 4441
    },
    {
      "epoch": 0.36376664721718927,
      "grad_norm": 0.20425009727478027,
      "learning_rate": 1.959947659737886e-05,
      "loss": 0.8315043449401855,
      "step": 4442
    },
    {
      "epoch": 0.3638485397537082,
      "grad_norm": 0.2228768914937973,
      "learning_rate": 1.959754027847954e-05,
      "loss": 0.5408908724784851,
      "step": 4443
    },
    {
      "epoch": 0.36393043229022715,
      "grad_norm": 0.19096027314662933,
      "learning_rate": 1.959560362935983e-05,
      "loss": 1.079308271408081,
      "step": 4444
    },
    {
      "epoch": 0.3640123248267461,
      "grad_norm": 0.20256954431533813,
      "learning_rate": 1.9593666650103895e-05,
      "loss": 0.8582855463027954,
      "step": 4445
    },
    {
      "epoch": 0.3640942173632651,
      "grad_norm": 0.14436964690685272,
      "learning_rate": 1.9591729340795937e-05,
      "loss": 0.6439790725708008,
      "step": 4446
    },
    {
      "epoch": 0.364176109899784,
      "grad_norm": 0.15410207211971283,
      "learning_rate": 1.9589791701520153e-05,
      "loss": 0.7378430962562561,
      "step": 4447
    },
    {
      "epoch": 0.36425800243630296,
      "grad_norm": 0.21897171437740326,
      "learning_rate": 1.9587853732360757e-05,
      "loss": 0.7293628454208374,
      "step": 4448
    },
    {
      "epoch": 0.36433989497282193,
      "grad_norm": 0.1672581285238266,
      "learning_rate": 1.958591543340199e-05,
      "loss": 0.8193011283874512,
      "step": 4449
    },
    {
      "epoch": 0.36442178750934084,
      "grad_norm": 0.20077411830425262,
      "learning_rate": 1.9583976804728095e-05,
      "loss": 0.538223385810852,
      "step": 4450
    },
    {
      "epoch": 0.3645036800458598,
      "grad_norm": 0.1502661556005478,
      "learning_rate": 1.958203784642333e-05,
      "loss": 0.9661727547645569,
      "step": 4451
    },
    {
      "epoch": 0.3645855725823788,
      "grad_norm": 0.2117607742547989,
      "learning_rate": 1.9580098558571974e-05,
      "loss": 1.007814884185791,
      "step": 4452
    },
    {
      "epoch": 0.36466746511889775,
      "grad_norm": 0.19347292184829712,
      "learning_rate": 1.957815894125831e-05,
      "loss": 1.0287089347839355,
      "step": 4453
    },
    {
      "epoch": 0.36474935765541666,
      "grad_norm": 0.164510115981102,
      "learning_rate": 1.9576218994566652e-05,
      "loss": 0.8123359680175781,
      "step": 4454
    },
    {
      "epoch": 0.36483125019193563,
      "grad_norm": 0.1980883628129959,
      "learning_rate": 1.957427871858131e-05,
      "loss": 0.6736574769020081,
      "step": 4455
    },
    {
      "epoch": 0.3649131427284546,
      "grad_norm": 0.20440638065338135,
      "learning_rate": 1.9572338113386615e-05,
      "loss": 0.7504818439483643,
      "step": 4456
    },
    {
      "epoch": 0.3649950352649735,
      "grad_norm": 0.1670597940683365,
      "learning_rate": 1.9570397179066922e-05,
      "loss": 0.8283258080482483,
      "step": 4457
    },
    {
      "epoch": 0.3650769278014925,
      "grad_norm": 0.20536115765571594,
      "learning_rate": 1.9568455915706585e-05,
      "loss": 1.1591969728469849,
      "step": 4458
    },
    {
      "epoch": 0.36515882033801145,
      "grad_norm": 0.2784768342971802,
      "learning_rate": 1.956651432338998e-05,
      "loss": 0.766333281993866,
      "step": 4459
    },
    {
      "epoch": 0.3652407128745304,
      "grad_norm": 0.2383124679327011,
      "learning_rate": 1.9564572402201505e-05,
      "loss": 1.1083064079284668,
      "step": 4460
    },
    {
      "epoch": 0.36532260541104933,
      "grad_norm": 0.2069420963525772,
      "learning_rate": 1.9562630152225553e-05,
      "loss": 0.6386774182319641,
      "step": 4461
    },
    {
      "epoch": 0.3654044979475683,
      "grad_norm": 0.20315736532211304,
      "learning_rate": 1.9560687573546546e-05,
      "loss": 1.005161166191101,
      "step": 4462
    },
    {
      "epoch": 0.36548639048408726,
      "grad_norm": 0.17940525710582733,
      "learning_rate": 1.9558744666248916e-05,
      "loss": 0.66644686460495,
      "step": 4463
    },
    {
      "epoch": 0.36556828302060623,
      "grad_norm": 0.17964328825473785,
      "learning_rate": 1.9556801430417114e-05,
      "loss": 1.230204463005066,
      "step": 4464
    },
    {
      "epoch": 0.36565017555712515,
      "grad_norm": 0.18787643313407898,
      "learning_rate": 1.9554857866135597e-05,
      "loss": 0.8045151233673096,
      "step": 4465
    },
    {
      "epoch": 0.3657320680936441,
      "grad_norm": 0.1781775951385498,
      "learning_rate": 1.955291397348884e-05,
      "loss": 0.8430500030517578,
      "step": 4466
    },
    {
      "epoch": 0.3658139606301631,
      "grad_norm": 0.16637390851974487,
      "learning_rate": 1.9550969752561335e-05,
      "loss": 0.8271950483322144,
      "step": 4467
    },
    {
      "epoch": 0.365895853166682,
      "grad_norm": 0.15370331704616547,
      "learning_rate": 1.954902520343758e-05,
      "loss": 0.8958649635314941,
      "step": 4468
    },
    {
      "epoch": 0.36597774570320096,
      "grad_norm": 0.1868627369403839,
      "learning_rate": 1.9547080326202098e-05,
      "loss": 0.6421324014663696,
      "step": 4469
    },
    {
      "epoch": 0.36605963823971993,
      "grad_norm": 0.17652401328086853,
      "learning_rate": 1.9545135120939425e-05,
      "loss": 0.6036357283592224,
      "step": 4470
    },
    {
      "epoch": 0.3661415307762389,
      "grad_norm": 0.18761026859283447,
      "learning_rate": 1.9543189587734105e-05,
      "loss": 0.7823152542114258,
      "step": 4471
    },
    {
      "epoch": 0.3662234233127578,
      "grad_norm": 0.15301331877708435,
      "learning_rate": 1.954124372667069e-05,
      "loss": 0.7712863087654114,
      "step": 4472
    },
    {
      "epoch": 0.3663053158492768,
      "grad_norm": 0.17974521219730377,
      "learning_rate": 1.9539297537833768e-05,
      "loss": 0.7622796297073364,
      "step": 4473
    },
    {
      "epoch": 0.36638720838579575,
      "grad_norm": 0.18023647367954254,
      "learning_rate": 1.9537351021307923e-05,
      "loss": 0.8440065383911133,
      "step": 4474
    },
    {
      "epoch": 0.3664691009223147,
      "grad_norm": 0.19292372465133667,
      "learning_rate": 1.9535404177177755e-05,
      "loss": 1.007076621055603,
      "step": 4475
    },
    {
      "epoch": 0.36655099345883363,
      "grad_norm": 0.18161523342132568,
      "learning_rate": 1.9533457005527887e-05,
      "loss": 0.9718837738037109,
      "step": 4476
    },
    {
      "epoch": 0.3666328859953526,
      "grad_norm": 0.13928084075450897,
      "learning_rate": 1.9531509506442953e-05,
      "loss": 0.7639694213867188,
      "step": 4477
    },
    {
      "epoch": 0.36671477853187157,
      "grad_norm": 0.2091611623764038,
      "learning_rate": 1.9529561680007588e-05,
      "loss": 1.0506787300109863,
      "step": 4478
    },
    {
      "epoch": 0.3667966710683905,
      "grad_norm": 0.2031441479921341,
      "learning_rate": 1.9527613526306468e-05,
      "loss": 0.8657310605049133,
      "step": 4479
    },
    {
      "epoch": 0.36687856360490945,
      "grad_norm": 0.1838437020778656,
      "learning_rate": 1.952566504542426e-05,
      "loss": 0.7657269835472107,
      "step": 4480
    },
    {
      "epoch": 0.3669604561414284,
      "grad_norm": 0.16421669721603394,
      "learning_rate": 1.9523716237445647e-05,
      "loss": 0.894124448299408,
      "step": 4481
    },
    {
      "epoch": 0.3670423486779474,
      "grad_norm": 0.1986609250307083,
      "learning_rate": 1.9521767102455343e-05,
      "loss": 0.8338479399681091,
      "step": 4482
    },
    {
      "epoch": 0.3671242412144663,
      "grad_norm": 0.1784464567899704,
      "learning_rate": 1.951981764053806e-05,
      "loss": 0.9913228154182434,
      "step": 4483
    },
    {
      "epoch": 0.36720613375098526,
      "grad_norm": 0.21472986042499542,
      "learning_rate": 1.951786785177853e-05,
      "loss": 0.8309758901596069,
      "step": 4484
    },
    {
      "epoch": 0.36728802628750423,
      "grad_norm": 0.15903311967849731,
      "learning_rate": 1.95159177362615e-05,
      "loss": 0.7856239676475525,
      "step": 4485
    },
    {
      "epoch": 0.3673699188240232,
      "grad_norm": 0.18653059005737305,
      "learning_rate": 1.9513967294071726e-05,
      "loss": 0.8931725025177002,
      "step": 4486
    },
    {
      "epoch": 0.3674518113605421,
      "grad_norm": 0.18549638986587524,
      "learning_rate": 1.951201652529399e-05,
      "loss": 0.727169394493103,
      "step": 4487
    },
    {
      "epoch": 0.3675337038970611,
      "grad_norm": 0.20123016834259033,
      "learning_rate": 1.9510065430013073e-05,
      "loss": 1.066023349761963,
      "step": 4488
    },
    {
      "epoch": 0.36761559643358005,
      "grad_norm": 0.22585538029670715,
      "learning_rate": 1.9508114008313783e-05,
      "loss": 0.7750298976898193,
      "step": 4489
    },
    {
      "epoch": 0.36769748897009896,
      "grad_norm": 0.16940414905548096,
      "learning_rate": 1.9506162260280935e-05,
      "loss": 0.4586896002292633,
      "step": 4490
    },
    {
      "epoch": 0.36777938150661793,
      "grad_norm": 0.16240467131137848,
      "learning_rate": 1.950421018599936e-05,
      "loss": 0.6740632057189941,
      "step": 4491
    },
    {
      "epoch": 0.3678612740431369,
      "grad_norm": 0.18858158588409424,
      "learning_rate": 1.95022577855539e-05,
      "loss": 0.8084611892700195,
      "step": 4492
    },
    {
      "epoch": 0.36794316657965587,
      "grad_norm": 0.18464578688144684,
      "learning_rate": 1.950030505902942e-05,
      "loss": 0.9965607523918152,
      "step": 4493
    },
    {
      "epoch": 0.3680250591161748,
      "grad_norm": 0.21004711091518402,
      "learning_rate": 1.9498352006510788e-05,
      "loss": 0.9133917093276978,
      "step": 4494
    },
    {
      "epoch": 0.36810695165269375,
      "grad_norm": 0.17623217403888702,
      "learning_rate": 1.9496398628082896e-05,
      "loss": 0.9391257762908936,
      "step": 4495
    },
    {
      "epoch": 0.3681888441892127,
      "grad_norm": 0.17982333898544312,
      "learning_rate": 1.9494444923830644e-05,
      "loss": 0.7609497308731079,
      "step": 4496
    },
    {
      "epoch": 0.3682707367257317,
      "grad_norm": 0.19167070090770721,
      "learning_rate": 1.9492490893838947e-05,
      "loss": 0.6060552597045898,
      "step": 4497
    },
    {
      "epoch": 0.3683526292622506,
      "grad_norm": 0.20592227578163147,
      "learning_rate": 1.9490536538192738e-05,
      "loss": 0.851462185382843,
      "step": 4498
    },
    {
      "epoch": 0.36843452179876957,
      "grad_norm": 0.17703038454055786,
      "learning_rate": 1.9488581856976958e-05,
      "loss": 0.7449119687080383,
      "step": 4499
    },
    {
      "epoch": 0.36851641433528853,
      "grad_norm": 0.18883945047855377,
      "learning_rate": 1.948662685027657e-05,
      "loss": 0.8299137353897095,
      "step": 4500
    },
    {
      "epoch": 0.36859830687180745,
      "grad_norm": 0.19982299208641052,
      "learning_rate": 1.948467151817654e-05,
      "loss": 0.7308027744293213,
      "step": 4501
    },
    {
      "epoch": 0.3686801994083264,
      "grad_norm": 0.19350045919418335,
      "learning_rate": 1.9482715860761864e-05,
      "loss": 0.8454387784004211,
      "step": 4502
    },
    {
      "epoch": 0.3687620919448454,
      "grad_norm": 0.1684250682592392,
      "learning_rate": 1.9480759878117535e-05,
      "loss": 0.6734793186187744,
      "step": 4503
    },
    {
      "epoch": 0.36884398448136435,
      "grad_norm": 0.18470391631126404,
      "learning_rate": 1.9478803570328566e-05,
      "loss": 0.9357970356941223,
      "step": 4504
    },
    {
      "epoch": 0.36892587701788326,
      "grad_norm": 0.21319517493247986,
      "learning_rate": 1.9476846937479995e-05,
      "loss": 1.3622959852218628,
      "step": 4505
    },
    {
      "epoch": 0.36900776955440223,
      "grad_norm": 0.15470333397388458,
      "learning_rate": 1.9474889979656862e-05,
      "loss": 0.7581462860107422,
      "step": 4506
    },
    {
      "epoch": 0.3690896620909212,
      "grad_norm": 0.18320509791374207,
      "learning_rate": 1.9472932696944227e-05,
      "loss": 0.8273270130157471,
      "step": 4507
    },
    {
      "epoch": 0.36917155462744017,
      "grad_norm": 0.22648395597934723,
      "learning_rate": 1.9470975089427155e-05,
      "loss": 0.8774992227554321,
      "step": 4508
    },
    {
      "epoch": 0.3692534471639591,
      "grad_norm": 0.22458767890930176,
      "learning_rate": 1.946901715719073e-05,
      "loss": 0.6364206671714783,
      "step": 4509
    },
    {
      "epoch": 0.36933533970047805,
      "grad_norm": 0.2132290154695511,
      "learning_rate": 1.9467058900320065e-05,
      "loss": 0.8736752271652222,
      "step": 4510
    },
    {
      "epoch": 0.369417232236997,
      "grad_norm": 0.17828500270843506,
      "learning_rate": 1.946510031890026e-05,
      "loss": 0.9607784748077393,
      "step": 4511
    },
    {
      "epoch": 0.36949912477351593,
      "grad_norm": 0.18401983380317688,
      "learning_rate": 1.946314141301645e-05,
      "loss": 1.117727518081665,
      "step": 4512
    },
    {
      "epoch": 0.3695810173100349,
      "grad_norm": 0.18003787100315094,
      "learning_rate": 1.9461182182753776e-05,
      "loss": 0.9633479118347168,
      "step": 4513
    },
    {
      "epoch": 0.36966290984655387,
      "grad_norm": 0.17685002088546753,
      "learning_rate": 1.9459222628197396e-05,
      "loss": 0.7344169020652771,
      "step": 4514
    },
    {
      "epoch": 0.36974480238307283,
      "grad_norm": 0.1823803335428238,
      "learning_rate": 1.9457262749432475e-05,
      "loss": 0.9310920238494873,
      "step": 4515
    },
    {
      "epoch": 0.36982669491959175,
      "grad_norm": 0.1701962649822235,
      "learning_rate": 1.9455302546544203e-05,
      "loss": 0.6976556777954102,
      "step": 4516
    },
    {
      "epoch": 0.3699085874561107,
      "grad_norm": 0.19401836395263672,
      "learning_rate": 1.9453342019617775e-05,
      "loss": 0.8871124982833862,
      "step": 4517
    },
    {
      "epoch": 0.3699904799926297,
      "grad_norm": 0.1735466569662094,
      "learning_rate": 1.9451381168738405e-05,
      "loss": 0.6942733526229858,
      "step": 4518
    },
    {
      "epoch": 0.37007237252914865,
      "grad_norm": 0.18954968452453613,
      "learning_rate": 1.944941999399132e-05,
      "loss": 0.8441452383995056,
      "step": 4519
    },
    {
      "epoch": 0.37015426506566756,
      "grad_norm": 0.1651827096939087,
      "learning_rate": 1.9447458495461763e-05,
      "loss": 0.7416768670082092,
      "step": 4520
    },
    {
      "epoch": 0.37023615760218653,
      "grad_norm": 0.19355501234531403,
      "learning_rate": 1.944549667323498e-05,
      "loss": 1.031585931777954,
      "step": 4521
    },
    {
      "epoch": 0.3703180501387055,
      "grad_norm": 0.19508761167526245,
      "learning_rate": 1.944353452739625e-05,
      "loss": 1.1424672603607178,
      "step": 4522
    },
    {
      "epoch": 0.3703999426752244,
      "grad_norm": 0.21336470544338226,
      "learning_rate": 1.9441572058030848e-05,
      "loss": 0.683007001876831,
      "step": 4523
    },
    {
      "epoch": 0.3704818352117434,
      "grad_norm": 0.175675168633461,
      "learning_rate": 1.9439609265224075e-05,
      "loss": 0.5875193476676941,
      "step": 4524
    },
    {
      "epoch": 0.37056372774826235,
      "grad_norm": 0.21545380353927612,
      "learning_rate": 1.9437646149061245e-05,
      "loss": 0.8703210949897766,
      "step": 4525
    },
    {
      "epoch": 0.3706456202847813,
      "grad_norm": 0.19088014960289001,
      "learning_rate": 1.943568270962768e-05,
      "loss": 0.7797892689704895,
      "step": 4526
    },
    {
      "epoch": 0.37072751282130023,
      "grad_norm": 0.14856427907943726,
      "learning_rate": 1.9433718947008716e-05,
      "loss": 0.9655696749687195,
      "step": 4527
    },
    {
      "epoch": 0.3708094053578192,
      "grad_norm": 0.1834736168384552,
      "learning_rate": 1.943175486128971e-05,
      "loss": 0.6576709747314453,
      "step": 4528
    },
    {
      "epoch": 0.37089129789433817,
      "grad_norm": 0.18833300471305847,
      "learning_rate": 1.9429790452556033e-05,
      "loss": 0.9876053333282471,
      "step": 4529
    },
    {
      "epoch": 0.3709731904308571,
      "grad_norm": 0.1802380532026291,
      "learning_rate": 1.942782572089306e-05,
      "loss": 0.8554468750953674,
      "step": 4530
    },
    {
      "epoch": 0.37105508296737605,
      "grad_norm": 0.2061423510313034,
      "learning_rate": 1.9425860666386187e-05,
      "loss": 0.9618027210235596,
      "step": 4531
    },
    {
      "epoch": 0.371136975503895,
      "grad_norm": 0.19142718613147736,
      "learning_rate": 1.9423895289120824e-05,
      "loss": 0.7149055600166321,
      "step": 4532
    },
    {
      "epoch": 0.371218868040414,
      "grad_norm": 0.18927709758281708,
      "learning_rate": 1.9421929589182393e-05,
      "loss": 0.773067831993103,
      "step": 4533
    },
    {
      "epoch": 0.3713007605769329,
      "grad_norm": 0.1648421436548233,
      "learning_rate": 1.941996356665634e-05,
      "loss": 0.8457905650138855,
      "step": 4534
    },
    {
      "epoch": 0.37138265311345187,
      "grad_norm": 0.18654143810272217,
      "learning_rate": 1.94179972216281e-05,
      "loss": 0.900816798210144,
      "step": 4535
    },
    {
      "epoch": 0.37146454564997083,
      "grad_norm": 0.20962196588516235,
      "learning_rate": 1.9416030554183155e-05,
      "loss": 0.6707956790924072,
      "step": 4536
    },
    {
      "epoch": 0.3715464381864898,
      "grad_norm": 0.1802428811788559,
      "learning_rate": 1.9414063564406977e-05,
      "loss": 0.8269150257110596,
      "step": 4537
    },
    {
      "epoch": 0.3716283307230087,
      "grad_norm": 0.19045491516590118,
      "learning_rate": 1.941209625238506e-05,
      "loss": 1.281597375869751,
      "step": 4538
    },
    {
      "epoch": 0.3717102232595277,
      "grad_norm": 0.21050530672073364,
      "learning_rate": 1.9410128618202913e-05,
      "loss": 0.8412893414497375,
      "step": 4539
    },
    {
      "epoch": 0.37179211579604665,
      "grad_norm": 0.18824219703674316,
      "learning_rate": 1.9408160661946052e-05,
      "loss": 1.096423864364624,
      "step": 4540
    },
    {
      "epoch": 0.37187400833256556,
      "grad_norm": 0.1591040939092636,
      "learning_rate": 1.940619238370002e-05,
      "loss": 0.9486916065216064,
      "step": 4541
    },
    {
      "epoch": 0.37195590086908453,
      "grad_norm": 0.1613236665725708,
      "learning_rate": 1.940422378355036e-05,
      "loss": 0.6722403764724731,
      "step": 4542
    },
    {
      "epoch": 0.3720377934056035,
      "grad_norm": 0.19576163589954376,
      "learning_rate": 1.9402254861582645e-05,
      "loss": 1.0907971858978271,
      "step": 4543
    },
    {
      "epoch": 0.37211968594212247,
      "grad_norm": 0.2036546766757965,
      "learning_rate": 1.940028561788244e-05,
      "loss": 1.0277293920516968,
      "step": 4544
    },
    {
      "epoch": 0.3722015784786414,
      "grad_norm": 0.20037733018398285,
      "learning_rate": 1.9398316052535346e-05,
      "loss": 0.9633432626724243,
      "step": 4545
    },
    {
      "epoch": 0.37228347101516035,
      "grad_norm": 0.19084274768829346,
      "learning_rate": 1.9396346165626964e-05,
      "loss": 0.6675033569335938,
      "step": 4546
    },
    {
      "epoch": 0.3723653635516793,
      "grad_norm": 0.16899316012859344,
      "learning_rate": 1.9394375957242915e-05,
      "loss": 0.7685231566429138,
      "step": 4547
    },
    {
      "epoch": 0.3724472560881983,
      "grad_norm": 0.1697678416967392,
      "learning_rate": 1.9392405427468832e-05,
      "loss": 0.4490240514278412,
      "step": 4548
    },
    {
      "epoch": 0.3725291486247172,
      "grad_norm": 0.2173643261194229,
      "learning_rate": 1.9390434576390355e-05,
      "loss": 1.0609660148620605,
      "step": 4549
    },
    {
      "epoch": 0.37261104116123617,
      "grad_norm": 0.2078549563884735,
      "learning_rate": 1.938846340409316e-05,
      "loss": 0.571779727935791,
      "step": 4550
    },
    {
      "epoch": 0.37269293369775514,
      "grad_norm": 0.16200584173202515,
      "learning_rate": 1.9386491910662916e-05,
      "loss": 0.8075960874557495,
      "step": 4551
    },
    {
      "epoch": 0.37277482623427405,
      "grad_norm": 0.16025245189666748,
      "learning_rate": 1.9384520096185306e-05,
      "loss": 0.9250539541244507,
      "step": 4552
    },
    {
      "epoch": 0.372856718770793,
      "grad_norm": 0.1940290778875351,
      "learning_rate": 1.9382547960746038e-05,
      "loss": 1.031877040863037,
      "step": 4553
    },
    {
      "epoch": 0.372938611307312,
      "grad_norm": 0.18438591063022614,
      "learning_rate": 1.938057550443083e-05,
      "loss": 0.6505870223045349,
      "step": 4554
    },
    {
      "epoch": 0.37302050384383095,
      "grad_norm": 0.14912840723991394,
      "learning_rate": 1.9378602727325416e-05,
      "loss": 0.5663623213768005,
      "step": 4555
    },
    {
      "epoch": 0.37310239638034987,
      "grad_norm": 0.19380109012126923,
      "learning_rate": 1.937662962951553e-05,
      "loss": 0.8810120224952698,
      "step": 4556
    },
    {
      "epoch": 0.37318428891686883,
      "grad_norm": 0.18831732869148254,
      "learning_rate": 1.9374656211086945e-05,
      "loss": 0.9577232599258423,
      "step": 4557
    },
    {
      "epoch": 0.3732661814533878,
      "grad_norm": 0.2217322140932083,
      "learning_rate": 1.9372682472125424e-05,
      "loss": 0.893231987953186,
      "step": 4558
    },
    {
      "epoch": 0.37334807398990677,
      "grad_norm": 0.203490749001503,
      "learning_rate": 1.9370708412716755e-05,
      "loss": 0.8800333142280579,
      "step": 4559
    },
    {
      "epoch": 0.3734299665264257,
      "grad_norm": 0.1817035973072052,
      "learning_rate": 1.9368734032946745e-05,
      "loss": 0.9259476661682129,
      "step": 4560
    },
    {
      "epoch": 0.37351185906294465,
      "grad_norm": 0.20660722255706787,
      "learning_rate": 1.93667593329012e-05,
      "loss": 0.9931461215019226,
      "step": 4561
    },
    {
      "epoch": 0.3735937515994636,
      "grad_norm": 0.17772835493087769,
      "learning_rate": 1.9364784312665957e-05,
      "loss": 0.7022066712379456,
      "step": 4562
    },
    {
      "epoch": 0.37367564413598253,
      "grad_norm": 0.1951800286769867,
      "learning_rate": 1.9362808972326853e-05,
      "loss": 0.8044015169143677,
      "step": 4563
    },
    {
      "epoch": 0.3737575366725015,
      "grad_norm": 0.18887825310230255,
      "learning_rate": 1.9360833311969742e-05,
      "loss": 0.7709892988204956,
      "step": 4564
    },
    {
      "epoch": 0.37383942920902047,
      "grad_norm": 0.20662137866020203,
      "learning_rate": 1.9358857331680505e-05,
      "loss": 0.8899393677711487,
      "step": 4565
    },
    {
      "epoch": 0.37392132174553944,
      "grad_norm": 0.1804330050945282,
      "learning_rate": 1.9356881031545017e-05,
      "loss": 0.6485239863395691,
      "step": 4566
    },
    {
      "epoch": 0.37400321428205835,
      "grad_norm": 0.23091301321983337,
      "learning_rate": 1.9354904411649174e-05,
      "loss": 0.7925410866737366,
      "step": 4567
    },
    {
      "epoch": 0.3740851068185773,
      "grad_norm": 0.19559821486473083,
      "learning_rate": 1.9352927472078898e-05,
      "loss": 0.8500974774360657,
      "step": 4568
    },
    {
      "epoch": 0.3741669993550963,
      "grad_norm": 0.15601348876953125,
      "learning_rate": 1.9350950212920107e-05,
      "loss": 0.9639931321144104,
      "step": 4569
    },
    {
      "epoch": 0.37424889189161525,
      "grad_norm": 0.2301827073097229,
      "learning_rate": 1.9348972634258746e-05,
      "loss": 0.8901113867759705,
      "step": 4570
    },
    {
      "epoch": 0.37433078442813417,
      "grad_norm": 0.20078466832637787,
      "learning_rate": 1.9346994736180763e-05,
      "loss": 0.9731923937797546,
      "step": 4571
    },
    {
      "epoch": 0.37441267696465313,
      "grad_norm": 0.20019513368606567,
      "learning_rate": 1.934501651877213e-05,
      "loss": 0.5382856726646423,
      "step": 4572
    },
    {
      "epoch": 0.3744945695011721,
      "grad_norm": 0.16807831823825836,
      "learning_rate": 1.9343037982118828e-05,
      "loss": 0.8273330926895142,
      "step": 4573
    },
    {
      "epoch": 0.374576462037691,
      "grad_norm": 0.19539517164230347,
      "learning_rate": 1.934105912630685e-05,
      "loss": 1.2040979862213135,
      "step": 4574
    },
    {
      "epoch": 0.37465835457421,
      "grad_norm": 0.17991510033607483,
      "learning_rate": 1.9339079951422208e-05,
      "loss": 0.7164191007614136,
      "step": 4575
    },
    {
      "epoch": 0.37474024711072895,
      "grad_norm": 0.21070437133312225,
      "learning_rate": 1.9337100457550927e-05,
      "loss": 0.8691115379333496,
      "step": 4576
    },
    {
      "epoch": 0.3748221396472479,
      "grad_norm": 0.17130343616008759,
      "learning_rate": 1.933512064477904e-05,
      "loss": 0.6696967482566833,
      "step": 4577
    },
    {
      "epoch": 0.37490403218376683,
      "grad_norm": 0.21378833055496216,
      "learning_rate": 1.933314051319259e-05,
      "loss": 0.6832566857337952,
      "step": 4578
    },
    {
      "epoch": 0.3749859247202858,
      "grad_norm": 0.15101757645606995,
      "learning_rate": 1.9331160062877657e-05,
      "loss": 0.6452093720436096,
      "step": 4579
    },
    {
      "epoch": 0.37506781725680477,
      "grad_norm": 0.22484837472438812,
      "learning_rate": 1.932917929392031e-05,
      "loss": 1.0838408470153809,
      "step": 4580
    },
    {
      "epoch": 0.37514970979332374,
      "grad_norm": 0.20229625701904297,
      "learning_rate": 1.932719820640665e-05,
      "loss": 1.0439257621765137,
      "step": 4581
    },
    {
      "epoch": 0.37523160232984265,
      "grad_norm": 0.18673290312290192,
      "learning_rate": 1.9325216800422774e-05,
      "loss": 0.7688818573951721,
      "step": 4582
    },
    {
      "epoch": 0.3753134948663616,
      "grad_norm": 0.1581125259399414,
      "learning_rate": 1.9323235076054804e-05,
      "loss": 0.7316241264343262,
      "step": 4583
    },
    {
      "epoch": 0.3753953874028806,
      "grad_norm": 0.2093316614627838,
      "learning_rate": 1.9321253033388876e-05,
      "loss": 0.7076972723007202,
      "step": 4584
    },
    {
      "epoch": 0.3754772799393995,
      "grad_norm": 0.18392010033130646,
      "learning_rate": 1.9319270672511137e-05,
      "loss": 0.8832443952560425,
      "step": 4585
    },
    {
      "epoch": 0.37555917247591847,
      "grad_norm": 0.16416677832603455,
      "learning_rate": 1.9317287993507753e-05,
      "loss": 0.6995161771774292,
      "step": 4586
    },
    {
      "epoch": 0.37564106501243744,
      "grad_norm": 0.17237384617328644,
      "learning_rate": 1.9315304996464893e-05,
      "loss": 0.7553730607032776,
      "step": 4587
    },
    {
      "epoch": 0.3757229575489564,
      "grad_norm": 0.18508848547935486,
      "learning_rate": 1.9313321681468748e-05,
      "loss": 0.8050664067268372,
      "step": 4588
    },
    {
      "epoch": 0.3758048500854753,
      "grad_norm": 0.21674981713294983,
      "learning_rate": 1.931133804860552e-05,
      "loss": 0.8104474544525146,
      "step": 4589
    },
    {
      "epoch": 0.3758867426219943,
      "grad_norm": 0.20441006124019623,
      "learning_rate": 1.9309354097961423e-05,
      "loss": 0.9267318248748779,
      "step": 4590
    },
    {
      "epoch": 0.37596863515851325,
      "grad_norm": 0.17830127477645874,
      "learning_rate": 1.93073698296227e-05,
      "loss": 0.8412303924560547,
      "step": 4591
    },
    {
      "epoch": 0.37605052769503217,
      "grad_norm": 0.19757241010665894,
      "learning_rate": 1.9305385243675582e-05,
      "loss": 0.6384515166282654,
      "step": 4592
    },
    {
      "epoch": 0.37613242023155113,
      "grad_norm": 0.21371901035308838,
      "learning_rate": 1.9303400340206334e-05,
      "loss": 0.6727374792098999,
      "step": 4593
    },
    {
      "epoch": 0.3762143127680701,
      "grad_norm": 0.1869383156299591,
      "learning_rate": 1.930141511930123e-05,
      "loss": 1.1401580572128296,
      "step": 4594
    },
    {
      "epoch": 0.37629620530458907,
      "grad_norm": 0.2347451001405716,
      "learning_rate": 1.929942958104655e-05,
      "loss": 0.909294068813324,
      "step": 4595
    },
    {
      "epoch": 0.376378097841108,
      "grad_norm": 0.15483404695987701,
      "learning_rate": 1.9297443725528595e-05,
      "loss": 0.9040797352790833,
      "step": 4596
    },
    {
      "epoch": 0.37645999037762695,
      "grad_norm": 0.21393480896949768,
      "learning_rate": 1.929545755283368e-05,
      "loss": 0.669298529624939,
      "step": 4597
    },
    {
      "epoch": 0.3765418829141459,
      "grad_norm": 0.16979260742664337,
      "learning_rate": 1.9293471063048133e-05,
      "loss": 0.9506065845489502,
      "step": 4598
    },
    {
      "epoch": 0.3766237754506649,
      "grad_norm": 0.21195824444293976,
      "learning_rate": 1.9291484256258295e-05,
      "loss": 0.9584797620773315,
      "step": 4599
    },
    {
      "epoch": 0.3767056679871838,
      "grad_norm": 0.1855274736881256,
      "learning_rate": 1.9289497132550516e-05,
      "loss": 0.8431603908538818,
      "step": 4600
    },
    {
      "epoch": 0.37678756052370277,
      "grad_norm": 0.1596568524837494,
      "learning_rate": 1.928750969201117e-05,
      "loss": 0.7100690603256226,
      "step": 4601
    },
    {
      "epoch": 0.37686945306022174,
      "grad_norm": 0.15324713289737701,
      "learning_rate": 1.928552193472664e-05,
      "loss": 0.7743615508079529,
      "step": 4602
    },
    {
      "epoch": 0.37695134559674065,
      "grad_norm": 0.16985802352428436,
      "learning_rate": 1.9283533860783317e-05,
      "loss": 0.7185080051422119,
      "step": 4603
    },
    {
      "epoch": 0.3770332381332596,
      "grad_norm": 0.2017555981874466,
      "learning_rate": 1.9281545470267615e-05,
      "loss": 0.7175133228302002,
      "step": 4604
    },
    {
      "epoch": 0.3771151306697786,
      "grad_norm": 0.13697998225688934,
      "learning_rate": 1.9279556763265956e-05,
      "loss": 0.5350152254104614,
      "step": 4605
    },
    {
      "epoch": 0.37719702320629755,
      "grad_norm": 0.20710907876491547,
      "learning_rate": 1.927756773986478e-05,
      "loss": 1.0026438236236572,
      "step": 4606
    },
    {
      "epoch": 0.37727891574281647,
      "grad_norm": 0.16106805205345154,
      "learning_rate": 1.9275578400150537e-05,
      "loss": 0.9975225329399109,
      "step": 4607
    },
    {
      "epoch": 0.37736080827933544,
      "grad_norm": 0.16031181812286377,
      "learning_rate": 1.927358874420969e-05,
      "loss": 0.6076956391334534,
      "step": 4608
    },
    {
      "epoch": 0.3774427008158544,
      "grad_norm": 0.18430650234222412,
      "learning_rate": 1.927159877212871e-05,
      "loss": 0.7946627140045166,
      "step": 4609
    },
    {
      "epoch": 0.37752459335237337,
      "grad_norm": 0.17545609176158905,
      "learning_rate": 1.926960848399411e-05,
      "loss": 0.6843754649162292,
      "step": 4610
    },
    {
      "epoch": 0.3776064858888923,
      "grad_norm": 0.19993361830711365,
      "learning_rate": 1.9267617879892376e-05,
      "loss": 0.6761384010314941,
      "step": 4611
    },
    {
      "epoch": 0.37768837842541125,
      "grad_norm": 0.18578436970710754,
      "learning_rate": 1.9265626959910037e-05,
      "loss": 0.8283423185348511,
      "step": 4612
    },
    {
      "epoch": 0.3777702709619302,
      "grad_norm": 0.19191323220729828,
      "learning_rate": 1.926363572413363e-05,
      "loss": 0.828531801700592,
      "step": 4613
    },
    {
      "epoch": 0.37785216349844913,
      "grad_norm": 0.19362658262252808,
      "learning_rate": 1.926164417264969e-05,
      "loss": 0.5216598510742188,
      "step": 4614
    },
    {
      "epoch": 0.3779340560349681,
      "grad_norm": 0.18186156451702118,
      "learning_rate": 1.925965230554479e-05,
      "loss": 0.49163955450057983,
      "step": 4615
    },
    {
      "epoch": 0.37801594857148707,
      "grad_norm": 0.1928747296333313,
      "learning_rate": 1.92576601229055e-05,
      "loss": 1.0166362524032593,
      "step": 4616
    },
    {
      "epoch": 0.37809784110800604,
      "grad_norm": 0.17852579057216644,
      "learning_rate": 1.925566762481841e-05,
      "loss": 0.7471522688865662,
      "step": 4617
    },
    {
      "epoch": 0.37817973364452495,
      "grad_norm": 0.21047402918338776,
      "learning_rate": 1.9253674811370123e-05,
      "loss": 0.6857478022575378,
      "step": 4618
    },
    {
      "epoch": 0.3782616261810439,
      "grad_norm": 0.17051324248313904,
      "learning_rate": 1.925168168264725e-05,
      "loss": 0.6565566062927246,
      "step": 4619
    },
    {
      "epoch": 0.3783435187175629,
      "grad_norm": 0.17616161704063416,
      "learning_rate": 1.9249688238736423e-05,
      "loss": 0.7236255407333374,
      "step": 4620
    },
    {
      "epoch": 0.37842541125408186,
      "grad_norm": 0.178658127784729,
      "learning_rate": 1.9247694479724288e-05,
      "loss": 0.8734599351882935,
      "step": 4621
    },
    {
      "epoch": 0.37850730379060077,
      "grad_norm": 0.17532174289226532,
      "learning_rate": 1.92457004056975e-05,
      "loss": 0.665255069732666,
      "step": 4622
    },
    {
      "epoch": 0.37858919632711974,
      "grad_norm": 0.20607908070087433,
      "learning_rate": 1.9243706016742728e-05,
      "loss": 0.9591212272644043,
      "step": 4623
    },
    {
      "epoch": 0.3786710888636387,
      "grad_norm": 0.2087327390909195,
      "learning_rate": 1.924171131294666e-05,
      "loss": 0.8994796276092529,
      "step": 4624
    },
    {
      "epoch": 0.3787529814001576,
      "grad_norm": 0.17072175443172455,
      "learning_rate": 1.923971629439599e-05,
      "loss": 0.6543944478034973,
      "step": 4625
    },
    {
      "epoch": 0.3788348739366766,
      "grad_norm": 0.22963057458400726,
      "learning_rate": 1.923772096117743e-05,
      "loss": 0.9494592547416687,
      "step": 4626
    },
    {
      "epoch": 0.37891676647319555,
      "grad_norm": 0.179121732711792,
      "learning_rate": 1.923572531337771e-05,
      "loss": 1.0978577136993408,
      "step": 4627
    },
    {
      "epoch": 0.3789986590097145,
      "grad_norm": 0.2187269777059555,
      "learning_rate": 1.9233729351083564e-05,
      "loss": 1.0316312313079834,
      "step": 4628
    },
    {
      "epoch": 0.37908055154623344,
      "grad_norm": 0.22152844071388245,
      "learning_rate": 1.923173307438175e-05,
      "loss": 0.7307385206222534,
      "step": 4629
    },
    {
      "epoch": 0.3791624440827524,
      "grad_norm": 0.16824860870838165,
      "learning_rate": 1.9229736483359028e-05,
      "loss": 0.7215285301208496,
      "step": 4630
    },
    {
      "epoch": 0.37924433661927137,
      "grad_norm": 0.19458675384521484,
      "learning_rate": 1.922773957810218e-05,
      "loss": 0.9631766676902771,
      "step": 4631
    },
    {
      "epoch": 0.37932622915579034,
      "grad_norm": 0.18557296693325043,
      "learning_rate": 1.922574235869801e-05,
      "loss": 0.732447624206543,
      "step": 4632
    },
    {
      "epoch": 0.37940812169230925,
      "grad_norm": 0.18249297142028809,
      "learning_rate": 1.9223744825233305e-05,
      "loss": 1.072929859161377,
      "step": 4633
    },
    {
      "epoch": 0.3794900142288282,
      "grad_norm": 0.23205098509788513,
      "learning_rate": 1.9221746977794904e-05,
      "loss": 0.9405452013015747,
      "step": 4634
    },
    {
      "epoch": 0.3795719067653472,
      "grad_norm": 0.1696392148733139,
      "learning_rate": 1.9219748816469633e-05,
      "loss": 0.612343430519104,
      "step": 4635
    },
    {
      "epoch": 0.3796537993018661,
      "grad_norm": 0.15711471438407898,
      "learning_rate": 1.9217750341344345e-05,
      "loss": 0.7110911011695862,
      "step": 4636
    },
    {
      "epoch": 0.37973569183838507,
      "grad_norm": 0.1825173795223236,
      "learning_rate": 1.9215751552505896e-05,
      "loss": 0.9842389225959778,
      "step": 4637
    },
    {
      "epoch": 0.37981758437490404,
      "grad_norm": 0.19044475257396698,
      "learning_rate": 1.9213752450041166e-05,
      "loss": 0.8584103584289551,
      "step": 4638
    },
    {
      "epoch": 0.379899476911423,
      "grad_norm": 0.15955668687820435,
      "learning_rate": 1.9211753034037046e-05,
      "loss": 0.6860961318016052,
      "step": 4639
    },
    {
      "epoch": 0.3799813694479419,
      "grad_norm": 0.1818350851535797,
      "learning_rate": 1.9209753304580432e-05,
      "loss": 0.6002848148345947,
      "step": 4640
    },
    {
      "epoch": 0.3800632619844609,
      "grad_norm": 0.23158234357833862,
      "learning_rate": 1.9207753261758247e-05,
      "loss": 1.1487622261047363,
      "step": 4641
    },
    {
      "epoch": 0.38014515452097986,
      "grad_norm": 0.18547779321670532,
      "learning_rate": 1.920575290565742e-05,
      "loss": 0.837938129901886,
      "step": 4642
    },
    {
      "epoch": 0.3802270470574988,
      "grad_norm": 0.20087409019470215,
      "learning_rate": 1.920375223636489e-05,
      "loss": 0.8864272236824036,
      "step": 4643
    },
    {
      "epoch": 0.38030893959401774,
      "grad_norm": 0.19217292964458466,
      "learning_rate": 1.920175125396762e-05,
      "loss": 0.9743700623512268,
      "step": 4644
    },
    {
      "epoch": 0.3803908321305367,
      "grad_norm": 0.17482365667819977,
      "learning_rate": 1.9199749958552576e-05,
      "loss": 0.7312459945678711,
      "step": 4645
    },
    {
      "epoch": 0.3804727246670557,
      "grad_norm": 0.19659122824668884,
      "learning_rate": 1.919774835020675e-05,
      "loss": 0.7501091957092285,
      "step": 4646
    },
    {
      "epoch": 0.3805546172035746,
      "grad_norm": 0.2303171157836914,
      "learning_rate": 1.9195746429017132e-05,
      "loss": 1.1049586534500122,
      "step": 4647
    },
    {
      "epoch": 0.38063650974009355,
      "grad_norm": 0.17912191152572632,
      "learning_rate": 1.919374419507074e-05,
      "loss": 0.6308512687683105,
      "step": 4648
    },
    {
      "epoch": 0.3807184022766125,
      "grad_norm": 0.1980789303779602,
      "learning_rate": 1.919174164845459e-05,
      "loss": 0.9254670143127441,
      "step": 4649
    },
    {
      "epoch": 0.3808002948131315,
      "grad_norm": 0.18950803577899933,
      "learning_rate": 1.9189738789255733e-05,
      "loss": 0.9841037392616272,
      "step": 4650
    },
    {
      "epoch": 0.3808821873496504,
      "grad_norm": 0.1781456172466278,
      "learning_rate": 1.9187735617561217e-05,
      "loss": 0.8448410034179688,
      "step": 4651
    },
    {
      "epoch": 0.38096407988616937,
      "grad_norm": 0.17821036279201508,
      "learning_rate": 1.9185732133458104e-05,
      "loss": 0.9637575149536133,
      "step": 4652
    },
    {
      "epoch": 0.38104597242268834,
      "grad_norm": 0.14005641639232635,
      "learning_rate": 1.9183728337033475e-05,
      "loss": 0.9429913759231567,
      "step": 4653
    },
    {
      "epoch": 0.3811278649592073,
      "grad_norm": 0.2177179902791977,
      "learning_rate": 1.9181724228374427e-05,
      "loss": 0.7686384916305542,
      "step": 4654
    },
    {
      "epoch": 0.3812097574957262,
      "grad_norm": 0.20252060890197754,
      "learning_rate": 1.9179719807568063e-05,
      "loss": 1.2005079984664917,
      "step": 4655
    },
    {
      "epoch": 0.3812916500322452,
      "grad_norm": 0.18166102468967438,
      "learning_rate": 1.9177715074701508e-05,
      "loss": 0.7227151989936829,
      "step": 4656
    },
    {
      "epoch": 0.38137354256876416,
      "grad_norm": 0.16207797825336456,
      "learning_rate": 1.917571002986189e-05,
      "loss": 1.0884654521942139,
      "step": 4657
    },
    {
      "epoch": 0.38145543510528307,
      "grad_norm": 0.15721853077411652,
      "learning_rate": 1.917370467313636e-05,
      "loss": 0.9087743759155273,
      "step": 4658
    },
    {
      "epoch": 0.38153732764180204,
      "grad_norm": 0.16655312478542328,
      "learning_rate": 1.9171699004612083e-05,
      "loss": 0.7234389781951904,
      "step": 4659
    },
    {
      "epoch": 0.381619220178321,
      "grad_norm": 0.19969238340854645,
      "learning_rate": 1.9169693024376224e-05,
      "loss": 0.7969765663146973,
      "step": 4660
    },
    {
      "epoch": 0.38170111271484,
      "grad_norm": 0.15778997540473938,
      "learning_rate": 1.9167686732515978e-05,
      "loss": 0.846939742565155,
      "step": 4661
    },
    {
      "epoch": 0.3817830052513589,
      "grad_norm": 0.1712370365858078,
      "learning_rate": 1.9165680129118545e-05,
      "loss": 0.7386664748191833,
      "step": 4662
    },
    {
      "epoch": 0.38186489778787785,
      "grad_norm": 0.17935208976268768,
      "learning_rate": 1.9163673214271143e-05,
      "loss": 0.8220565319061279,
      "step": 4663
    },
    {
      "epoch": 0.3819467903243968,
      "grad_norm": 0.22253113985061646,
      "learning_rate": 1.9161665988060996e-05,
      "loss": 0.7763327956199646,
      "step": 4664
    },
    {
      "epoch": 0.38202868286091574,
      "grad_norm": 0.2299795299768448,
      "learning_rate": 1.915965845057535e-05,
      "loss": 0.9015088081359863,
      "step": 4665
    },
    {
      "epoch": 0.3821105753974347,
      "grad_norm": 0.17065173387527466,
      "learning_rate": 1.9157650601901454e-05,
      "loss": 0.8723979592323303,
      "step": 4666
    },
    {
      "epoch": 0.3821924679339537,
      "grad_norm": 0.1794518083333969,
      "learning_rate": 1.9155642442126585e-05,
      "loss": 1.0887919664382935,
      "step": 4667
    },
    {
      "epoch": 0.38227436047047264,
      "grad_norm": 0.41670772433280945,
      "learning_rate": 1.9153633971338027e-05,
      "loss": 0.6485389471054077,
      "step": 4668
    },
    {
      "epoch": 0.38235625300699155,
      "grad_norm": 0.1808953881263733,
      "learning_rate": 1.915162518962307e-05,
      "loss": 0.7988380789756775,
      "step": 4669
    },
    {
      "epoch": 0.3824381455435105,
      "grad_norm": 0.18424618244171143,
      "learning_rate": 1.9149616097069028e-05,
      "loss": 1.096545934677124,
      "step": 4670
    },
    {
      "epoch": 0.3825200380800295,
      "grad_norm": 0.19670920073986053,
      "learning_rate": 1.9147606693763222e-05,
      "loss": 0.7661637663841248,
      "step": 4671
    },
    {
      "epoch": 0.38260193061654846,
      "grad_norm": 0.1931157261133194,
      "learning_rate": 1.914559697979299e-05,
      "loss": 1.1491210460662842,
      "step": 4672
    },
    {
      "epoch": 0.38268382315306737,
      "grad_norm": 0.20335866510868073,
      "learning_rate": 1.9143586955245685e-05,
      "loss": 0.9681087136268616,
      "step": 4673
    },
    {
      "epoch": 0.38276571568958634,
      "grad_norm": 0.18682818114757538,
      "learning_rate": 1.9141576620208665e-05,
      "loss": 0.642471194267273,
      "step": 4674
    },
    {
      "epoch": 0.3828476082261053,
      "grad_norm": 0.22111478447914124,
      "learning_rate": 1.9139565974769314e-05,
      "loss": 1.2741645574569702,
      "step": 4675
    },
    {
      "epoch": 0.3829295007626242,
      "grad_norm": 0.19419409334659576,
      "learning_rate": 1.913755501901502e-05,
      "loss": 0.922049343585968,
      "step": 4676
    },
    {
      "epoch": 0.3830113932991432,
      "grad_norm": 0.15282124280929565,
      "learning_rate": 1.9135543753033186e-05,
      "loss": 0.9016849398612976,
      "step": 4677
    },
    {
      "epoch": 0.38309328583566216,
      "grad_norm": 0.19066882133483887,
      "learning_rate": 1.913353217691123e-05,
      "loss": 0.6194227337837219,
      "step": 4678
    },
    {
      "epoch": 0.3831751783721811,
      "grad_norm": 0.18405954539775848,
      "learning_rate": 1.913152029073659e-05,
      "loss": 0.8733087182044983,
      "step": 4679
    },
    {
      "epoch": 0.38325707090870004,
      "grad_norm": 0.2081102877855301,
      "learning_rate": 1.91295080945967e-05,
      "loss": 1.0447919368743896,
      "step": 4680
    },
    {
      "epoch": 0.383338963445219,
      "grad_norm": 0.1806064248085022,
      "learning_rate": 1.912749558857903e-05,
      "loss": 0.8776271343231201,
      "step": 4681
    },
    {
      "epoch": 0.383420855981738,
      "grad_norm": 0.1844988465309143,
      "learning_rate": 1.912548277277104e-05,
      "loss": 0.9014125466346741,
      "step": 4682
    },
    {
      "epoch": 0.38350274851825694,
      "grad_norm": 0.157343789935112,
      "learning_rate": 1.9123469647260223e-05,
      "loss": 0.8187368512153625,
      "step": 4683
    },
    {
      "epoch": 0.38358464105477585,
      "grad_norm": 0.2054157257080078,
      "learning_rate": 1.9121456212134075e-05,
      "loss": 0.9915117025375366,
      "step": 4684
    },
    {
      "epoch": 0.3836665335912948,
      "grad_norm": 0.17829643189907074,
      "learning_rate": 1.9119442467480116e-05,
      "loss": 0.9117102026939392,
      "step": 4685
    },
    {
      "epoch": 0.3837484261278138,
      "grad_norm": 0.18098647892475128,
      "learning_rate": 1.9117428413385855e-05,
      "loss": 1.1551471948623657,
      "step": 4686
    },
    {
      "epoch": 0.3838303186643327,
      "grad_norm": 0.16925470530986786,
      "learning_rate": 1.911541404993884e-05,
      "loss": 0.6812540292739868,
      "step": 4687
    },
    {
      "epoch": 0.38391221120085167,
      "grad_norm": 0.23048144578933716,
      "learning_rate": 1.9113399377226634e-05,
      "loss": 1.3978283405303955,
      "step": 4688
    },
    {
      "epoch": 0.38399410373737064,
      "grad_norm": 0.2023683488368988,
      "learning_rate": 1.9111384395336783e-05,
      "loss": 0.897840142250061,
      "step": 4689
    },
    {
      "epoch": 0.3840759962738896,
      "grad_norm": 0.16986438632011414,
      "learning_rate": 1.9109369104356884e-05,
      "loss": 0.540365993976593,
      "step": 4690
    },
    {
      "epoch": 0.3841578888104085,
      "grad_norm": 0.2009609192609787,
      "learning_rate": 1.9107353504374517e-05,
      "loss": 1.0243202447891235,
      "step": 4691
    },
    {
      "epoch": 0.3842397813469275,
      "grad_norm": 0.1699782907962799,
      "learning_rate": 1.9105337595477297e-05,
      "loss": 0.7374356389045715,
      "step": 4692
    },
    {
      "epoch": 0.38432167388344646,
      "grad_norm": 0.17886433005332947,
      "learning_rate": 1.9103321377752837e-05,
      "loss": 0.9126213788986206,
      "step": 4693
    },
    {
      "epoch": 0.3844035664199654,
      "grad_norm": 0.17297737300395966,
      "learning_rate": 1.9101304851288777e-05,
      "loss": 0.576627254486084,
      "step": 4694
    },
    {
      "epoch": 0.38448545895648434,
      "grad_norm": 0.20151056349277496,
      "learning_rate": 1.9099288016172758e-05,
      "loss": 0.6510732173919678,
      "step": 4695
    },
    {
      "epoch": 0.3845673514930033,
      "grad_norm": 0.18612340092658997,
      "learning_rate": 1.9097270872492446e-05,
      "loss": 0.8395634889602661,
      "step": 4696
    },
    {
      "epoch": 0.3846492440295223,
      "grad_norm": 0.1726856529712677,
      "learning_rate": 1.9095253420335504e-05,
      "loss": 0.7868666648864746,
      "step": 4697
    },
    {
      "epoch": 0.3847311365660412,
      "grad_norm": 0.17768368124961853,
      "learning_rate": 1.9093235659789627e-05,
      "loss": 0.7823092341423035,
      "step": 4698
    },
    {
      "epoch": 0.38481302910256016,
      "grad_norm": 0.21562795341014862,
      "learning_rate": 1.9091217590942518e-05,
      "loss": 0.8404867053031921,
      "step": 4699
    },
    {
      "epoch": 0.3848949216390791,
      "grad_norm": 0.6321312785148621,
      "learning_rate": 1.908919921388188e-05,
      "loss": 1.3623120784759521,
      "step": 4700
    },
    {
      "epoch": 0.3849768141755981,
      "grad_norm": 0.19024516642093658,
      "learning_rate": 1.908718052869545e-05,
      "loss": 0.3984249532222748,
      "step": 4701
    },
    {
      "epoch": 0.385058706712117,
      "grad_norm": 0.17382001876831055,
      "learning_rate": 1.908516153547096e-05,
      "loss": 0.777703583240509,
      "step": 4702
    },
    {
      "epoch": 0.385140599248636,
      "grad_norm": 0.16616584360599518,
      "learning_rate": 1.9083142234296173e-05,
      "loss": 0.7914732098579407,
      "step": 4703
    },
    {
      "epoch": 0.38522249178515494,
      "grad_norm": 0.18576990067958832,
      "learning_rate": 1.9081122625258847e-05,
      "loss": 0.7434723377227783,
      "step": 4704
    },
    {
      "epoch": 0.3853043843216739,
      "grad_norm": 0.1827581822872162,
      "learning_rate": 1.907910270844677e-05,
      "loss": 0.790871798992157,
      "step": 4705
    },
    {
      "epoch": 0.3853862768581928,
      "grad_norm": 0.25097396969795227,
      "learning_rate": 1.9077082483947732e-05,
      "loss": 0.5377322435379028,
      "step": 4706
    },
    {
      "epoch": 0.3854681693947118,
      "grad_norm": 0.18662287294864655,
      "learning_rate": 1.907506195184954e-05,
      "loss": 0.8455860018730164,
      "step": 4707
    },
    {
      "epoch": 0.38555006193123076,
      "grad_norm": 0.1949547678232193,
      "learning_rate": 1.9073041112240015e-05,
      "loss": 0.7009474635124207,
      "step": 4708
    },
    {
      "epoch": 0.38563195446774967,
      "grad_norm": 0.2103237807750702,
      "learning_rate": 1.907101996520699e-05,
      "loss": 1.0106903314590454,
      "step": 4709
    },
    {
      "epoch": 0.38571384700426864,
      "grad_norm": 0.2117520123720169,
      "learning_rate": 1.9068998510838318e-05,
      "loss": 1.0766915082931519,
      "step": 4710
    },
    {
      "epoch": 0.3857957395407876,
      "grad_norm": 0.20286214351654053,
      "learning_rate": 1.9066976749221855e-05,
      "loss": 0.6734803915023804,
      "step": 4711
    },
    {
      "epoch": 0.3858776320773066,
      "grad_norm": 0.16865183413028717,
      "learning_rate": 1.9064954680445472e-05,
      "loss": 0.651240885257721,
      "step": 4712
    },
    {
      "epoch": 0.3859595246138255,
      "grad_norm": 0.1968625783920288,
      "learning_rate": 1.9062932304597064e-05,
      "loss": 0.8508142232894897,
      "step": 4713
    },
    {
      "epoch": 0.38604141715034446,
      "grad_norm": 0.1921038031578064,
      "learning_rate": 1.9060909621764525e-05,
      "loss": 1.035552978515625,
      "step": 4714
    },
    {
      "epoch": 0.3861233096868634,
      "grad_norm": 0.18270500004291534,
      "learning_rate": 1.9058886632035775e-05,
      "loss": 0.8478269577026367,
      "step": 4715
    },
    {
      "epoch": 0.3862052022233824,
      "grad_norm": 0.18830478191375732,
      "learning_rate": 1.9056863335498738e-05,
      "loss": 0.8543262481689453,
      "step": 4716
    },
    {
      "epoch": 0.3862870947599013,
      "grad_norm": 0.1982220858335495,
      "learning_rate": 1.905483973224135e-05,
      "loss": 0.832301914691925,
      "step": 4717
    },
    {
      "epoch": 0.3863689872964203,
      "grad_norm": 0.17266348004341125,
      "learning_rate": 1.905281582235158e-05,
      "loss": 0.6463779807090759,
      "step": 4718
    },
    {
      "epoch": 0.38645087983293924,
      "grad_norm": 0.18379566073417664,
      "learning_rate": 1.905079160591738e-05,
      "loss": 0.953780472278595,
      "step": 4719
    },
    {
      "epoch": 0.38653277236945816,
      "grad_norm": 0.17711329460144043,
      "learning_rate": 1.9048767083026738e-05,
      "loss": 0.8314529061317444,
      "step": 4720
    },
    {
      "epoch": 0.3866146649059771,
      "grad_norm": 0.16410169005393982,
      "learning_rate": 1.9046742253767646e-05,
      "loss": 0.8281599283218384,
      "step": 4721
    },
    {
      "epoch": 0.3866965574424961,
      "grad_norm": 0.17608238756656647,
      "learning_rate": 1.9044717118228114e-05,
      "loss": 0.8168250322341919,
      "step": 4722
    },
    {
      "epoch": 0.38677844997901506,
      "grad_norm": 0.20266146957874298,
      "learning_rate": 1.904269167649616e-05,
      "loss": 0.751883864402771,
      "step": 4723
    },
    {
      "epoch": 0.386860342515534,
      "grad_norm": 0.17009034752845764,
      "learning_rate": 1.9040665928659818e-05,
      "loss": 0.5345026254653931,
      "step": 4724
    },
    {
      "epoch": 0.38694223505205294,
      "grad_norm": 0.19455698132514954,
      "learning_rate": 1.9038639874807135e-05,
      "loss": 0.8817864656448364,
      "step": 4725
    },
    {
      "epoch": 0.3870241275885719,
      "grad_norm": 0.18844927847385406,
      "learning_rate": 1.903661351502618e-05,
      "loss": 1.080299973487854,
      "step": 4726
    },
    {
      "epoch": 0.3871060201250909,
      "grad_norm": 0.18124818801879883,
      "learning_rate": 1.903458684940501e-05,
      "loss": 1.2788214683532715,
      "step": 4727
    },
    {
      "epoch": 0.3871879126616098,
      "grad_norm": 0.17130951583385468,
      "learning_rate": 1.9032559878031727e-05,
      "loss": 0.8856240510940552,
      "step": 4728
    },
    {
      "epoch": 0.38726980519812876,
      "grad_norm": 0.17440466582775116,
      "learning_rate": 1.9030532600994428e-05,
      "loss": 1.0971019268035889,
      "step": 4729
    },
    {
      "epoch": 0.3873516977346477,
      "grad_norm": 0.1842314451932907,
      "learning_rate": 1.9028505018381224e-05,
      "loss": 0.704480767250061,
      "step": 4730
    },
    {
      "epoch": 0.38743359027116664,
      "grad_norm": 0.17579196393489838,
      "learning_rate": 1.9026477130280246e-05,
      "loss": 0.9885772466659546,
      "step": 4731
    },
    {
      "epoch": 0.3875154828076856,
      "grad_norm": 0.180941641330719,
      "learning_rate": 1.9024448936779627e-05,
      "loss": 0.8051034212112427,
      "step": 4732
    },
    {
      "epoch": 0.3875973753442046,
      "grad_norm": 0.15753179788589478,
      "learning_rate": 1.902242043796753e-05,
      "loss": 0.557755708694458,
      "step": 4733
    },
    {
      "epoch": 0.38767926788072354,
      "grad_norm": 0.2828572392463684,
      "learning_rate": 1.9020391633932116e-05,
      "loss": 0.7578437924385071,
      "step": 4734
    },
    {
      "epoch": 0.38776116041724246,
      "grad_norm": 0.24470220506191254,
      "learning_rate": 1.9018362524761566e-05,
      "loss": 1.21573805809021,
      "step": 4735
    },
    {
      "epoch": 0.3878430529537614,
      "grad_norm": 0.19364018738269806,
      "learning_rate": 1.9016333110544078e-05,
      "loss": 0.7295354604721069,
      "step": 4736
    },
    {
      "epoch": 0.3879249454902804,
      "grad_norm": 0.19651104509830475,
      "learning_rate": 1.901430339136785e-05,
      "loss": 0.8116263151168823,
      "step": 4737
    },
    {
      "epoch": 0.3880068380267993,
      "grad_norm": 0.17374886572360992,
      "learning_rate": 1.901227336732111e-05,
      "loss": 0.978779137134552,
      "step": 4738
    },
    {
      "epoch": 0.3880887305633183,
      "grad_norm": 0.1617298573255539,
      "learning_rate": 1.9010243038492088e-05,
      "loss": 0.7317852973937988,
      "step": 4739
    },
    {
      "epoch": 0.38817062309983724,
      "grad_norm": 0.18309368193149567,
      "learning_rate": 1.900821240496903e-05,
      "loss": 0.9934594035148621,
      "step": 4740
    },
    {
      "epoch": 0.3882525156363562,
      "grad_norm": 0.1813001185655594,
      "learning_rate": 1.9006181466840192e-05,
      "loss": 0.9230425953865051,
      "step": 4741
    },
    {
      "epoch": 0.3883344081728751,
      "grad_norm": 0.15505822002887726,
      "learning_rate": 1.900415022419386e-05,
      "loss": 0.7632986307144165,
      "step": 4742
    },
    {
      "epoch": 0.3884163007093941,
      "grad_norm": 0.1734817624092102,
      "learning_rate": 1.9002118677118302e-05,
      "loss": 0.8062170147895813,
      "step": 4743
    },
    {
      "epoch": 0.38849819324591306,
      "grad_norm": 0.18184848129749298,
      "learning_rate": 1.900008682570183e-05,
      "loss": 0.8907952308654785,
      "step": 4744
    },
    {
      "epoch": 0.388580085782432,
      "grad_norm": 0.1758464127779007,
      "learning_rate": 1.8998054670032756e-05,
      "loss": 0.7911707758903503,
      "step": 4745
    },
    {
      "epoch": 0.38866197831895094,
      "grad_norm": 0.19586613774299622,
      "learning_rate": 1.89960222101994e-05,
      "loss": 1.026349425315857,
      "step": 4746
    },
    {
      "epoch": 0.3887438708554699,
      "grad_norm": 0.17799249291419983,
      "learning_rate": 1.8993989446290104e-05,
      "loss": 0.5929654240608215,
      "step": 4747
    },
    {
      "epoch": 0.3888257633919889,
      "grad_norm": 0.1711270809173584,
      "learning_rate": 1.8991956378393223e-05,
      "loss": 0.7344236969947815,
      "step": 4748
    },
    {
      "epoch": 0.3889076559285078,
      "grad_norm": 0.19303074479103088,
      "learning_rate": 1.898992300659712e-05,
      "loss": 0.6305198073387146,
      "step": 4749
    },
    {
      "epoch": 0.38898954846502676,
      "grad_norm": 0.20798610150814056,
      "learning_rate": 1.8987889330990174e-05,
      "loss": 0.8729308247566223,
      "step": 4750
    },
    {
      "epoch": 0.3890714410015457,
      "grad_norm": 0.21643230319023132,
      "learning_rate": 1.8985855351660776e-05,
      "loss": 0.9534658789634705,
      "step": 4751
    },
    {
      "epoch": 0.3891533335380647,
      "grad_norm": 0.1691722720861435,
      "learning_rate": 1.8983821068697327e-05,
      "loss": 0.7010846734046936,
      "step": 4752
    },
    {
      "epoch": 0.3892352260745836,
      "grad_norm": 0.24139350652694702,
      "learning_rate": 1.8981786482188257e-05,
      "loss": 0.8685301542282104,
      "step": 4753
    },
    {
      "epoch": 0.3893171186111026,
      "grad_norm": 0.20303493738174438,
      "learning_rate": 1.8979751592221988e-05,
      "loss": 0.9798063039779663,
      "step": 4754
    },
    {
      "epoch": 0.38939901114762154,
      "grad_norm": 0.1991591602563858,
      "learning_rate": 1.8977716398886965e-05,
      "loss": 0.816617488861084,
      "step": 4755
    },
    {
      "epoch": 0.3894809036841405,
      "grad_norm": 0.16839754581451416,
      "learning_rate": 1.8975680902271653e-05,
      "loss": 0.70870041847229,
      "step": 4756
    },
    {
      "epoch": 0.3895627962206594,
      "grad_norm": 0.17853528261184692,
      "learning_rate": 1.8973645102464516e-05,
      "loss": 0.8785358667373657,
      "step": 4757
    },
    {
      "epoch": 0.3896446887571784,
      "grad_norm": 0.18125510215759277,
      "learning_rate": 1.8971608999554038e-05,
      "loss": 0.6509623527526855,
      "step": 4758
    },
    {
      "epoch": 0.38972658129369736,
      "grad_norm": 0.16728413105010986,
      "learning_rate": 1.896957259362872e-05,
      "loss": 0.9230778813362122,
      "step": 4759
    },
    {
      "epoch": 0.3898084738302163,
      "grad_norm": 0.178557351231575,
      "learning_rate": 1.8967535884777074e-05,
      "loss": 0.6155305504798889,
      "step": 4760
    },
    {
      "epoch": 0.38989036636673524,
      "grad_norm": 0.19686660170555115,
      "learning_rate": 1.896549887308762e-05,
      "loss": 0.825183629989624,
      "step": 4761
    },
    {
      "epoch": 0.3899722589032542,
      "grad_norm": 0.2109401822090149,
      "learning_rate": 1.8963461558648894e-05,
      "loss": 0.8403916954994202,
      "step": 4762
    },
    {
      "epoch": 0.3900541514397732,
      "grad_norm": 0.17190447449684143,
      "learning_rate": 1.896142394154945e-05,
      "loss": 0.8841496706008911,
      "step": 4763
    },
    {
      "epoch": 0.3901360439762921,
      "grad_norm": 0.18996299803256989,
      "learning_rate": 1.8959386021877853e-05,
      "loss": 0.8261072039604187,
      "step": 4764
    },
    {
      "epoch": 0.39021793651281106,
      "grad_norm": 0.1672254055738449,
      "learning_rate": 1.895734779972267e-05,
      "loss": 1.0292366743087769,
      "step": 4765
    },
    {
      "epoch": 0.39029982904933,
      "grad_norm": 0.1517544686794281,
      "learning_rate": 1.8955309275172503e-05,
      "loss": 0.5616741180419922,
      "step": 4766
    },
    {
      "epoch": 0.390381721585849,
      "grad_norm": 0.18326613306999207,
      "learning_rate": 1.8953270448315943e-05,
      "loss": 0.7764319181442261,
      "step": 4767
    },
    {
      "epoch": 0.3904636141223679,
      "grad_norm": 0.18155787885189056,
      "learning_rate": 1.895123131924161e-05,
      "loss": 0.9292797446250916,
      "step": 4768
    },
    {
      "epoch": 0.3905455066588869,
      "grad_norm": 0.21552330255508423,
      "learning_rate": 1.8949191888038134e-05,
      "loss": 1.0428134202957153,
      "step": 4769
    },
    {
      "epoch": 0.39062739919540584,
      "grad_norm": 0.19898544251918793,
      "learning_rate": 1.8947152154794162e-05,
      "loss": 0.5966057181358337,
      "step": 4770
    },
    {
      "epoch": 0.39070929173192476,
      "grad_norm": 0.1756182610988617,
      "learning_rate": 1.894511211959834e-05,
      "loss": 0.9428150057792664,
      "step": 4771
    },
    {
      "epoch": 0.3907911842684437,
      "grad_norm": 0.15361586213111877,
      "learning_rate": 1.8943071782539343e-05,
      "loss": 0.9658079743385315,
      "step": 4772
    },
    {
      "epoch": 0.3908730768049627,
      "grad_norm": 0.17854666709899902,
      "learning_rate": 1.894103114370585e-05,
      "loss": 0.7041339874267578,
      "step": 4773
    },
    {
      "epoch": 0.39095496934148166,
      "grad_norm": 0.19075332581996918,
      "learning_rate": 1.893899020318655e-05,
      "loss": 1.0539426803588867,
      "step": 4774
    },
    {
      "epoch": 0.3910368618780006,
      "grad_norm": 0.17244793474674225,
      "learning_rate": 1.893694896107016e-05,
      "loss": 0.6864275932312012,
      "step": 4775
    },
    {
      "epoch": 0.39111875441451954,
      "grad_norm": 0.1847013384103775,
      "learning_rate": 1.89349074174454e-05,
      "loss": 0.798041820526123,
      "step": 4776
    },
    {
      "epoch": 0.3912006469510385,
      "grad_norm": 0.201450377702713,
      "learning_rate": 1.8932865572400997e-05,
      "loss": 0.7214220762252808,
      "step": 4777
    },
    {
      "epoch": 0.3912825394875575,
      "grad_norm": 0.18268105387687683,
      "learning_rate": 1.89308234260257e-05,
      "loss": 1.269468903541565,
      "step": 4778
    },
    {
      "epoch": 0.3913644320240764,
      "grad_norm": 0.1874200701713562,
      "learning_rate": 1.8928780978408275e-05,
      "loss": 1.0264439582824707,
      "step": 4779
    },
    {
      "epoch": 0.39144632456059536,
      "grad_norm": 0.20627617835998535,
      "learning_rate": 1.892673822963749e-05,
      "loss": 0.5321992635726929,
      "step": 4780
    },
    {
      "epoch": 0.39152821709711433,
      "grad_norm": 0.1619804948568344,
      "learning_rate": 1.8924695179802135e-05,
      "loss": 0.5075188279151917,
      "step": 4781
    },
    {
      "epoch": 0.39161010963363324,
      "grad_norm": 0.188926100730896,
      "learning_rate": 1.8922651828991003e-05,
      "loss": 1.024842381477356,
      "step": 4782
    },
    {
      "epoch": 0.3916920021701522,
      "grad_norm": 0.15851546823978424,
      "learning_rate": 1.8920608177292908e-05,
      "loss": 0.6178660988807678,
      "step": 4783
    },
    {
      "epoch": 0.3917738947066712,
      "grad_norm": 0.18076932430267334,
      "learning_rate": 1.8918564224796686e-05,
      "loss": 0.6195169687271118,
      "step": 4784
    },
    {
      "epoch": 0.39185578724319015,
      "grad_norm": 0.22339071333408356,
      "learning_rate": 1.8916519971591158e-05,
      "loss": 0.7771727442741394,
      "step": 4785
    },
    {
      "epoch": 0.39193767977970906,
      "grad_norm": 0.26517030596733093,
      "learning_rate": 1.8914475417765193e-05,
      "loss": 0.6652868986129761,
      "step": 4786
    },
    {
      "epoch": 0.392019572316228,
      "grad_norm": 0.16020995378494263,
      "learning_rate": 1.8912430563407642e-05,
      "loss": 0.8639443516731262,
      "step": 4787
    },
    {
      "epoch": 0.392101464852747,
      "grad_norm": 0.15542225539684296,
      "learning_rate": 1.891038540860739e-05,
      "loss": 0.8990994691848755,
      "step": 4788
    },
    {
      "epoch": 0.39218335738926596,
      "grad_norm": 0.1965998113155365,
      "learning_rate": 1.8908339953453326e-05,
      "loss": 0.6839345097541809,
      "step": 4789
    },
    {
      "epoch": 0.3922652499257849,
      "grad_norm": 0.19546347856521606,
      "learning_rate": 1.8906294198034354e-05,
      "loss": 0.8501741290092468,
      "step": 4790
    },
    {
      "epoch": 0.39234714246230384,
      "grad_norm": 0.24416853487491608,
      "learning_rate": 1.8904248142439394e-05,
      "loss": 1.0741746425628662,
      "step": 4791
    },
    {
      "epoch": 0.3924290349988228,
      "grad_norm": 0.2053661048412323,
      "learning_rate": 1.890220178675737e-05,
      "loss": 0.8482263684272766,
      "step": 4792
    },
    {
      "epoch": 0.3925109275353417,
      "grad_norm": 0.1658380627632141,
      "learning_rate": 1.890015513107723e-05,
      "loss": 1.1489648818969727,
      "step": 4793
    },
    {
      "epoch": 0.3925928200718607,
      "grad_norm": 0.18956169486045837,
      "learning_rate": 1.8898108175487927e-05,
      "loss": 0.9440225958824158,
      "step": 4794
    },
    {
      "epoch": 0.39267471260837966,
      "grad_norm": 0.2088089883327484,
      "learning_rate": 1.8896060920078433e-05,
      "loss": 0.7082890868186951,
      "step": 4795
    },
    {
      "epoch": 0.39275660514489863,
      "grad_norm": 0.18618738651275635,
      "learning_rate": 1.889401336493773e-05,
      "loss": 0.7174160480499268,
      "step": 4796
    },
    {
      "epoch": 0.39283849768141754,
      "grad_norm": 0.2095162570476532,
      "learning_rate": 1.889196551015481e-05,
      "loss": 0.8309505581855774,
      "step": 4797
    },
    {
      "epoch": 0.3929203902179365,
      "grad_norm": 0.20121712982654572,
      "learning_rate": 1.888991735581868e-05,
      "loss": 0.8457232713699341,
      "step": 4798
    },
    {
      "epoch": 0.3930022827544555,
      "grad_norm": 0.21801941096782684,
      "learning_rate": 1.888786890201837e-05,
      "loss": 0.7142794132232666,
      "step": 4799
    },
    {
      "epoch": 0.3930841752909744,
      "grad_norm": 0.20744477212429047,
      "learning_rate": 1.8885820148842902e-05,
      "loss": 0.896957278251648,
      "step": 4800
    },
    {
      "epoch": 0.39316606782749336,
      "grad_norm": 0.1974206119775772,
      "learning_rate": 1.8883771096381336e-05,
      "loss": 1.000659465789795,
      "step": 4801
    },
    {
      "epoch": 0.39324796036401233,
      "grad_norm": 0.15925173461437225,
      "learning_rate": 1.888172174472272e-05,
      "loss": 0.7107816934585571,
      "step": 4802
    },
    {
      "epoch": 0.3933298529005313,
      "grad_norm": 0.15333260595798492,
      "learning_rate": 1.8879672093956142e-05,
      "loss": 0.7949489951133728,
      "step": 4803
    },
    {
      "epoch": 0.3934117454370502,
      "grad_norm": 0.19944031536579132,
      "learning_rate": 1.8877622144170673e-05,
      "loss": 0.9496010541915894,
      "step": 4804
    },
    {
      "epoch": 0.3934936379735692,
      "grad_norm": 0.19722013175487518,
      "learning_rate": 1.887557189545542e-05,
      "loss": 0.5283668041229248,
      "step": 4805
    },
    {
      "epoch": 0.39357553051008815,
      "grad_norm": 0.1944478154182434,
      "learning_rate": 1.8873521347899493e-05,
      "loss": 0.8928956985473633,
      "step": 4806
    },
    {
      "epoch": 0.3936574230466071,
      "grad_norm": 0.2396780103445053,
      "learning_rate": 1.887147050159202e-05,
      "loss": 1.3415393829345703,
      "step": 4807
    },
    {
      "epoch": 0.393739315583126,
      "grad_norm": 0.22172248363494873,
      "learning_rate": 1.8869419356622133e-05,
      "loss": 0.6701005101203918,
      "step": 4808
    },
    {
      "epoch": 0.393821208119645,
      "grad_norm": 0.18499638140201569,
      "learning_rate": 1.8867367913078994e-05,
      "loss": 0.6846961975097656,
      "step": 4809
    },
    {
      "epoch": 0.39390310065616396,
      "grad_norm": 0.15962941944599152,
      "learning_rate": 1.886531617105176e-05,
      "loss": 0.82304847240448,
      "step": 4810
    },
    {
      "epoch": 0.3939849931926829,
      "grad_norm": 0.1600765585899353,
      "learning_rate": 1.8863264130629602e-05,
      "loss": 0.8350710272789001,
      "step": 4811
    },
    {
      "epoch": 0.39406688572920184,
      "grad_norm": 0.16347472369670868,
      "learning_rate": 1.8861211791901726e-05,
      "loss": 1.1845346689224243,
      "step": 4812
    },
    {
      "epoch": 0.3941487782657208,
      "grad_norm": 0.2014016956090927,
      "learning_rate": 1.885915915495732e-05,
      "loss": 1.119883418083191,
      "step": 4813
    },
    {
      "epoch": 0.3942306708022398,
      "grad_norm": 0.1755836457014084,
      "learning_rate": 1.8857106219885612e-05,
      "loss": 0.8172536492347717,
      "step": 4814
    },
    {
      "epoch": 0.3943125633387587,
      "grad_norm": 0.17551127076148987,
      "learning_rate": 1.885505298677582e-05,
      "loss": 1.0441190004348755,
      "step": 4815
    },
    {
      "epoch": 0.39439445587527766,
      "grad_norm": 0.17074929177761078,
      "learning_rate": 1.8852999455717195e-05,
      "loss": 0.8962393999099731,
      "step": 4816
    },
    {
      "epoch": 0.39447634841179663,
      "grad_norm": 0.20979388058185577,
      "learning_rate": 1.8850945626798986e-05,
      "loss": 0.9057430624961853,
      "step": 4817
    },
    {
      "epoch": 0.3945582409483156,
      "grad_norm": 0.22108319401741028,
      "learning_rate": 1.8848891500110465e-05,
      "loss": 0.6513183116912842,
      "step": 4818
    },
    {
      "epoch": 0.3946401334848345,
      "grad_norm": 0.22447112202644348,
      "learning_rate": 1.884683707574091e-05,
      "loss": 0.9307804703712463,
      "step": 4819
    },
    {
      "epoch": 0.3947220260213535,
      "grad_norm": 0.19502268731594086,
      "learning_rate": 1.8844782353779614e-05,
      "loss": 0.6807905435562134,
      "step": 4820
    },
    {
      "epoch": 0.39480391855787245,
      "grad_norm": 0.1621302217245102,
      "learning_rate": 1.8842727334315887e-05,
      "loss": 0.8481922149658203,
      "step": 4821
    },
    {
      "epoch": 0.39488581109439136,
      "grad_norm": 0.1866615116596222,
      "learning_rate": 1.8840672017439047e-05,
      "loss": 1.0245755910873413,
      "step": 4822
    },
    {
      "epoch": 0.3949677036309103,
      "grad_norm": 0.1874581277370453,
      "learning_rate": 1.8838616403238427e-05,
      "loss": 0.7871371507644653,
      "step": 4823
    },
    {
      "epoch": 0.3950495961674293,
      "grad_norm": 0.16598296165466309,
      "learning_rate": 1.883656049180337e-05,
      "loss": 0.7117646932601929,
      "step": 4824
    },
    {
      "epoch": 0.39513148870394826,
      "grad_norm": 0.2599584758281708,
      "learning_rate": 1.8834504283223238e-05,
      "loss": 0.8345670700073242,
      "step": 4825
    },
    {
      "epoch": 0.3952133812404672,
      "grad_norm": 0.1995057612657547,
      "learning_rate": 1.88324477775874e-05,
      "loss": 0.7166072726249695,
      "step": 4826
    },
    {
      "epoch": 0.39529527377698614,
      "grad_norm": 0.1992579698562622,
      "learning_rate": 1.8830390974985244e-05,
      "loss": 0.6348736882209778,
      "step": 4827
    },
    {
      "epoch": 0.3953771663135051,
      "grad_norm": 0.21752823889255524,
      "learning_rate": 1.8828333875506157e-05,
      "loss": 0.8026854991912842,
      "step": 4828
    },
    {
      "epoch": 0.3954590588500241,
      "grad_norm": 0.19239723682403564,
      "learning_rate": 1.882627647923956e-05,
      "loss": 1.0187113285064697,
      "step": 4829
    },
    {
      "epoch": 0.395540951386543,
      "grad_norm": 0.17154355347156525,
      "learning_rate": 1.8824218786274872e-05,
      "loss": 0.5761831998825073,
      "step": 4830
    },
    {
      "epoch": 0.39562284392306196,
      "grad_norm": 0.17544662952423096,
      "learning_rate": 1.882216079670153e-05,
      "loss": 0.6359361410140991,
      "step": 4831
    },
    {
      "epoch": 0.39570473645958093,
      "grad_norm": 0.19478167593479156,
      "learning_rate": 1.882010251060898e-05,
      "loss": 0.9323430061340332,
      "step": 4832
    },
    {
      "epoch": 0.39578662899609984,
      "grad_norm": 0.22067810595035553,
      "learning_rate": 1.881804392808668e-05,
      "loss": 0.7781283855438232,
      "step": 4833
    },
    {
      "epoch": 0.3958685215326188,
      "grad_norm": 0.19773642718791962,
      "learning_rate": 1.8815985049224113e-05,
      "loss": 1.1480196714401245,
      "step": 4834
    },
    {
      "epoch": 0.3959504140691378,
      "grad_norm": 0.19072774052619934,
      "learning_rate": 1.8813925874110766e-05,
      "loss": 0.6420587301254272,
      "step": 4835
    },
    {
      "epoch": 0.39603230660565675,
      "grad_norm": 0.18430760502815247,
      "learning_rate": 1.8811866402836136e-05,
      "loss": 0.8467979431152344,
      "step": 4836
    },
    {
      "epoch": 0.39611419914217566,
      "grad_norm": 0.17710460722446442,
      "learning_rate": 1.8809806635489732e-05,
      "loss": 0.6361441612243652,
      "step": 4837
    },
    {
      "epoch": 0.39619609167869463,
      "grad_norm": 0.1872129589319229,
      "learning_rate": 1.8807746572161085e-05,
      "loss": 0.5884761214256287,
      "step": 4838
    },
    {
      "epoch": 0.3962779842152136,
      "grad_norm": 0.17545996606349945,
      "learning_rate": 1.8805686212939733e-05,
      "loss": 1.0176620483398438,
      "step": 4839
    },
    {
      "epoch": 0.39635987675173256,
      "grad_norm": 0.18100392818450928,
      "learning_rate": 1.880362555791523e-05,
      "loss": 0.8562783598899841,
      "step": 4840
    },
    {
      "epoch": 0.3964417692882515,
      "grad_norm": 0.15924102067947388,
      "learning_rate": 1.8801564607177138e-05,
      "loss": 0.672008216381073,
      "step": 4841
    },
    {
      "epoch": 0.39652366182477045,
      "grad_norm": 0.19947659969329834,
      "learning_rate": 1.8799503360815033e-05,
      "loss": 0.7233729958534241,
      "step": 4842
    },
    {
      "epoch": 0.3966055543612894,
      "grad_norm": 0.18508578836917877,
      "learning_rate": 1.8797441818918507e-05,
      "loss": 1.0334047079086304,
      "step": 4843
    },
    {
      "epoch": 0.3966874468978083,
      "grad_norm": 0.17228274047374725,
      "learning_rate": 1.8795379981577164e-05,
      "loss": 0.5747145414352417,
      "step": 4844
    },
    {
      "epoch": 0.3967693394343273,
      "grad_norm": 0.1550668627023697,
      "learning_rate": 1.8793317848880616e-05,
      "loss": 1.0177005529403687,
      "step": 4845
    },
    {
      "epoch": 0.39685123197084626,
      "grad_norm": 0.22186262905597687,
      "learning_rate": 1.8791255420918496e-05,
      "loss": 0.7089474201202393,
      "step": 4846
    },
    {
      "epoch": 0.39693312450736523,
      "grad_norm": 0.29002153873443604,
      "learning_rate": 1.8789192697780447e-05,
      "loss": 1.2491763830184937,
      "step": 4847
    },
    {
      "epoch": 0.39701501704388414,
      "grad_norm": 0.1888546496629715,
      "learning_rate": 1.8787129679556117e-05,
      "loss": 0.7779920101165771,
      "step": 4848
    },
    {
      "epoch": 0.3970969095804031,
      "grad_norm": 0.21422606706619263,
      "learning_rate": 1.878506636633518e-05,
      "loss": 1.074955940246582,
      "step": 4849
    },
    {
      "epoch": 0.3971788021169221,
      "grad_norm": 0.19364796578884125,
      "learning_rate": 1.8783002758207308e-05,
      "loss": 0.6064584255218506,
      "step": 4850
    },
    {
      "epoch": 0.39726069465344105,
      "grad_norm": 0.17778833210468292,
      "learning_rate": 1.8780938855262203e-05,
      "loss": 1.032112717628479,
      "step": 4851
    },
    {
      "epoch": 0.39734258718995996,
      "grad_norm": 0.2084885835647583,
      "learning_rate": 1.877887465758957e-05,
      "loss": 0.8352402448654175,
      "step": 4852
    },
    {
      "epoch": 0.39742447972647893,
      "grad_norm": 0.15326200425624847,
      "learning_rate": 1.877681016527912e-05,
      "loss": 0.6935020089149475,
      "step": 4853
    },
    {
      "epoch": 0.3975063722629979,
      "grad_norm": 0.19249418377876282,
      "learning_rate": 1.8774745378420587e-05,
      "loss": 0.7612571120262146,
      "step": 4854
    },
    {
      "epoch": 0.3975882647995168,
      "grad_norm": 0.22246062755584717,
      "learning_rate": 1.8772680297103717e-05,
      "loss": 1.0527467727661133,
      "step": 4855
    },
    {
      "epoch": 0.3976701573360358,
      "grad_norm": 0.19392932951450348,
      "learning_rate": 1.877061492141827e-05,
      "loss": 0.985914945602417,
      "step": 4856
    },
    {
      "epoch": 0.39775204987255475,
      "grad_norm": 0.19676290452480316,
      "learning_rate": 1.876854925145401e-05,
      "loss": 1.0880986452102661,
      "step": 4857
    },
    {
      "epoch": 0.3978339424090737,
      "grad_norm": 0.19492173194885254,
      "learning_rate": 1.8766483287300725e-05,
      "loss": 0.8911033868789673,
      "step": 4858
    },
    {
      "epoch": 0.39791583494559263,
      "grad_norm": 0.20369091629981995,
      "learning_rate": 1.87644170290482e-05,
      "loss": 0.6007199883460999,
      "step": 4859
    },
    {
      "epoch": 0.3979977274821116,
      "grad_norm": 0.19511395692825317,
      "learning_rate": 1.876235047678626e-05,
      "loss": 0.8877812623977661,
      "step": 4860
    },
    {
      "epoch": 0.39807962001863056,
      "grad_norm": 0.18075543642044067,
      "learning_rate": 1.876028363060471e-05,
      "loss": 0.954290509223938,
      "step": 4861
    },
    {
      "epoch": 0.39816151255514953,
      "grad_norm": 0.18347269296646118,
      "learning_rate": 1.8758216490593392e-05,
      "loss": 0.9476075172424316,
      "step": 4862
    },
    {
      "epoch": 0.39824340509166845,
      "grad_norm": 0.29638779163360596,
      "learning_rate": 1.875614905684215e-05,
      "loss": 0.7463136911392212,
      "step": 4863
    },
    {
      "epoch": 0.3983252976281874,
      "grad_norm": 0.201890230178833,
      "learning_rate": 1.8754081329440843e-05,
      "loss": 1.0599281787872314,
      "step": 4864
    },
    {
      "epoch": 0.3984071901647064,
      "grad_norm": 0.18756718933582306,
      "learning_rate": 1.875201330847934e-05,
      "loss": 0.9535694122314453,
      "step": 4865
    },
    {
      "epoch": 0.3984890827012253,
      "grad_norm": 0.17451325058937073,
      "learning_rate": 1.8749944994047535e-05,
      "loss": 0.9142393469810486,
      "step": 4866
    },
    {
      "epoch": 0.39857097523774426,
      "grad_norm": 0.16986796259880066,
      "learning_rate": 1.874787638623532e-05,
      "loss": 0.7894538044929504,
      "step": 4867
    },
    {
      "epoch": 0.39865286777426323,
      "grad_norm": 0.1724662333726883,
      "learning_rate": 1.87458074851326e-05,
      "loss": 0.6923216581344604,
      "step": 4868
    },
    {
      "epoch": 0.3987347603107822,
      "grad_norm": 0.18401780724525452,
      "learning_rate": 1.8743738290829307e-05,
      "loss": 1.099830985069275,
      "step": 4869
    },
    {
      "epoch": 0.3988166528473011,
      "grad_norm": 0.1732831746339798,
      "learning_rate": 1.8741668803415368e-05,
      "loss": 0.976685643196106,
      "step": 4870
    },
    {
      "epoch": 0.3988985453838201,
      "grad_norm": 0.20462198555469513,
      "learning_rate": 1.873959902298074e-05,
      "loss": 0.8694584369659424,
      "step": 4871
    },
    {
      "epoch": 0.39898043792033905,
      "grad_norm": 0.20911945402622223,
      "learning_rate": 1.8737528949615378e-05,
      "loss": 1.0574796199798584,
      "step": 4872
    },
    {
      "epoch": 0.39906233045685796,
      "grad_norm": 0.16782048344612122,
      "learning_rate": 1.873545858340926e-05,
      "loss": 0.9791169166564941,
      "step": 4873
    },
    {
      "epoch": 0.39914422299337693,
      "grad_norm": 0.16992226243019104,
      "learning_rate": 1.8733387924452365e-05,
      "loss": 0.8472546339035034,
      "step": 4874
    },
    {
      "epoch": 0.3992261155298959,
      "grad_norm": 0.1652415543794632,
      "learning_rate": 1.8731316972834703e-05,
      "loss": 0.5662095546722412,
      "step": 4875
    },
    {
      "epoch": 0.39930800806641487,
      "grad_norm": 0.1790178269147873,
      "learning_rate": 1.872924572864628e-05,
      "loss": 0.48817360401153564,
      "step": 4876
    },
    {
      "epoch": 0.3993899006029338,
      "grad_norm": 0.20772720873355865,
      "learning_rate": 1.8727174191977122e-05,
      "loss": 0.9243919849395752,
      "step": 4877
    },
    {
      "epoch": 0.39947179313945275,
      "grad_norm": 0.167026087641716,
      "learning_rate": 1.8725102362917267e-05,
      "loss": 0.7697576284408569,
      "step": 4878
    },
    {
      "epoch": 0.3995536856759717,
      "grad_norm": 0.2021031528711319,
      "learning_rate": 1.8723030241556763e-05,
      "loss": 0.9909370541572571,
      "step": 4879
    },
    {
      "epoch": 0.3996355782124907,
      "grad_norm": 0.18447256088256836,
      "learning_rate": 1.8720957827985675e-05,
      "loss": 0.8108887672424316,
      "step": 4880
    },
    {
      "epoch": 0.3997174707490096,
      "grad_norm": 0.16770540177822113,
      "learning_rate": 1.8718885122294082e-05,
      "loss": 0.8009738922119141,
      "step": 4881
    },
    {
      "epoch": 0.39979936328552856,
      "grad_norm": 0.15608267486095428,
      "learning_rate": 1.871681212457206e-05,
      "loss": 0.4319993257522583,
      "step": 4882
    },
    {
      "epoch": 0.39988125582204753,
      "grad_norm": 0.19232289493083954,
      "learning_rate": 1.8714738834909727e-05,
      "loss": 0.6601657271385193,
      "step": 4883
    },
    {
      "epoch": 0.39996314835856644,
      "grad_norm": 0.19573992490768433,
      "learning_rate": 1.8712665253397186e-05,
      "loss": 1.050797700881958,
      "step": 4884
    },
    {
      "epoch": 0.4000450408950854,
      "grad_norm": 0.20168934762477875,
      "learning_rate": 1.8710591380124564e-05,
      "loss": 1.1181604862213135,
      "step": 4885
    },
    {
      "epoch": 0.4001269334316044,
      "grad_norm": 0.1804691106081009,
      "learning_rate": 1.8708517215182004e-05,
      "loss": 0.9934780597686768,
      "step": 4886
    },
    {
      "epoch": 0.40020882596812335,
      "grad_norm": 0.18792091310024261,
      "learning_rate": 1.8706442758659654e-05,
      "loss": 0.9834442138671875,
      "step": 4887
    },
    {
      "epoch": 0.40029071850464226,
      "grad_norm": 0.5209029316902161,
      "learning_rate": 1.8704368010647678e-05,
      "loss": 1.3392693996429443,
      "step": 4888
    },
    {
      "epoch": 0.40037261104116123,
      "grad_norm": 0.18379555642604828,
      "learning_rate": 1.870229297123626e-05,
      "loss": 0.7264520525932312,
      "step": 4889
    },
    {
      "epoch": 0.4004545035776802,
      "grad_norm": 0.17947925627231598,
      "learning_rate": 1.8700217640515582e-05,
      "loss": 0.8933248519897461,
      "step": 4890
    },
    {
      "epoch": 0.40053639611419917,
      "grad_norm": 0.20687562227249146,
      "learning_rate": 1.8698142018575844e-05,
      "loss": 0.6400417685508728,
      "step": 4891
    },
    {
      "epoch": 0.4006182886507181,
      "grad_norm": 0.17644000053405762,
      "learning_rate": 1.8696066105507273e-05,
      "loss": 1.0227482318878174,
      "step": 4892
    },
    {
      "epoch": 0.40070018118723705,
      "grad_norm": 0.19244930148124695,
      "learning_rate": 1.8693989901400086e-05,
      "loss": 1.3238481283187866,
      "step": 4893
    },
    {
      "epoch": 0.400782073723756,
      "grad_norm": 0.19676131010055542,
      "learning_rate": 1.8691913406344533e-05,
      "loss": 0.8338460922241211,
      "step": 4894
    },
    {
      "epoch": 0.40086396626027493,
      "grad_norm": 0.17388290166854858,
      "learning_rate": 1.8689836620430855e-05,
      "loss": 0.8799936771392822,
      "step": 4895
    },
    {
      "epoch": 0.4009458587967939,
      "grad_norm": 0.1800016611814499,
      "learning_rate": 1.868775954374933e-05,
      "loss": 0.6964021325111389,
      "step": 4896
    },
    {
      "epoch": 0.40102775133331287,
      "grad_norm": 0.20619326829910278,
      "learning_rate": 1.8685682176390224e-05,
      "loss": 1.0306931734085083,
      "step": 4897
    },
    {
      "epoch": 0.40110964386983183,
      "grad_norm": 0.27117466926574707,
      "learning_rate": 1.868360451844384e-05,
      "loss": 0.8637059926986694,
      "step": 4898
    },
    {
      "epoch": 0.40119153640635075,
      "grad_norm": 0.1894320249557495,
      "learning_rate": 1.868152657000047e-05,
      "loss": 1.009373664855957,
      "step": 4899
    },
    {
      "epoch": 0.4012734289428697,
      "grad_norm": 0.1927165985107422,
      "learning_rate": 1.8679448331150442e-05,
      "loss": 0.8266853094100952,
      "step": 4900
    },
    {
      "epoch": 0.4013553214793887,
      "grad_norm": 0.1999220848083496,
      "learning_rate": 1.8677369801984076e-05,
      "loss": 0.8183196187019348,
      "step": 4901
    },
    {
      "epoch": 0.40143721401590765,
      "grad_norm": 0.1815393567085266,
      "learning_rate": 1.8675290982591717e-05,
      "loss": 0.6955007910728455,
      "step": 4902
    },
    {
      "epoch": 0.40151910655242656,
      "grad_norm": 0.21682676672935486,
      "learning_rate": 1.8673211873063716e-05,
      "loss": 1.1195605993270874,
      "step": 4903
    },
    {
      "epoch": 0.40160099908894553,
      "grad_norm": 0.16370873153209686,
      "learning_rate": 1.8671132473490448e-05,
      "loss": 0.8337005972862244,
      "step": 4904
    },
    {
      "epoch": 0.4016828916254645,
      "grad_norm": 0.14623259007930756,
      "learning_rate": 1.866905278396228e-05,
      "loss": 0.6407696604728699,
      "step": 4905
    },
    {
      "epoch": 0.4017647841619834,
      "grad_norm": 0.16124309599399567,
      "learning_rate": 1.8666972804569613e-05,
      "loss": 0.6979441046714783,
      "step": 4906
    },
    {
      "epoch": 0.4018466766985024,
      "grad_norm": 0.17841477692127228,
      "learning_rate": 1.8664892535402848e-05,
      "loss": 0.7008439898490906,
      "step": 4907
    },
    {
      "epoch": 0.40192856923502135,
      "grad_norm": 0.15273445844650269,
      "learning_rate": 1.8662811976552403e-05,
      "loss": 0.541215717792511,
      "step": 4908
    },
    {
      "epoch": 0.4020104617715403,
      "grad_norm": 0.17031942307949066,
      "learning_rate": 1.866073112810871e-05,
      "loss": 0.8181204199790955,
      "step": 4909
    },
    {
      "epoch": 0.40209235430805923,
      "grad_norm": 0.16850562393665314,
      "learning_rate": 1.8658649990162205e-05,
      "loss": 0.7695010900497437,
      "step": 4910
    },
    {
      "epoch": 0.4021742468445782,
      "grad_norm": 0.18291139602661133,
      "learning_rate": 1.8656568562803347e-05,
      "loss": 1.1851997375488281,
      "step": 4911
    },
    {
      "epoch": 0.40225613938109717,
      "grad_norm": 0.16229744255542755,
      "learning_rate": 1.8654486846122604e-05,
      "loss": 0.8543393611907959,
      "step": 4912
    },
    {
      "epoch": 0.40233803191761613,
      "grad_norm": 0.18266651034355164,
      "learning_rate": 1.8652404840210455e-05,
      "loss": 0.6875848174095154,
      "step": 4913
    },
    {
      "epoch": 0.40241992445413505,
      "grad_norm": 0.19899767637252808,
      "learning_rate": 1.8650322545157388e-05,
      "loss": 0.8558641672134399,
      "step": 4914
    },
    {
      "epoch": 0.402501816990654,
      "grad_norm": 0.1952270269393921,
      "learning_rate": 1.8648239961053918e-05,
      "loss": 0.7933339476585388,
      "step": 4915
    },
    {
      "epoch": 0.402583709527173,
      "grad_norm": 0.18190136551856995,
      "learning_rate": 1.8646157087990555e-05,
      "loss": 0.976254403591156,
      "step": 4916
    },
    {
      "epoch": 0.4026656020636919,
      "grad_norm": 0.1763625293970108,
      "learning_rate": 1.864407392605783e-05,
      "loss": 0.9067272543907166,
      "step": 4917
    },
    {
      "epoch": 0.40274749460021086,
      "grad_norm": 0.17689669132232666,
      "learning_rate": 1.8641990475346292e-05,
      "loss": 0.8053146004676819,
      "step": 4918
    },
    {
      "epoch": 0.40282938713672983,
      "grad_norm": 0.211796835064888,
      "learning_rate": 1.8639906735946488e-05,
      "loss": 0.9738070964813232,
      "step": 4919
    },
    {
      "epoch": 0.4029112796732488,
      "grad_norm": 0.17949987947940826,
      "learning_rate": 1.8637822707948992e-05,
      "loss": 0.8623453974723816,
      "step": 4920
    },
    {
      "epoch": 0.4029931722097677,
      "grad_norm": 0.17677032947540283,
      "learning_rate": 1.863573839144438e-05,
      "loss": 0.9536089897155762,
      "step": 4921
    },
    {
      "epoch": 0.4030750647462867,
      "grad_norm": 0.18126145005226135,
      "learning_rate": 1.8633653786523247e-05,
      "loss": 0.7138976454734802,
      "step": 4922
    },
    {
      "epoch": 0.40315695728280565,
      "grad_norm": 0.17649222910404205,
      "learning_rate": 1.8631568893276198e-05,
      "loss": 0.7391239404678345,
      "step": 4923
    },
    {
      "epoch": 0.4032388498193246,
      "grad_norm": 0.1705232858657837,
      "learning_rate": 1.862948371179385e-05,
      "loss": 0.9113979339599609,
      "step": 4924
    },
    {
      "epoch": 0.40332074235584353,
      "grad_norm": 0.1746300309896469,
      "learning_rate": 1.8627398242166837e-05,
      "loss": 0.8079941272735596,
      "step": 4925
    },
    {
      "epoch": 0.4034026348923625,
      "grad_norm": 0.18022660911083221,
      "learning_rate": 1.8625312484485802e-05,
      "loss": 1.0175292491912842,
      "step": 4926
    },
    {
      "epoch": 0.40348452742888147,
      "grad_norm": 0.20292019844055176,
      "learning_rate": 1.8623226438841396e-05,
      "loss": 0.9352402687072754,
      "step": 4927
    },
    {
      "epoch": 0.4035664199654004,
      "grad_norm": 0.19311539828777313,
      "learning_rate": 1.8621140105324288e-05,
      "loss": 0.7845450043678284,
      "step": 4928
    },
    {
      "epoch": 0.40364831250191935,
      "grad_norm": 0.22226370871067047,
      "learning_rate": 1.8619053484025167e-05,
      "loss": 0.8938077092170715,
      "step": 4929
    },
    {
      "epoch": 0.4037302050384383,
      "grad_norm": 0.19201205670833588,
      "learning_rate": 1.861696657503472e-05,
      "loss": 0.7553132772445679,
      "step": 4930
    },
    {
      "epoch": 0.4038120975749573,
      "grad_norm": 0.19559839367866516,
      "learning_rate": 1.861487937844364e-05,
      "loss": 1.0512011051177979,
      "step": 4931
    },
    {
      "epoch": 0.4038939901114762,
      "grad_norm": 0.1607467383146286,
      "learning_rate": 1.861279189434267e-05,
      "loss": 0.6537278294563293,
      "step": 4932
    },
    {
      "epoch": 0.40397588264799517,
      "grad_norm": 0.20775189995765686,
      "learning_rate": 1.8610704122822524e-05,
      "loss": 0.6359946131706238,
      "step": 4933
    },
    {
      "epoch": 0.40405777518451413,
      "grad_norm": 0.21767739951610565,
      "learning_rate": 1.860861606397395e-05,
      "loss": 0.8081493377685547,
      "step": 4934
    },
    {
      "epoch": 0.40413966772103305,
      "grad_norm": 0.1786140650510788,
      "learning_rate": 1.86065277178877e-05,
      "loss": 1.2672450542449951,
      "step": 4935
    },
    {
      "epoch": 0.404221560257552,
      "grad_norm": 0.2459297925233841,
      "learning_rate": 1.8604439084654547e-05,
      "loss": 0.7733134627342224,
      "step": 4936
    },
    {
      "epoch": 0.404303452794071,
      "grad_norm": 0.20963795483112335,
      "learning_rate": 1.860235016436527e-05,
      "loss": 0.8020097017288208,
      "step": 4937
    },
    {
      "epoch": 0.40438534533058995,
      "grad_norm": 0.21937990188598633,
      "learning_rate": 1.8600260957110665e-05,
      "loss": 0.5839344263076782,
      "step": 4938
    },
    {
      "epoch": 0.40446723786710886,
      "grad_norm": 0.16929015517234802,
      "learning_rate": 1.859817146298153e-05,
      "loss": 0.9136120080947876,
      "step": 4939
    },
    {
      "epoch": 0.40454913040362783,
      "grad_norm": 0.16853730380535126,
      "learning_rate": 1.8596081682068688e-05,
      "loss": 0.883001446723938,
      "step": 4940
    },
    {
      "epoch": 0.4046310229401468,
      "grad_norm": 0.18358813226222992,
      "learning_rate": 1.859399161446297e-05,
      "loss": 0.6945838332176208,
      "step": 4941
    },
    {
      "epoch": 0.40471291547666577,
      "grad_norm": 0.2062784880399704,
      "learning_rate": 1.8591901260255217e-05,
      "loss": 0.7648672461509705,
      "step": 4942
    },
    {
      "epoch": 0.4047948080131847,
      "grad_norm": 0.19497710466384888,
      "learning_rate": 1.8589810619536286e-05,
      "loss": 0.6843498349189758,
      "step": 4943
    },
    {
      "epoch": 0.40487670054970365,
      "grad_norm": 0.1966712474822998,
      "learning_rate": 1.8587719692397044e-05,
      "loss": 0.7911531329154968,
      "step": 4944
    },
    {
      "epoch": 0.4049585930862226,
      "grad_norm": 0.1806543469429016,
      "learning_rate": 1.8585628478928366e-05,
      "loss": 1.0327527523040771,
      "step": 4945
    },
    {
      "epoch": 0.40504048562274153,
      "grad_norm": 0.18884876370429993,
      "learning_rate": 1.858353697922116e-05,
      "loss": 0.9346252679824829,
      "step": 4946
    },
    {
      "epoch": 0.4051223781592605,
      "grad_norm": 0.1834551841020584,
      "learning_rate": 1.8581445193366313e-05,
      "loss": 0.8489859104156494,
      "step": 4947
    },
    {
      "epoch": 0.40520427069577947,
      "grad_norm": 0.1718529760837555,
      "learning_rate": 1.8579353121454755e-05,
      "loss": 0.8416170477867126,
      "step": 4948
    },
    {
      "epoch": 0.40528616323229844,
      "grad_norm": 0.20401811599731445,
      "learning_rate": 1.857726076357741e-05,
      "loss": 0.9367818832397461,
      "step": 4949
    },
    {
      "epoch": 0.40536805576881735,
      "grad_norm": 0.2048729956150055,
      "learning_rate": 1.8575168119825225e-05,
      "loss": 0.9329394102096558,
      "step": 4950
    },
    {
      "epoch": 0.4054499483053363,
      "grad_norm": 0.19837123155593872,
      "learning_rate": 1.857307519028915e-05,
      "loss": 1.1233603954315186,
      "step": 4951
    },
    {
      "epoch": 0.4055318408418553,
      "grad_norm": 0.18282349407672882,
      "learning_rate": 1.8570981975060155e-05,
      "loss": 0.6810227036476135,
      "step": 4952
    },
    {
      "epoch": 0.40561373337837425,
      "grad_norm": 0.15881504118442535,
      "learning_rate": 1.856888847422922e-05,
      "loss": 0.726540744304657,
      "step": 4953
    },
    {
      "epoch": 0.40569562591489317,
      "grad_norm": 0.22410526871681213,
      "learning_rate": 1.856679468788734e-05,
      "loss": 0.9761060476303101,
      "step": 4954
    },
    {
      "epoch": 0.40577751845141213,
      "grad_norm": 0.1948442906141281,
      "learning_rate": 1.8564700616125513e-05,
      "loss": 0.8350445032119751,
      "step": 4955
    },
    {
      "epoch": 0.4058594109879311,
      "grad_norm": 0.1629355400800705,
      "learning_rate": 1.8562606259034754e-05,
      "loss": 0.8301382064819336,
      "step": 4956
    },
    {
      "epoch": 0.40594130352445,
      "grad_norm": 0.16960804164409637,
      "learning_rate": 1.8560511616706108e-05,
      "loss": 0.7986688017845154,
      "step": 4957
    },
    {
      "epoch": 0.406023196060969,
      "grad_norm": 0.1749076098203659,
      "learning_rate": 1.85584166892306e-05,
      "loss": 0.8358690738677979,
      "step": 4958
    },
    {
      "epoch": 0.40610508859748795,
      "grad_norm": 0.16591525077819824,
      "learning_rate": 1.8556321476699292e-05,
      "loss": 0.7329019904136658,
      "step": 4959
    },
    {
      "epoch": 0.4061869811340069,
      "grad_norm": 0.1741303652524948,
      "learning_rate": 1.8554225979203248e-05,
      "loss": 0.6649019718170166,
      "step": 4960
    },
    {
      "epoch": 0.40626887367052583,
      "grad_norm": 0.1892394870519638,
      "learning_rate": 1.8552130196833547e-05,
      "loss": 0.8816577196121216,
      "step": 4961
    },
    {
      "epoch": 0.4063507662070448,
      "grad_norm": 0.199670672416687,
      "learning_rate": 1.8550034129681286e-05,
      "loss": 0.8123025298118591,
      "step": 4962
    },
    {
      "epoch": 0.40643265874356377,
      "grad_norm": 0.19663968682289124,
      "learning_rate": 1.854793777783756e-05,
      "loss": 0.9683864712715149,
      "step": 4963
    },
    {
      "epoch": 0.40651455128008274,
      "grad_norm": 0.21916139125823975,
      "learning_rate": 1.8545841141393492e-05,
      "loss": 0.884762704372406,
      "step": 4964
    },
    {
      "epoch": 0.40659644381660165,
      "grad_norm": 0.20365190505981445,
      "learning_rate": 1.8543744220440206e-05,
      "loss": 0.9140419960021973,
      "step": 4965
    },
    {
      "epoch": 0.4066783363531206,
      "grad_norm": 0.22361786663532257,
      "learning_rate": 1.8541647015068846e-05,
      "loss": 1.1997884511947632,
      "step": 4966
    },
    {
      "epoch": 0.4067602288896396,
      "grad_norm": 0.2001107633113861,
      "learning_rate": 1.853954952537056e-05,
      "loss": 0.769279956817627,
      "step": 4967
    },
    {
      "epoch": 0.4068421214261585,
      "grad_norm": 0.19657249748706818,
      "learning_rate": 1.853745175143652e-05,
      "loss": 0.7329741716384888,
      "step": 4968
    },
    {
      "epoch": 0.40692401396267747,
      "grad_norm": 0.2194504588842392,
      "learning_rate": 1.85353536933579e-05,
      "loss": 0.9167615175247192,
      "step": 4969
    },
    {
      "epoch": 0.40700590649919643,
      "grad_norm": 0.19511324167251587,
      "learning_rate": 1.8533255351225892e-05,
      "loss": 0.9569521546363831,
      "step": 4970
    },
    {
      "epoch": 0.4070877990357154,
      "grad_norm": 0.18368768692016602,
      "learning_rate": 1.85311567251317e-05,
      "loss": 0.924453616142273,
      "step": 4971
    },
    {
      "epoch": 0.4071696915722343,
      "grad_norm": 0.1920098513364792,
      "learning_rate": 1.8529057815166535e-05,
      "loss": 0.9231491088867188,
      "step": 4972
    },
    {
      "epoch": 0.4072515841087533,
      "grad_norm": 0.20510846376419067,
      "learning_rate": 1.852695862142162e-05,
      "loss": 0.7535781860351562,
      "step": 4973
    },
    {
      "epoch": 0.40733347664527225,
      "grad_norm": 0.2008444368839264,
      "learning_rate": 1.852485914398821e-05,
      "loss": 0.9512158632278442,
      "step": 4974
    },
    {
      "epoch": 0.4074153691817912,
      "grad_norm": 0.17124027013778687,
      "learning_rate": 1.8522759382957547e-05,
      "loss": 0.7773808836936951,
      "step": 4975
    },
    {
      "epoch": 0.40749726171831013,
      "grad_norm": 0.18196497857570648,
      "learning_rate": 1.8520659338420892e-05,
      "loss": 0.8907803893089294,
      "step": 4976
    },
    {
      "epoch": 0.4075791542548291,
      "grad_norm": 0.1609843522310257,
      "learning_rate": 1.8518559010469528e-05,
      "loss": 1.0434606075286865,
      "step": 4977
    },
    {
      "epoch": 0.40766104679134807,
      "grad_norm": 0.178365558385849,
      "learning_rate": 1.851645839919474e-05,
      "loss": 0.8374481797218323,
      "step": 4978
    },
    {
      "epoch": 0.407742939327867,
      "grad_norm": 0.15459269285202026,
      "learning_rate": 1.851435750468783e-05,
      "loss": 0.7163717150688171,
      "step": 4979
    },
    {
      "epoch": 0.40782483186438595,
      "grad_norm": 0.2024887651205063,
      "learning_rate": 1.8512256327040114e-05,
      "loss": 0.6893372535705566,
      "step": 4980
    },
    {
      "epoch": 0.4079067244009049,
      "grad_norm": 0.18016675114631653,
      "learning_rate": 1.8510154866342915e-05,
      "loss": 0.45104050636291504,
      "step": 4981
    },
    {
      "epoch": 0.4079886169374239,
      "grad_norm": 0.18132303655147552,
      "learning_rate": 1.850805312268757e-05,
      "loss": 0.6311023235321045,
      "step": 4982
    },
    {
      "epoch": 0.4080705094739428,
      "grad_norm": 0.17496518790721893,
      "learning_rate": 1.8505951096165434e-05,
      "loss": 0.9249585270881653,
      "step": 4983
    },
    {
      "epoch": 0.40815240201046177,
      "grad_norm": 0.17304642498493195,
      "learning_rate": 1.850384878686787e-05,
      "loss": 0.7712814211845398,
      "step": 4984
    },
    {
      "epoch": 0.40823429454698074,
      "grad_norm": 0.22274519503116608,
      "learning_rate": 1.8501746194886244e-05,
      "loss": 0.6546491980552673,
      "step": 4985
    },
    {
      "epoch": 0.4083161870834997,
      "grad_norm": 0.24287113547325134,
      "learning_rate": 1.8499643320311956e-05,
      "loss": 0.8947634100914001,
      "step": 4986
    },
    {
      "epoch": 0.4083980796200186,
      "grad_norm": 0.18471942842006683,
      "learning_rate": 1.8497540163236395e-05,
      "loss": 0.9118187427520752,
      "step": 4987
    },
    {
      "epoch": 0.4084799721565376,
      "grad_norm": 0.21892130374908447,
      "learning_rate": 1.849543672375098e-05,
      "loss": 1.0773539543151855,
      "step": 4988
    },
    {
      "epoch": 0.40856186469305655,
      "grad_norm": 0.19448097050189972,
      "learning_rate": 1.849333300194713e-05,
      "loss": 0.8315266370773315,
      "step": 4989
    },
    {
      "epoch": 0.40864375722957547,
      "grad_norm": 0.18658170104026794,
      "learning_rate": 1.8491228997916285e-05,
      "loss": 1.0829107761383057,
      "step": 4990
    },
    {
      "epoch": 0.40872564976609443,
      "grad_norm": 0.21908164024353027,
      "learning_rate": 1.848912471174989e-05,
      "loss": 0.9035988450050354,
      "step": 4991
    },
    {
      "epoch": 0.4088075423026134,
      "grad_norm": 0.1565021425485611,
      "learning_rate": 1.8487020143539412e-05,
      "loss": 0.7623292207717896,
      "step": 4992
    },
    {
      "epoch": 0.40888943483913237,
      "grad_norm": 0.1700085550546646,
      "learning_rate": 1.848491529337632e-05,
      "loss": 1.0076258182525635,
      "step": 4993
    },
    {
      "epoch": 0.4089713273756513,
      "grad_norm": 0.1764584481716156,
      "learning_rate": 1.8482810161352095e-05,
      "loss": 1.2546398639678955,
      "step": 4994
    },
    {
      "epoch": 0.40905321991217025,
      "grad_norm": 0.13372963666915894,
      "learning_rate": 1.8480704747558247e-05,
      "loss": 0.6615644693374634,
      "step": 4995
    },
    {
      "epoch": 0.4091351124486892,
      "grad_norm": 0.1648058444261551,
      "learning_rate": 1.8478599052086272e-05,
      "loss": 1.0163546800613403,
      "step": 4996
    },
    {
      "epoch": 0.4092170049852082,
      "grad_norm": 0.2006029635667801,
      "learning_rate": 1.84764930750277e-05,
      "loss": 1.329639196395874,
      "step": 4997
    },
    {
      "epoch": 0.4092988975217271,
      "grad_norm": 0.1914386749267578,
      "learning_rate": 1.8474386816474068e-05,
      "loss": 0.6329330801963806,
      "step": 4998
    },
    {
      "epoch": 0.40938079005824607,
      "grad_norm": 0.17491298913955688,
      "learning_rate": 1.8472280276516916e-05,
      "loss": 1.0190798044204712,
      "step": 4999
    },
    {
      "epoch": 0.40946268259476504,
      "grad_norm": 0.1948796510696411,
      "learning_rate": 1.8470173455247806e-05,
      "loss": 0.9776408672332764,
      "step": 5000
    },
    {
      "epoch": 0.40954457513128395,
      "grad_norm": 0.1562151461839676,
      "learning_rate": 1.846806635275831e-05,
      "loss": 0.9279755353927612,
      "step": 5001
    },
    {
      "epoch": 0.4096264676678029,
      "grad_norm": 0.1939138025045395,
      "learning_rate": 1.8465958969140008e-05,
      "loss": 0.7593216896057129,
      "step": 5002
    },
    {
      "epoch": 0.4097083602043219,
      "grad_norm": 0.17962585389614105,
      "learning_rate": 1.8463851304484503e-05,
      "loss": 0.8716633319854736,
      "step": 5003
    },
    {
      "epoch": 0.40979025274084085,
      "grad_norm": 0.20440377295017242,
      "learning_rate": 1.846174335888339e-05,
      "loss": 1.3017947673797607,
      "step": 5004
    },
    {
      "epoch": 0.40987214527735977,
      "grad_norm": 0.2181839495897293,
      "learning_rate": 1.84596351324283e-05,
      "loss": 0.6828992962837219,
      "step": 5005
    },
    {
      "epoch": 0.40995403781387874,
      "grad_norm": 0.19064843654632568,
      "learning_rate": 1.8457526625210867e-05,
      "loss": 0.7747812271118164,
      "step": 5006
    },
    {
      "epoch": 0.4100359303503977,
      "grad_norm": 0.1992618441581726,
      "learning_rate": 1.8455417837322724e-05,
      "loss": 0.8848176002502441,
      "step": 5007
    },
    {
      "epoch": 0.4101178228869166,
      "grad_norm": 0.2140759378671646,
      "learning_rate": 1.8453308768855533e-05,
      "loss": 1.458719253540039,
      "step": 5008
    },
    {
      "epoch": 0.4101997154234356,
      "grad_norm": 0.1962522268295288,
      "learning_rate": 1.8451199419900972e-05,
      "loss": 0.8300123810768127,
      "step": 5009
    },
    {
      "epoch": 0.41028160795995455,
      "grad_norm": 0.15152139961719513,
      "learning_rate": 1.8449089790550706e-05,
      "loss": 0.6351271867752075,
      "step": 5010
    },
    {
      "epoch": 0.4103635004964735,
      "grad_norm": 0.22866128385066986,
      "learning_rate": 1.8446979880896437e-05,
      "loss": 0.7984527349472046,
      "step": 5011
    },
    {
      "epoch": 0.41044539303299243,
      "grad_norm": 0.16357088088989258,
      "learning_rate": 1.8444869691029874e-05,
      "loss": 0.8507169485092163,
      "step": 5012
    },
    {
      "epoch": 0.4105272855695114,
      "grad_norm": 0.18283525109291077,
      "learning_rate": 1.844275922104272e-05,
      "loss": 1.0137220621109009,
      "step": 5013
    },
    {
      "epoch": 0.41060917810603037,
      "grad_norm": 0.17136558890342712,
      "learning_rate": 1.8440648471026727e-05,
      "loss": 0.794072687625885,
      "step": 5014
    },
    {
      "epoch": 0.41069107064254934,
      "grad_norm": 0.1886119395494461,
      "learning_rate": 1.8438537441073614e-05,
      "loss": 0.8772546052932739,
      "step": 5015
    },
    {
      "epoch": 0.41077296317906825,
      "grad_norm": 0.1752847284078598,
      "learning_rate": 1.8436426131275148e-05,
      "loss": 0.895545244216919,
      "step": 5016
    },
    {
      "epoch": 0.4108548557155872,
      "grad_norm": 0.17248259484767914,
      "learning_rate": 1.84343145417231e-05,
      "loss": 0.953667402267456,
      "step": 5017
    },
    {
      "epoch": 0.4109367482521062,
      "grad_norm": 0.17045527696609497,
      "learning_rate": 1.8432202672509234e-05,
      "loss": 0.9176346659660339,
      "step": 5018
    },
    {
      "epoch": 0.4110186407886251,
      "grad_norm": 0.1872190684080124,
      "learning_rate": 1.8430090523725348e-05,
      "loss": 0.91285640001297,
      "step": 5019
    },
    {
      "epoch": 0.41110053332514407,
      "grad_norm": 0.18455931544303894,
      "learning_rate": 1.8427978095463243e-05,
      "loss": 0.7770208120346069,
      "step": 5020
    },
    {
      "epoch": 0.41118242586166304,
      "grad_norm": 0.15735073387622833,
      "learning_rate": 1.842586538781474e-05,
      "loss": 0.6984356641769409,
      "step": 5021
    },
    {
      "epoch": 0.411264318398182,
      "grad_norm": 0.1825786828994751,
      "learning_rate": 1.8423752400871654e-05,
      "loss": 0.833448052406311,
      "step": 5022
    },
    {
      "epoch": 0.4113462109347009,
      "grad_norm": 0.17885322868824005,
      "learning_rate": 1.8421639134725835e-05,
      "loss": 0.6901897192001343,
      "step": 5023
    },
    {
      "epoch": 0.4114281034712199,
      "grad_norm": 0.18414925038814545,
      "learning_rate": 1.841952558946913e-05,
      "loss": 0.7179949283599854,
      "step": 5024
    },
    {
      "epoch": 0.41150999600773885,
      "grad_norm": 0.17916403710842133,
      "learning_rate": 1.8417411765193408e-05,
      "loss": 1.1610106229782104,
      "step": 5025
    },
    {
      "epoch": 0.4115918885442578,
      "grad_norm": 0.18813838064670563,
      "learning_rate": 1.841529766199053e-05,
      "loss": 0.8597434759140015,
      "step": 5026
    },
    {
      "epoch": 0.41167378108077674,
      "grad_norm": 0.1834438592195511,
      "learning_rate": 1.8413183279952397e-05,
      "loss": 0.8743181824684143,
      "step": 5027
    },
    {
      "epoch": 0.4117556736172957,
      "grad_norm": 0.18378406763076782,
      "learning_rate": 1.8411068619170906e-05,
      "loss": 0.8814966082572937,
      "step": 5028
    },
    {
      "epoch": 0.41183756615381467,
      "grad_norm": 0.18118485808372498,
      "learning_rate": 1.8408953679737968e-05,
      "loss": 0.9799172282218933,
      "step": 5029
    },
    {
      "epoch": 0.4119194586903336,
      "grad_norm": 0.16234096884727478,
      "learning_rate": 1.8406838461745505e-05,
      "loss": 0.8437620997428894,
      "step": 5030
    },
    {
      "epoch": 0.41200135122685255,
      "grad_norm": 0.1751297265291214,
      "learning_rate": 1.8404722965285456e-05,
      "loss": 0.6059399247169495,
      "step": 5031
    },
    {
      "epoch": 0.4120832437633715,
      "grad_norm": 0.181343674659729,
      "learning_rate": 1.8402607190449766e-05,
      "loss": 0.7560007572174072,
      "step": 5032
    },
    {
      "epoch": 0.4121651362998905,
      "grad_norm": 0.16960038244724274,
      "learning_rate": 1.84004911373304e-05,
      "loss": 0.7235829830169678,
      "step": 5033
    },
    {
      "epoch": 0.4122470288364094,
      "grad_norm": 0.16263945400714874,
      "learning_rate": 1.8398374806019323e-05,
      "loss": 0.6315873265266418,
      "step": 5034
    },
    {
      "epoch": 0.41232892137292837,
      "grad_norm": 0.16230539977550507,
      "learning_rate": 1.8396258196608524e-05,
      "loss": 0.940045952796936,
      "step": 5035
    },
    {
      "epoch": 0.41241081390944734,
      "grad_norm": 0.17629235982894897,
      "learning_rate": 1.8394141309190003e-05,
      "loss": 0.7686565518379211,
      "step": 5036
    },
    {
      "epoch": 0.4124927064459663,
      "grad_norm": 0.14415593445301056,
      "learning_rate": 1.8392024143855764e-05,
      "loss": 0.6679525971412659,
      "step": 5037
    },
    {
      "epoch": 0.4125745989824852,
      "grad_norm": 0.22276516258716583,
      "learning_rate": 1.8389906700697832e-05,
      "loss": 0.7229623794555664,
      "step": 5038
    },
    {
      "epoch": 0.4126564915190042,
      "grad_norm": 0.19961333274841309,
      "learning_rate": 1.8387788979808232e-05,
      "loss": 0.7958537936210632,
      "step": 5039
    },
    {
      "epoch": 0.41273838405552316,
      "grad_norm": 0.1747620701789856,
      "learning_rate": 1.8385670981279017e-05,
      "loss": 0.9156761169433594,
      "step": 5040
    },
    {
      "epoch": 0.41282027659204207,
      "grad_norm": 0.20118634402751923,
      "learning_rate": 1.8383552705202237e-05,
      "loss": 0.9581171870231628,
      "step": 5041
    },
    {
      "epoch": 0.41290216912856104,
      "grad_norm": 0.1832311451435089,
      "learning_rate": 1.8381434151669966e-05,
      "loss": 0.8034359216690063,
      "step": 5042
    },
    {
      "epoch": 0.41298406166508,
      "grad_norm": 0.17819732427597046,
      "learning_rate": 1.8379315320774284e-05,
      "loss": 1.0520960092544556,
      "step": 5043
    },
    {
      "epoch": 0.413065954201599,
      "grad_norm": 0.15707549452781677,
      "learning_rate": 1.8377196212607286e-05,
      "loss": 0.8873629570007324,
      "step": 5044
    },
    {
      "epoch": 0.4131478467381179,
      "grad_norm": 0.14208269119262695,
      "learning_rate": 1.837507682726107e-05,
      "loss": 0.8027006983757019,
      "step": 5045
    },
    {
      "epoch": 0.41322973927463685,
      "grad_norm": 0.21803416311740875,
      "learning_rate": 1.837295716482776e-05,
      "loss": 1.0342291593551636,
      "step": 5046
    },
    {
      "epoch": 0.4133116318111558,
      "grad_norm": 0.1716996133327484,
      "learning_rate": 1.8370837225399484e-05,
      "loss": 0.7716652750968933,
      "step": 5047
    },
    {
      "epoch": 0.4133935243476748,
      "grad_norm": 0.17768065631389618,
      "learning_rate": 1.836871700906838e-05,
      "loss": 0.9595019221305847,
      "step": 5048
    },
    {
      "epoch": 0.4134754168841937,
      "grad_norm": 0.1855856031179428,
      "learning_rate": 1.8366596515926606e-05,
      "loss": 0.9740675687789917,
      "step": 5049
    },
    {
      "epoch": 0.41355730942071267,
      "grad_norm": 0.20895196497440338,
      "learning_rate": 1.8364475746066327e-05,
      "loss": 1.0079102516174316,
      "step": 5050
    },
    {
      "epoch": 0.41363920195723164,
      "grad_norm": 0.20975518226623535,
      "learning_rate": 1.836235469957972e-05,
      "loss": 0.8894243836402893,
      "step": 5051
    },
    {
      "epoch": 0.41372109449375055,
      "grad_norm": 0.19174417853355408,
      "learning_rate": 1.8360233376558967e-05,
      "loss": 0.8962051868438721,
      "step": 5052
    },
    {
      "epoch": 0.4138029870302695,
      "grad_norm": 0.1846431940793991,
      "learning_rate": 1.835811177709628e-05,
      "loss": 0.6075901389122009,
      "step": 5053
    },
    {
      "epoch": 0.4138848795667885,
      "grad_norm": 0.18730808794498444,
      "learning_rate": 1.8355989901283868e-05,
      "loss": 0.981726348400116,
      "step": 5054
    },
    {
      "epoch": 0.41396677210330746,
      "grad_norm": 0.198908731341362,
      "learning_rate": 1.8353867749213958e-05,
      "loss": 0.7639051675796509,
      "step": 5055
    },
    {
      "epoch": 0.41404866463982637,
      "grad_norm": 0.15319058299064636,
      "learning_rate": 1.835174532097878e-05,
      "loss": 0.7714362144470215,
      "step": 5056
    },
    {
      "epoch": 0.41413055717634534,
      "grad_norm": 0.19342046976089478,
      "learning_rate": 1.8349622616670595e-05,
      "loss": 0.8694853782653809,
      "step": 5057
    },
    {
      "epoch": 0.4142124497128643,
      "grad_norm": 0.19516195356845856,
      "learning_rate": 1.8347499636381658e-05,
      "loss": 0.5231842398643494,
      "step": 5058
    },
    {
      "epoch": 0.4142943422493833,
      "grad_norm": 0.1792737990617752,
      "learning_rate": 1.834537638020424e-05,
      "loss": 0.9029706120491028,
      "step": 5059
    },
    {
      "epoch": 0.4143762347859022,
      "grad_norm": 0.19359585642814636,
      "learning_rate": 1.834325284823063e-05,
      "loss": 0.8035160303115845,
      "step": 5060
    },
    {
      "epoch": 0.41445812732242115,
      "grad_norm": 0.17062629759311676,
      "learning_rate": 1.834112904055313e-05,
      "loss": 0.9138513803482056,
      "step": 5061
    },
    {
      "epoch": 0.4145400198589401,
      "grad_norm": 0.23629418015480042,
      "learning_rate": 1.8339004957264042e-05,
      "loss": 1.0429738759994507,
      "step": 5062
    },
    {
      "epoch": 0.41462191239545904,
      "grad_norm": 0.2184596061706543,
      "learning_rate": 1.8336880598455692e-05,
      "loss": 0.7987755537033081,
      "step": 5063
    },
    {
      "epoch": 0.414703804931978,
      "grad_norm": 0.20879052579402924,
      "learning_rate": 1.8334755964220406e-05,
      "loss": 0.8705446720123291,
      "step": 5064
    },
    {
      "epoch": 0.41478569746849697,
      "grad_norm": 0.21428747475147247,
      "learning_rate": 1.8332631054650538e-05,
      "loss": 0.7276716232299805,
      "step": 5065
    },
    {
      "epoch": 0.41486759000501594,
      "grad_norm": 0.18660998344421387,
      "learning_rate": 1.8330505869838446e-05,
      "loss": 0.7780908942222595,
      "step": 5066
    },
    {
      "epoch": 0.41494948254153485,
      "grad_norm": 0.201173335313797,
      "learning_rate": 1.8328380409876493e-05,
      "loss": 0.9767258167266846,
      "step": 5067
    },
    {
      "epoch": 0.4150313750780538,
      "grad_norm": 0.1836206167936325,
      "learning_rate": 1.8326254674857057e-05,
      "loss": 0.7083567976951599,
      "step": 5068
    },
    {
      "epoch": 0.4151132676145728,
      "grad_norm": 0.18724898993968964,
      "learning_rate": 1.8324128664872545e-05,
      "loss": 0.7906497120857239,
      "step": 5069
    },
    {
      "epoch": 0.4151951601510917,
      "grad_norm": 0.17939847707748413,
      "learning_rate": 1.832200238001535e-05,
      "loss": 0.8438933491706848,
      "step": 5070
    },
    {
      "epoch": 0.41527705268761067,
      "grad_norm": 0.1575000137090683,
      "learning_rate": 1.8319875820377897e-05,
      "loss": 1.034073829650879,
      "step": 5071
    },
    {
      "epoch": 0.41535894522412964,
      "grad_norm": 0.18241645395755768,
      "learning_rate": 1.8317748986052606e-05,
      "loss": 0.7690894603729248,
      "step": 5072
    },
    {
      "epoch": 0.4154408377606486,
      "grad_norm": 0.17777486145496368,
      "learning_rate": 1.8315621877131926e-05,
      "loss": 0.932884693145752,
      "step": 5073
    },
    {
      "epoch": 0.4155227302971675,
      "grad_norm": 0.1905587762594223,
      "learning_rate": 1.8313494493708305e-05,
      "loss": 0.8105591535568237,
      "step": 5074
    },
    {
      "epoch": 0.4156046228336865,
      "grad_norm": 0.18785883486270905,
      "learning_rate": 1.8311366835874212e-05,
      "loss": 0.9248709678649902,
      "step": 5075
    },
    {
      "epoch": 0.41568651537020546,
      "grad_norm": 0.17590418457984924,
      "learning_rate": 1.830923890372212e-05,
      "loss": 0.7090432643890381,
      "step": 5076
    },
    {
      "epoch": 0.4157684079067244,
      "grad_norm": 0.19362188875675201,
      "learning_rate": 1.830711069734452e-05,
      "loss": 0.9690373539924622,
      "step": 5077
    },
    {
      "epoch": 0.41585030044324334,
      "grad_norm": 0.19288070499897003,
      "learning_rate": 1.8304982216833915e-05,
      "loss": 0.9549434781074524,
      "step": 5078
    },
    {
      "epoch": 0.4159321929797623,
      "grad_norm": 0.17955997586250305,
      "learning_rate": 1.8302853462282812e-05,
      "loss": 1.018237829208374,
      "step": 5079
    },
    {
      "epoch": 0.4160140855162813,
      "grad_norm": 0.18209630250930786,
      "learning_rate": 1.8300724433783738e-05,
      "loss": 0.9378691911697388,
      "step": 5080
    },
    {
      "epoch": 0.4160959780528002,
      "grad_norm": 0.17391470074653625,
      "learning_rate": 1.8298595131429228e-05,
      "loss": 0.947184681892395,
      "step": 5081
    },
    {
      "epoch": 0.41617787058931915,
      "grad_norm": 0.19187791645526886,
      "learning_rate": 1.8296465555311833e-05,
      "loss": 1.1741151809692383,
      "step": 5082
    },
    {
      "epoch": 0.4162597631258381,
      "grad_norm": 0.19200529158115387,
      "learning_rate": 1.8294335705524114e-05,
      "loss": 0.744033932685852,
      "step": 5083
    },
    {
      "epoch": 0.4163416556623571,
      "grad_norm": 0.18049094080924988,
      "learning_rate": 1.8292205582158634e-05,
      "loss": 0.6440449953079224,
      "step": 5084
    },
    {
      "epoch": 0.416423548198876,
      "grad_norm": 0.18438820540905,
      "learning_rate": 1.829007518530799e-05,
      "loss": 0.6096867918968201,
      "step": 5085
    },
    {
      "epoch": 0.41650544073539497,
      "grad_norm": 0.17387816309928894,
      "learning_rate": 1.828794451506477e-05,
      "loss": 0.8521718978881836,
      "step": 5086
    },
    {
      "epoch": 0.41658733327191394,
      "grad_norm": 0.21966509521007538,
      "learning_rate": 1.828581357152158e-05,
      "loss": 0.903580367565155,
      "step": 5087
    },
    {
      "epoch": 0.4166692258084329,
      "grad_norm": 0.1836947500705719,
      "learning_rate": 1.828368235477105e-05,
      "loss": 0.8803563117980957,
      "step": 5088
    },
    {
      "epoch": 0.4167511183449518,
      "grad_norm": 0.1807958036661148,
      "learning_rate": 1.8281550864905798e-05,
      "loss": 1.0072920322418213,
      "step": 5089
    },
    {
      "epoch": 0.4168330108814708,
      "grad_norm": 0.1803923100233078,
      "learning_rate": 1.8279419102018476e-05,
      "loss": 0.8528308272361755,
      "step": 5090
    },
    {
      "epoch": 0.41691490341798976,
      "grad_norm": 0.20516814291477203,
      "learning_rate": 1.8277287066201737e-05,
      "loss": 0.9267112016677856,
      "step": 5091
    },
    {
      "epoch": 0.41699679595450867,
      "grad_norm": 0.19773732125759125,
      "learning_rate": 1.8275154757548245e-05,
      "loss": 0.809674084186554,
      "step": 5092
    },
    {
      "epoch": 0.41707868849102764,
      "grad_norm": 0.1936405599117279,
      "learning_rate": 1.8273022176150686e-05,
      "loss": 0.9244376420974731,
      "step": 5093
    },
    {
      "epoch": 0.4171605810275466,
      "grad_norm": 0.18972590565681458,
      "learning_rate": 1.8270889322101742e-05,
      "loss": 0.7975389361381531,
      "step": 5094
    },
    {
      "epoch": 0.4172424735640656,
      "grad_norm": 0.18869681656360626,
      "learning_rate": 1.8268756195494123e-05,
      "loss": 0.9766507744789124,
      "step": 5095
    },
    {
      "epoch": 0.4173243661005845,
      "grad_norm": 0.21369363367557526,
      "learning_rate": 1.8266622796420542e-05,
      "loss": 0.9319756627082825,
      "step": 5096
    },
    {
      "epoch": 0.41740625863710346,
      "grad_norm": 0.20810289680957794,
      "learning_rate": 1.8264489124973722e-05,
      "loss": 1.1251598596572876,
      "step": 5097
    },
    {
      "epoch": 0.4174881511736224,
      "grad_norm": 0.17675898969173431,
      "learning_rate": 1.8262355181246406e-05,
      "loss": 0.7165467739105225,
      "step": 5098
    },
    {
      "epoch": 0.4175700437101414,
      "grad_norm": 0.16627661883831024,
      "learning_rate": 1.8260220965331334e-05,
      "loss": 0.7169197797775269,
      "step": 5099
    },
    {
      "epoch": 0.4176519362466603,
      "grad_norm": 0.1938011795282364,
      "learning_rate": 1.825808647732128e-05,
      "loss": 0.8337584733963013,
      "step": 5100
    },
    {
      "epoch": 0.4177338287831793,
      "grad_norm": 0.16278977692127228,
      "learning_rate": 1.8255951717309013e-05,
      "loss": 0.844434380531311,
      "step": 5101
    },
    {
      "epoch": 0.41781572131969824,
      "grad_norm": 0.2081368863582611,
      "learning_rate": 1.825381668538732e-05,
      "loss": 0.7591282725334167,
      "step": 5102
    },
    {
      "epoch": 0.41789761385621715,
      "grad_norm": 0.17814123630523682,
      "learning_rate": 1.825168138164899e-05,
      "loss": 1.009148120880127,
      "step": 5103
    },
    {
      "epoch": 0.4179795063927361,
      "grad_norm": 0.1793467104434967,
      "learning_rate": 1.824954580618684e-05,
      "loss": 0.8043340444564819,
      "step": 5104
    },
    {
      "epoch": 0.4180613989292551,
      "grad_norm": 0.16902127861976624,
      "learning_rate": 1.824740995909369e-05,
      "loss": 0.5436157584190369,
      "step": 5105
    },
    {
      "epoch": 0.41814329146577406,
      "grad_norm": 0.20043110847473145,
      "learning_rate": 1.8245273840462373e-05,
      "loss": 0.7089431881904602,
      "step": 5106
    },
    {
      "epoch": 0.41822518400229297,
      "grad_norm": 0.21063454449176788,
      "learning_rate": 1.8243137450385734e-05,
      "loss": 0.7714178562164307,
      "step": 5107
    },
    {
      "epoch": 0.41830707653881194,
      "grad_norm": 0.17908529937267303,
      "learning_rate": 1.8241000788956625e-05,
      "loss": 0.6990824937820435,
      "step": 5108
    },
    {
      "epoch": 0.4183889690753309,
      "grad_norm": 0.1587780863046646,
      "learning_rate": 1.823886385626792e-05,
      "loss": 0.8366127610206604,
      "step": 5109
    },
    {
      "epoch": 0.4184708616118499,
      "grad_norm": 0.19029811024665833,
      "learning_rate": 1.823672665241249e-05,
      "loss": 0.8395830392837524,
      "step": 5110
    },
    {
      "epoch": 0.4185527541483688,
      "grad_norm": 0.17339667677879333,
      "learning_rate": 1.8234589177483238e-05,
      "loss": 1.0161033868789673,
      "step": 5111
    },
    {
      "epoch": 0.41863464668488776,
      "grad_norm": 0.2131277322769165,
      "learning_rate": 1.823245143157306e-05,
      "loss": 0.7818050384521484,
      "step": 5112
    },
    {
      "epoch": 0.4187165392214067,
      "grad_norm": 0.17857001721858978,
      "learning_rate": 1.8230313414774873e-05,
      "loss": 0.858956515789032,
      "step": 5113
    },
    {
      "epoch": 0.41879843175792564,
      "grad_norm": 0.18848836421966553,
      "learning_rate": 1.8228175127181605e-05,
      "loss": 0.7782373428344727,
      "step": 5114
    },
    {
      "epoch": 0.4188803242944446,
      "grad_norm": 0.2088644951581955,
      "learning_rate": 1.8226036568886196e-05,
      "loss": 0.6854217648506165,
      "step": 5115
    },
    {
      "epoch": 0.4189622168309636,
      "grad_norm": 0.17247751355171204,
      "learning_rate": 1.8223897739981588e-05,
      "loss": 0.6193524599075317,
      "step": 5116
    },
    {
      "epoch": 0.41904410936748254,
      "grad_norm": 0.20458747446537018,
      "learning_rate": 1.8221758640560756e-05,
      "loss": 0.8224255442619324,
      "step": 5117
    },
    {
      "epoch": 0.41912600190400146,
      "grad_norm": 0.1811729073524475,
      "learning_rate": 1.8219619270716666e-05,
      "loss": 0.7989381551742554,
      "step": 5118
    },
    {
      "epoch": 0.4192078944405204,
      "grad_norm": 0.15227505564689636,
      "learning_rate": 1.8217479630542306e-05,
      "loss": 0.6109708547592163,
      "step": 5119
    },
    {
      "epoch": 0.4192897869770394,
      "grad_norm": 0.17618891596794128,
      "learning_rate": 1.8215339720130673e-05,
      "loss": 1.030480146408081,
      "step": 5120
    },
    {
      "epoch": 0.41937167951355836,
      "grad_norm": 0.17340907454490662,
      "learning_rate": 1.8213199539574778e-05,
      "loss": 0.7121025919914246,
      "step": 5121
    },
    {
      "epoch": 0.4194535720500773,
      "grad_norm": 0.19075623154640198,
      "learning_rate": 1.8211059088967635e-05,
      "loss": 0.9103667736053467,
      "step": 5122
    },
    {
      "epoch": 0.41953546458659624,
      "grad_norm": 0.18273910880088806,
      "learning_rate": 1.820891836840229e-05,
      "loss": 0.9746288657188416,
      "step": 5123
    },
    {
      "epoch": 0.4196173571231152,
      "grad_norm": 0.20412300527095795,
      "learning_rate": 1.8206777377971777e-05,
      "loss": 0.6894181370735168,
      "step": 5124
    },
    {
      "epoch": 0.4196992496596341,
      "grad_norm": 0.19168631732463837,
      "learning_rate": 1.8204636117769154e-05,
      "loss": 0.6336537003517151,
      "step": 5125
    },
    {
      "epoch": 0.4197811421961531,
      "grad_norm": 0.18541893362998962,
      "learning_rate": 1.8202494587887494e-05,
      "loss": 0.8271810412406921,
      "step": 5126
    },
    {
      "epoch": 0.41986303473267206,
      "grad_norm": 0.16419188678264618,
      "learning_rate": 1.8200352788419868e-05,
      "loss": 1.0343014001846313,
      "step": 5127
    },
    {
      "epoch": 0.419944927269191,
      "grad_norm": 0.1834053099155426,
      "learning_rate": 1.8198210719459376e-05,
      "loss": 0.6832797527313232,
      "step": 5128
    },
    {
      "epoch": 0.42002681980570994,
      "grad_norm": 0.15223999321460724,
      "learning_rate": 1.819606838109912e-05,
      "loss": 0.9008648991584778,
      "step": 5129
    },
    {
      "epoch": 0.4201087123422289,
      "grad_norm": 0.16065111756324768,
      "learning_rate": 1.8193925773432205e-05,
      "loss": 0.7688698768615723,
      "step": 5130
    },
    {
      "epoch": 0.4201906048787479,
      "grad_norm": 0.1646135151386261,
      "learning_rate": 1.819178289655177e-05,
      "loss": 0.9256729483604431,
      "step": 5131
    },
    {
      "epoch": 0.42027249741526684,
      "grad_norm": 0.1854640245437622,
      "learning_rate": 1.8189639750550947e-05,
      "loss": 0.9076009392738342,
      "step": 5132
    },
    {
      "epoch": 0.42035438995178576,
      "grad_norm": 0.15605102479457855,
      "learning_rate": 1.8187496335522887e-05,
      "loss": 1.2095659971237183,
      "step": 5133
    },
    {
      "epoch": 0.4204362824883047,
      "grad_norm": 0.18665140867233276,
      "learning_rate": 1.8185352651560754e-05,
      "loss": 0.6672487854957581,
      "step": 5134
    },
    {
      "epoch": 0.4205181750248237,
      "grad_norm": 0.16953518986701965,
      "learning_rate": 1.818320869875772e-05,
      "loss": 0.7401173710823059,
      "step": 5135
    },
    {
      "epoch": 0.4206000675613426,
      "grad_norm": 0.18107870221138,
      "learning_rate": 1.8181064477206964e-05,
      "loss": 0.9411282539367676,
      "step": 5136
    },
    {
      "epoch": 0.4206819600978616,
      "grad_norm": 0.22464562952518463,
      "learning_rate": 1.817891998700169e-05,
      "loss": 1.0147037506103516,
      "step": 5137
    },
    {
      "epoch": 0.42076385263438054,
      "grad_norm": 0.18624348938465118,
      "learning_rate": 1.8176775228235105e-05,
      "loss": 0.776630163192749,
      "step": 5138
    },
    {
      "epoch": 0.4208457451708995,
      "grad_norm": 0.20334188640117645,
      "learning_rate": 1.8174630201000426e-05,
      "loss": 0.8975825905799866,
      "step": 5139
    },
    {
      "epoch": 0.4209276377074184,
      "grad_norm": 0.20415005087852478,
      "learning_rate": 1.8172484905390894e-05,
      "loss": 1.018853783607483,
      "step": 5140
    },
    {
      "epoch": 0.4210095302439374,
      "grad_norm": 0.1865503340959549,
      "learning_rate": 1.8170339341499738e-05,
      "loss": 0.5734263062477112,
      "step": 5141
    },
    {
      "epoch": 0.42109142278045636,
      "grad_norm": 0.15297000110149384,
      "learning_rate": 1.816819350942022e-05,
      "loss": 1.0201913118362427,
      "step": 5142
    },
    {
      "epoch": 0.42117331531697527,
      "grad_norm": 0.23985448479652405,
      "learning_rate": 1.8166047409245612e-05,
      "loss": 0.9153088331222534,
      "step": 5143
    },
    {
      "epoch": 0.42125520785349424,
      "grad_norm": 0.159892737865448,
      "learning_rate": 1.8163901041069185e-05,
      "loss": 0.6670908331871033,
      "step": 5144
    },
    {
      "epoch": 0.4213371003900132,
      "grad_norm": 0.19944003224372864,
      "learning_rate": 1.8161754404984237e-05,
      "loss": 0.7307386994361877,
      "step": 5145
    },
    {
      "epoch": 0.4214189929265322,
      "grad_norm": 0.1786630004644394,
      "learning_rate": 1.815960750108406e-05,
      "loss": 0.8546532988548279,
      "step": 5146
    },
    {
      "epoch": 0.4215008854630511,
      "grad_norm": 0.17497839033603668,
      "learning_rate": 1.8157460329461966e-05,
      "loss": 0.9227883815765381,
      "step": 5147
    },
    {
      "epoch": 0.42158277799957006,
      "grad_norm": 0.15835429728031158,
      "learning_rate": 1.815531289021129e-05,
      "loss": 0.6922209858894348,
      "step": 5148
    },
    {
      "epoch": 0.421664670536089,
      "grad_norm": 0.18352779746055603,
      "learning_rate": 1.8153165183425366e-05,
      "loss": 0.6159884333610535,
      "step": 5149
    },
    {
      "epoch": 0.421746563072608,
      "grad_norm": 0.15086840093135834,
      "learning_rate": 1.8151017209197537e-05,
      "loss": 0.6185681223869324,
      "step": 5150
    },
    {
      "epoch": 0.4218284556091269,
      "grad_norm": 0.27050885558128357,
      "learning_rate": 1.814886896762117e-05,
      "loss": 0.9331297874450684,
      "step": 5151
    },
    {
      "epoch": 0.4219103481456459,
      "grad_norm": 0.18095380067825317,
      "learning_rate": 1.8146720458789628e-05,
      "loss": 0.8188208937644958,
      "step": 5152
    },
    {
      "epoch": 0.42199224068216484,
      "grad_norm": 0.2031712681055069,
      "learning_rate": 1.81445716827963e-05,
      "loss": 0.8433481454849243,
      "step": 5153
    },
    {
      "epoch": 0.42207413321868376,
      "grad_norm": 0.19730126857757568,
      "learning_rate": 1.8142422639734577e-05,
      "loss": 0.6817410588264465,
      "step": 5154
    },
    {
      "epoch": 0.4221560257552027,
      "grad_norm": 0.20351417362689972,
      "learning_rate": 1.8140273329697867e-05,
      "loss": 0.9335296154022217,
      "step": 5155
    },
    {
      "epoch": 0.4222379182917217,
      "grad_norm": 0.23672421276569366,
      "learning_rate": 1.8138123752779595e-05,
      "loss": 0.847127377986908,
      "step": 5156
    },
    {
      "epoch": 0.42231981082824066,
      "grad_norm": 0.15737968683242798,
      "learning_rate": 1.813597390907318e-05,
      "loss": 1.0522080659866333,
      "step": 5157
    },
    {
      "epoch": 0.4224017033647596,
      "grad_norm": 0.19796104729175568,
      "learning_rate": 1.813382379867206e-05,
      "loss": 1.0770142078399658,
      "step": 5158
    },
    {
      "epoch": 0.42248359590127854,
      "grad_norm": 0.2659813165664673,
      "learning_rate": 1.8131673421669702e-05,
      "loss": 0.826482355594635,
      "step": 5159
    },
    {
      "epoch": 0.4225654884377975,
      "grad_norm": 0.16272585093975067,
      "learning_rate": 1.812952277815956e-05,
      "loss": 1.1627418994903564,
      "step": 5160
    },
    {
      "epoch": 0.4226473809743165,
      "grad_norm": 0.168051615357399,
      "learning_rate": 1.8127371868235116e-05,
      "loss": 1.096964955329895,
      "step": 5161
    },
    {
      "epoch": 0.4227292735108354,
      "grad_norm": 0.16542716324329376,
      "learning_rate": 1.812522069198985e-05,
      "loss": 0.7036202549934387,
      "step": 5162
    },
    {
      "epoch": 0.42281116604735436,
      "grad_norm": 0.18782448768615723,
      "learning_rate": 1.8123069249517263e-05,
      "loss": 1.015702724456787,
      "step": 5163
    },
    {
      "epoch": 0.4228930585838733,
      "grad_norm": 0.19121849536895752,
      "learning_rate": 1.812091754091087e-05,
      "loss": 0.7503167390823364,
      "step": 5164
    },
    {
      "epoch": 0.42297495112039224,
      "grad_norm": 0.17587770521640778,
      "learning_rate": 1.8118765566264193e-05,
      "loss": 0.5838730335235596,
      "step": 5165
    },
    {
      "epoch": 0.4230568436569112,
      "grad_norm": 0.20884041488170624,
      "learning_rate": 1.8116613325670762e-05,
      "loss": 0.634531557559967,
      "step": 5166
    },
    {
      "epoch": 0.4231387361934302,
      "grad_norm": 0.1666855663061142,
      "learning_rate": 1.811446081922412e-05,
      "loss": 0.7154521942138672,
      "step": 5167
    },
    {
      "epoch": 0.42322062872994914,
      "grad_norm": 0.14254380762577057,
      "learning_rate": 1.8112308047017828e-05,
      "loss": 0.6127246618270874,
      "step": 5168
    },
    {
      "epoch": 0.42330252126646806,
      "grad_norm": 0.15958784520626068,
      "learning_rate": 1.8110155009145457e-05,
      "loss": 0.8781707286834717,
      "step": 5169
    },
    {
      "epoch": 0.423384413802987,
      "grad_norm": 0.16516119241714478,
      "learning_rate": 1.810800170570058e-05,
      "loss": 1.0303021669387817,
      "step": 5170
    },
    {
      "epoch": 0.423466306339506,
      "grad_norm": 0.18929176032543182,
      "learning_rate": 1.8105848136776793e-05,
      "loss": 0.9102661609649658,
      "step": 5171
    },
    {
      "epoch": 0.42354819887602496,
      "grad_norm": 0.14856991171836853,
      "learning_rate": 1.8103694302467698e-05,
      "loss": 0.766535222530365,
      "step": 5172
    },
    {
      "epoch": 0.4236300914125439,
      "grad_norm": 0.19249460101127625,
      "learning_rate": 1.810154020286691e-05,
      "loss": 1.1857414245605469,
      "step": 5173
    },
    {
      "epoch": 0.42371198394906284,
      "grad_norm": 0.168783038854599,
      "learning_rate": 1.8099385838068053e-05,
      "loss": 0.7672098875045776,
      "step": 5174
    },
    {
      "epoch": 0.4237938764855818,
      "grad_norm": 0.18971367180347443,
      "learning_rate": 1.8097231208164766e-05,
      "loss": 0.7717016339302063,
      "step": 5175
    },
    {
      "epoch": 0.4238757690221007,
      "grad_norm": 0.18256278336048126,
      "learning_rate": 1.80950763132507e-05,
      "loss": 0.900749921798706,
      "step": 5176
    },
    {
      "epoch": 0.4239576615586197,
      "grad_norm": 0.20790952444076538,
      "learning_rate": 1.8092921153419512e-05,
      "loss": 0.8541547060012817,
      "step": 5177
    },
    {
      "epoch": 0.42403955409513866,
      "grad_norm": 0.1926247924566269,
      "learning_rate": 1.8090765728764875e-05,
      "loss": 1.0725033283233643,
      "step": 5178
    },
    {
      "epoch": 0.42412144663165763,
      "grad_norm": 0.18151718378067017,
      "learning_rate": 1.8088610039380475e-05,
      "loss": 0.728358805179596,
      "step": 5179
    },
    {
      "epoch": 0.42420333916817654,
      "grad_norm": 0.1894334852695465,
      "learning_rate": 1.8086454085360004e-05,
      "loss": 0.9969973564147949,
      "step": 5180
    },
    {
      "epoch": 0.4242852317046955,
      "grad_norm": 0.17966806888580322,
      "learning_rate": 1.808429786679717e-05,
      "loss": 0.793081521987915,
      "step": 5181
    },
    {
      "epoch": 0.4243671242412145,
      "grad_norm": 0.15464387834072113,
      "learning_rate": 1.808214138378569e-05,
      "loss": 0.534359335899353,
      "step": 5182
    },
    {
      "epoch": 0.42444901677773345,
      "grad_norm": 0.16161811351776123,
      "learning_rate": 1.80799846364193e-05,
      "loss": 0.654643714427948,
      "step": 5183
    },
    {
      "epoch": 0.42453090931425236,
      "grad_norm": 0.16348862648010254,
      "learning_rate": 1.807782762479173e-05,
      "loss": 0.5545058250427246,
      "step": 5184
    },
    {
      "epoch": 0.4246128018507713,
      "grad_norm": 0.18262702226638794,
      "learning_rate": 1.8075670348996746e-05,
      "loss": 0.6415174007415771,
      "step": 5185
    },
    {
      "epoch": 0.4246946943872903,
      "grad_norm": 0.1428273618221283,
      "learning_rate": 1.80735128091281e-05,
      "loss": 0.6356614828109741,
      "step": 5186
    },
    {
      "epoch": 0.4247765869238092,
      "grad_norm": 0.18506789207458496,
      "learning_rate": 1.807135500527957e-05,
      "loss": 1.241161584854126,
      "step": 5187
    },
    {
      "epoch": 0.4248584794603282,
      "grad_norm": 0.21634933352470398,
      "learning_rate": 1.8069196937544952e-05,
      "loss": 0.7496830821037292,
      "step": 5188
    },
    {
      "epoch": 0.42494037199684714,
      "grad_norm": 0.19860310852527618,
      "learning_rate": 1.8067038606018036e-05,
      "loss": 0.9316244721412659,
      "step": 5189
    },
    {
      "epoch": 0.4250222645333661,
      "grad_norm": 0.1529722362756729,
      "learning_rate": 1.806488001079263e-05,
      "loss": 0.8022573590278625,
      "step": 5190
    },
    {
      "epoch": 0.425104157069885,
      "grad_norm": 0.24162186682224274,
      "learning_rate": 1.8062721151962562e-05,
      "loss": 1.0487141609191895,
      "step": 5191
    },
    {
      "epoch": 0.425186049606404,
      "grad_norm": 0.2286836802959442,
      "learning_rate": 1.8060562029621664e-05,
      "loss": 1.109235405921936,
      "step": 5192
    },
    {
      "epoch": 0.42526794214292296,
      "grad_norm": 0.19610245525836945,
      "learning_rate": 1.8058402643863772e-05,
      "loss": 0.9658858180046082,
      "step": 5193
    },
    {
      "epoch": 0.42534983467944193,
      "grad_norm": 0.2430948168039322,
      "learning_rate": 1.8056242994782754e-05,
      "loss": 1.4655661582946777,
      "step": 5194
    },
    {
      "epoch": 0.42543172721596084,
      "grad_norm": 0.1682863086462021,
      "learning_rate": 1.805408308247247e-05,
      "loss": 0.6658735871315002,
      "step": 5195
    },
    {
      "epoch": 0.4255136197524798,
      "grad_norm": 0.19950735569000244,
      "learning_rate": 1.80519229070268e-05,
      "loss": 0.9865747690200806,
      "step": 5196
    },
    {
      "epoch": 0.4255955122889988,
      "grad_norm": 0.15903159976005554,
      "learning_rate": 1.804976246853964e-05,
      "loss": 0.7308022975921631,
      "step": 5197
    },
    {
      "epoch": 0.4256774048255177,
      "grad_norm": 0.20391543209552765,
      "learning_rate": 1.8047601767104878e-05,
      "loss": 1.0937438011169434,
      "step": 5198
    },
    {
      "epoch": 0.42575929736203666,
      "grad_norm": 0.2078532576560974,
      "learning_rate": 1.8045440802816436e-05,
      "loss": 1.2011927366256714,
      "step": 5199
    },
    {
      "epoch": 0.4258411898985556,
      "grad_norm": 0.20719796419143677,
      "learning_rate": 1.8043279575768243e-05,
      "loss": 0.8820013999938965,
      "step": 5200
    },
    {
      "epoch": 0.4259230824350746,
      "grad_norm": 0.1719372719526291,
      "learning_rate": 1.8041118086054223e-05,
      "loss": 0.7421293258666992,
      "step": 5201
    },
    {
      "epoch": 0.4260049749715935,
      "grad_norm": 0.17443060874938965,
      "learning_rate": 1.8038956333768336e-05,
      "loss": 0.9194854497909546,
      "step": 5202
    },
    {
      "epoch": 0.4260868675081125,
      "grad_norm": 0.1749100685119629,
      "learning_rate": 1.803679431900453e-05,
      "loss": 0.8144937753677368,
      "step": 5203
    },
    {
      "epoch": 0.42616876004463145,
      "grad_norm": 0.16796676814556122,
      "learning_rate": 1.8034632041856776e-05,
      "loss": 0.9969911575317383,
      "step": 5204
    },
    {
      "epoch": 0.42625065258115036,
      "grad_norm": 0.20621591806411743,
      "learning_rate": 1.8032469502419066e-05,
      "loss": 0.8193541169166565,
      "step": 5205
    },
    {
      "epoch": 0.4263325451176693,
      "grad_norm": 0.18196383118629456,
      "learning_rate": 1.8030306700785383e-05,
      "loss": 0.9046236872673035,
      "step": 5206
    },
    {
      "epoch": 0.4264144376541883,
      "grad_norm": 0.21178656816482544,
      "learning_rate": 1.8028143637049728e-05,
      "loss": 0.8544080257415771,
      "step": 5207
    },
    {
      "epoch": 0.42649633019070726,
      "grad_norm": 0.1592809408903122,
      "learning_rate": 1.802598031130613e-05,
      "loss": 0.7195807695388794,
      "step": 5208
    },
    {
      "epoch": 0.4265782227272262,
      "grad_norm": 0.14795784652233124,
      "learning_rate": 1.8023816723648604e-05,
      "loss": 0.8261533975601196,
      "step": 5209
    },
    {
      "epoch": 0.42666011526374514,
      "grad_norm": 0.18316765129566193,
      "learning_rate": 1.802165287417119e-05,
      "loss": 1.1495039463043213,
      "step": 5210
    },
    {
      "epoch": 0.4267420078002641,
      "grad_norm": 0.1899278163909912,
      "learning_rate": 1.801948876296795e-05,
      "loss": 0.6749752163887024,
      "step": 5211
    },
    {
      "epoch": 0.4268239003367831,
      "grad_norm": 0.17670032382011414,
      "learning_rate": 1.8017324390132927e-05,
      "loss": 0.6554186344146729,
      "step": 5212
    },
    {
      "epoch": 0.426905792873302,
      "grad_norm": 0.20175275206565857,
      "learning_rate": 1.801515975576021e-05,
      "loss": 0.5411480069160461,
      "step": 5213
    },
    {
      "epoch": 0.42698768540982096,
      "grad_norm": 0.2086649239063263,
      "learning_rate": 1.801299485994387e-05,
      "loss": 0.9328161478042603,
      "step": 5214
    },
    {
      "epoch": 0.42706957794633993,
      "grad_norm": 0.17418532073497772,
      "learning_rate": 1.8010829702778008e-05,
      "loss": 0.6841269135475159,
      "step": 5215
    },
    {
      "epoch": 0.42715147048285884,
      "grad_norm": 0.17946818470954895,
      "learning_rate": 1.8008664284356734e-05,
      "loss": 0.774822473526001,
      "step": 5216
    },
    {
      "epoch": 0.4272333630193778,
      "grad_norm": 0.19635938107967377,
      "learning_rate": 1.8006498604774158e-05,
      "loss": 1.1369215250015259,
      "step": 5217
    },
    {
      "epoch": 0.4273152555558968,
      "grad_norm": 0.22370025515556335,
      "learning_rate": 1.8004332664124416e-05,
      "loss": 1.0345852375030518,
      "step": 5218
    },
    {
      "epoch": 0.42739714809241575,
      "grad_norm": 0.22223836183547974,
      "learning_rate": 1.800216646250165e-05,
      "loss": 0.8757798075675964,
      "step": 5219
    },
    {
      "epoch": 0.42747904062893466,
      "grad_norm": 0.16669832170009613,
      "learning_rate": 1.8e-05,
      "loss": 0.723968505859375,
      "step": 5220
    },
    {
      "epoch": 0.4275609331654536,
      "grad_norm": 0.22766569256782532,
      "learning_rate": 1.7997833276713642e-05,
      "loss": 0.668658971786499,
      "step": 5221
    },
    {
      "epoch": 0.4276428257019726,
      "grad_norm": 0.1738009750843048,
      "learning_rate": 1.799566629273675e-05,
      "loss": 1.0621371269226074,
      "step": 5222
    },
    {
      "epoch": 0.42772471823849156,
      "grad_norm": 0.1554383486509323,
      "learning_rate": 1.7993499048163503e-05,
      "loss": 0.7264142036437988,
      "step": 5223
    },
    {
      "epoch": 0.4278066107750105,
      "grad_norm": 0.17040498554706573,
      "learning_rate": 1.79913315430881e-05,
      "loss": 0.8015718460083008,
      "step": 5224
    },
    {
      "epoch": 0.42788850331152944,
      "grad_norm": 0.21791017055511475,
      "learning_rate": 1.7989163777604756e-05,
      "loss": 0.9434969425201416,
      "step": 5225
    },
    {
      "epoch": 0.4279703958480484,
      "grad_norm": 0.2375526875257492,
      "learning_rate": 1.7986995751807686e-05,
      "loss": 0.8470356464385986,
      "step": 5226
    },
    {
      "epoch": 0.4280522883845673,
      "grad_norm": 0.184286966919899,
      "learning_rate": 1.7984827465791117e-05,
      "loss": 0.8053234219551086,
      "step": 5227
    },
    {
      "epoch": 0.4281341809210863,
      "grad_norm": 0.18212683498859406,
      "learning_rate": 1.79826589196493e-05,
      "loss": 0.8717324137687683,
      "step": 5228
    },
    {
      "epoch": 0.42821607345760526,
      "grad_norm": 0.22836649417877197,
      "learning_rate": 1.798049011347649e-05,
      "loss": 1.044623613357544,
      "step": 5229
    },
    {
      "epoch": 0.42829796599412423,
      "grad_norm": 0.17592276632785797,
      "learning_rate": 1.797832104736694e-05,
      "loss": 0.6819548606872559,
      "step": 5230
    },
    {
      "epoch": 0.42837985853064314,
      "grad_norm": 0.20201385021209717,
      "learning_rate": 1.7976151721414937e-05,
      "loss": 0.9130160212516785,
      "step": 5231
    },
    {
      "epoch": 0.4284617510671621,
      "grad_norm": 0.24081097543239594,
      "learning_rate": 1.7973982135714768e-05,
      "loss": 0.9813926815986633,
      "step": 5232
    },
    {
      "epoch": 0.4285436436036811,
      "grad_norm": 0.17714235186576843,
      "learning_rate": 1.7971812290360732e-05,
      "loss": 0.773067057132721,
      "step": 5233
    },
    {
      "epoch": 0.42862553614020005,
      "grad_norm": 0.18605205416679382,
      "learning_rate": 1.7969642185447133e-05,
      "loss": 0.7597056031227112,
      "step": 5234
    },
    {
      "epoch": 0.42870742867671896,
      "grad_norm": 0.22878563404083252,
      "learning_rate": 1.7967471821068302e-05,
      "loss": 0.9603187441825867,
      "step": 5235
    },
    {
      "epoch": 0.42878932121323793,
      "grad_norm": 0.19997872412204742,
      "learning_rate": 1.7965301197318563e-05,
      "loss": 0.9045911431312561,
      "step": 5236
    },
    {
      "epoch": 0.4288712137497569,
      "grad_norm": 0.2051328718662262,
      "learning_rate": 1.796313031429227e-05,
      "loss": 1.0697424411773682,
      "step": 5237
    },
    {
      "epoch": 0.4289531062862758,
      "grad_norm": 0.15966010093688965,
      "learning_rate": 1.796095917208377e-05,
      "loss": 0.7443054914474487,
      "step": 5238
    },
    {
      "epoch": 0.4290349988227948,
      "grad_norm": 0.21561506390571594,
      "learning_rate": 1.795878777078743e-05,
      "loss": 0.9222891926765442,
      "step": 5239
    },
    {
      "epoch": 0.42911689135931375,
      "grad_norm": 0.16359250247478485,
      "learning_rate": 1.7956616110497637e-05,
      "loss": 0.7263847589492798,
      "step": 5240
    },
    {
      "epoch": 0.4291987838958327,
      "grad_norm": 0.18921780586242676,
      "learning_rate": 1.7954444191308773e-05,
      "loss": 0.8206148147583008,
      "step": 5241
    },
    {
      "epoch": 0.4292806764323516,
      "grad_norm": 0.22242067754268646,
      "learning_rate": 1.7952272013315237e-05,
      "loss": 1.5340698957443237,
      "step": 5242
    },
    {
      "epoch": 0.4293625689688706,
      "grad_norm": 0.1659547984600067,
      "learning_rate": 1.7950099576611445e-05,
      "loss": 0.644098162651062,
      "step": 5243
    },
    {
      "epoch": 0.42944446150538956,
      "grad_norm": 0.16177275776863098,
      "learning_rate": 1.794792688129182e-05,
      "loss": 0.7638753652572632,
      "step": 5244
    },
    {
      "epoch": 0.42952635404190853,
      "grad_norm": 0.15325792133808136,
      "learning_rate": 1.7945753927450796e-05,
      "loss": 0.7638987302780151,
      "step": 5245
    },
    {
      "epoch": 0.42960824657842744,
      "grad_norm": 0.14615435898303986,
      "learning_rate": 1.7943580715182813e-05,
      "loss": 0.7906946539878845,
      "step": 5246
    },
    {
      "epoch": 0.4296901391149464,
      "grad_norm": 0.1929064393043518,
      "learning_rate": 1.7941407244582335e-05,
      "loss": 0.9481295347213745,
      "step": 5247
    },
    {
      "epoch": 0.4297720316514654,
      "grad_norm": 0.19395139813423157,
      "learning_rate": 1.7939233515743826e-05,
      "loss": 0.9269990921020508,
      "step": 5248
    },
    {
      "epoch": 0.4298539241879843,
      "grad_norm": 0.19673453271389008,
      "learning_rate": 1.793705952876177e-05,
      "loss": 0.9051194190979004,
      "step": 5249
    },
    {
      "epoch": 0.42993581672450326,
      "grad_norm": 0.19067056477069855,
      "learning_rate": 1.793488528373065e-05,
      "loss": 1.0638176202774048,
      "step": 5250
    },
    {
      "epoch": 0.43001770926102223,
      "grad_norm": 0.19593147933483124,
      "learning_rate": 1.7932710780744975e-05,
      "loss": 0.9559282064437866,
      "step": 5251
    },
    {
      "epoch": 0.4300996017975412,
      "grad_norm": 0.19393077492713928,
      "learning_rate": 1.793053601989925e-05,
      "loss": 0.7309958934783936,
      "step": 5252
    },
    {
      "epoch": 0.4301814943340601,
      "grad_norm": 0.1893109530210495,
      "learning_rate": 1.7928361001288008e-05,
      "loss": 0.653713047504425,
      "step": 5253
    },
    {
      "epoch": 0.4302633868705791,
      "grad_norm": 0.25740256905555725,
      "learning_rate": 1.792618572500578e-05,
      "loss": 0.8142497539520264,
      "step": 5254
    },
    {
      "epoch": 0.43034527940709805,
      "grad_norm": 0.19734449684619904,
      "learning_rate": 1.7924010191147107e-05,
      "loss": 1.0365540981292725,
      "step": 5255
    },
    {
      "epoch": 0.430427171943617,
      "grad_norm": 0.2064865082502365,
      "learning_rate": 1.7921834399806556e-05,
      "loss": 0.9764235019683838,
      "step": 5256
    },
    {
      "epoch": 0.43050906448013593,
      "grad_norm": 0.17908696830272675,
      "learning_rate": 1.791965835107869e-05,
      "loss": 0.6567720174789429,
      "step": 5257
    },
    {
      "epoch": 0.4305909570166549,
      "grad_norm": 0.16605861485004425,
      "learning_rate": 1.791748204505809e-05,
      "loss": 0.7692967057228088,
      "step": 5258
    },
    {
      "epoch": 0.43067284955317386,
      "grad_norm": 0.17985761165618896,
      "learning_rate": 1.791530548183935e-05,
      "loss": 0.6868066191673279,
      "step": 5259
    },
    {
      "epoch": 0.4307547420896928,
      "grad_norm": 0.17758452892303467,
      "learning_rate": 1.791312866151707e-05,
      "loss": 0.9222252368927002,
      "step": 5260
    },
    {
      "epoch": 0.43083663462621175,
      "grad_norm": 0.16254809498786926,
      "learning_rate": 1.7910951584185865e-05,
      "loss": 0.5568316578865051,
      "step": 5261
    },
    {
      "epoch": 0.4309185271627307,
      "grad_norm": 0.18595828115940094,
      "learning_rate": 1.7908774249940357e-05,
      "loss": 0.7172431349754333,
      "step": 5262
    },
    {
      "epoch": 0.4310004196992497,
      "grad_norm": 0.19052202999591827,
      "learning_rate": 1.7906596658875188e-05,
      "loss": 0.9660304188728333,
      "step": 5263
    },
    {
      "epoch": 0.4310823122357686,
      "grad_norm": 0.17811733484268188,
      "learning_rate": 1.790441881108499e-05,
      "loss": 0.8613092303276062,
      "step": 5264
    },
    {
      "epoch": 0.43116420477228756,
      "grad_norm": 0.19627846777439117,
      "learning_rate": 1.7902240706664442e-05,
      "loss": 0.8554889559745789,
      "step": 5265
    },
    {
      "epoch": 0.43124609730880653,
      "grad_norm": 0.17916545271873474,
      "learning_rate": 1.79000623457082e-05,
      "loss": 1.122860074043274,
      "step": 5266
    },
    {
      "epoch": 0.4313279898453255,
      "grad_norm": 0.18732695281505585,
      "learning_rate": 1.789788372831095e-05,
      "loss": 0.9140031337738037,
      "step": 5267
    },
    {
      "epoch": 0.4314098823818444,
      "grad_norm": 0.19233152270317078,
      "learning_rate": 1.789570485456738e-05,
      "loss": 0.9495373964309692,
      "step": 5268
    },
    {
      "epoch": 0.4314917749183634,
      "grad_norm": 0.17256949841976166,
      "learning_rate": 1.7893525724572192e-05,
      "loss": 0.8867323994636536,
      "step": 5269
    },
    {
      "epoch": 0.43157366745488235,
      "grad_norm": 0.21394529938697815,
      "learning_rate": 1.7891346338420103e-05,
      "loss": 0.8856604099273682,
      "step": 5270
    },
    {
      "epoch": 0.43165555999140126,
      "grad_norm": 0.19805340468883514,
      "learning_rate": 1.788916669620584e-05,
      "loss": 0.8674989342689514,
      "step": 5271
    },
    {
      "epoch": 0.43173745252792023,
      "grad_norm": 0.20426052808761597,
      "learning_rate": 1.7886986798024138e-05,
      "loss": 0.5844146609306335,
      "step": 5272
    },
    {
      "epoch": 0.4318193450644392,
      "grad_norm": 0.21376396715641022,
      "learning_rate": 1.788480664396974e-05,
      "loss": 0.7621214985847473,
      "step": 5273
    },
    {
      "epoch": 0.43190123760095817,
      "grad_norm": 0.19605118036270142,
      "learning_rate": 1.7882626234137408e-05,
      "loss": 0.8814871907234192,
      "step": 5274
    },
    {
      "epoch": 0.4319831301374771,
      "grad_norm": 0.22234995663166046,
      "learning_rate": 1.788044556862191e-05,
      "loss": 1.3437964916229248,
      "step": 5275
    },
    {
      "epoch": 0.43206502267399605,
      "grad_norm": 0.1702258288860321,
      "learning_rate": 1.787826464751803e-05,
      "loss": 0.7519149780273438,
      "step": 5276
    },
    {
      "epoch": 0.432146915210515,
      "grad_norm": 0.20419761538505554,
      "learning_rate": 1.7876083470920557e-05,
      "loss": 1.1047908067703247,
      "step": 5277
    },
    {
      "epoch": 0.4322288077470339,
      "grad_norm": 0.16860273480415344,
      "learning_rate": 1.7873902038924292e-05,
      "loss": 0.9084750413894653,
      "step": 5278
    },
    {
      "epoch": 0.4323107002835529,
      "grad_norm": 0.20364035665988922,
      "learning_rate": 1.7871720351624056e-05,
      "loss": 0.8341184258460999,
      "step": 5279
    },
    {
      "epoch": 0.43239259282007186,
      "grad_norm": 0.16052241623401642,
      "learning_rate": 1.7869538409114667e-05,
      "loss": 0.7386298179626465,
      "step": 5280
    },
    {
      "epoch": 0.43247448535659083,
      "grad_norm": 0.19251850247383118,
      "learning_rate": 1.786735621149096e-05,
      "loss": 0.7212740182876587,
      "step": 5281
    },
    {
      "epoch": 0.43255637789310974,
      "grad_norm": 0.1846843957901001,
      "learning_rate": 1.7865173758847797e-05,
      "loss": 0.924606442451477,
      "step": 5282
    },
    {
      "epoch": 0.4326382704296287,
      "grad_norm": 0.19072383642196655,
      "learning_rate": 1.786299105128002e-05,
      "loss": 1.2306139469146729,
      "step": 5283
    },
    {
      "epoch": 0.4327201629661477,
      "grad_norm": 0.18118098378181458,
      "learning_rate": 1.78608080888825e-05,
      "loss": 0.679756224155426,
      "step": 5284
    },
    {
      "epoch": 0.43280205550266665,
      "grad_norm": 0.20475806295871735,
      "learning_rate": 1.7858624871750125e-05,
      "loss": 0.6254398822784424,
      "step": 5285
    },
    {
      "epoch": 0.43288394803918556,
      "grad_norm": 0.1910928636789322,
      "learning_rate": 1.785644139997778e-05,
      "loss": 0.7806975245475769,
      "step": 5286
    },
    {
      "epoch": 0.43296584057570453,
      "grad_norm": 0.2629445791244507,
      "learning_rate": 1.785425767366038e-05,
      "loss": 0.6889681816101074,
      "step": 5287
    },
    {
      "epoch": 0.4330477331122235,
      "grad_norm": 0.226944699883461,
      "learning_rate": 1.785207369289282e-05,
      "loss": 0.8932806849479675,
      "step": 5288
    },
    {
      "epoch": 0.4331296256487424,
      "grad_norm": 0.1821507215499878,
      "learning_rate": 1.7849889457770035e-05,
      "loss": 0.6528094410896301,
      "step": 5289
    },
    {
      "epoch": 0.4332115181852614,
      "grad_norm": 0.1574363261461258,
      "learning_rate": 1.7847704968386964e-05,
      "loss": 0.8717193603515625,
      "step": 5290
    },
    {
      "epoch": 0.43329341072178035,
      "grad_norm": 0.21368974447250366,
      "learning_rate": 1.7845520224838552e-05,
      "loss": 0.8764718174934387,
      "step": 5291
    },
    {
      "epoch": 0.4333753032582993,
      "grad_norm": 0.16383175551891327,
      "learning_rate": 1.7843335227219747e-05,
      "loss": 0.7858221530914307,
      "step": 5292
    },
    {
      "epoch": 0.43345719579481823,
      "grad_norm": 0.19847773015499115,
      "learning_rate": 1.7841149975625534e-05,
      "loss": 0.740852952003479,
      "step": 5293
    },
    {
      "epoch": 0.4335390883313372,
      "grad_norm": 0.1982891857624054,
      "learning_rate": 1.7838964470150884e-05,
      "loss": 0.8177998661994934,
      "step": 5294
    },
    {
      "epoch": 0.43362098086785617,
      "grad_norm": 0.18227092921733856,
      "learning_rate": 1.7836778710890787e-05,
      "loss": 0.918961763381958,
      "step": 5295
    },
    {
      "epoch": 0.43370287340437513,
      "grad_norm": 0.2009030431509018,
      "learning_rate": 1.7834592697940246e-05,
      "loss": 0.8913455605506897,
      "step": 5296
    },
    {
      "epoch": 0.43378476594089405,
      "grad_norm": 0.2048221230506897,
      "learning_rate": 1.783240643139428e-05,
      "loss": 0.6601479053497314,
      "step": 5297
    },
    {
      "epoch": 0.433866658477413,
      "grad_norm": 0.19115330278873444,
      "learning_rate": 1.7830219911347904e-05,
      "loss": 0.8063157796859741,
      "step": 5298
    },
    {
      "epoch": 0.433948551013932,
      "grad_norm": 0.15937402844429016,
      "learning_rate": 1.782803313789616e-05,
      "loss": 0.8701393008232117,
      "step": 5299
    },
    {
      "epoch": 0.4340304435504509,
      "grad_norm": 0.18525800108909607,
      "learning_rate": 1.782584611113409e-05,
      "loss": 0.9214959144592285,
      "step": 5300
    },
    {
      "epoch": 0.43411233608696986,
      "grad_norm": 0.20054341852664948,
      "learning_rate": 1.7823658831156757e-05,
      "loss": 0.7140713334083557,
      "step": 5301
    },
    {
      "epoch": 0.43419422862348883,
      "grad_norm": 0.20579785108566284,
      "learning_rate": 1.782147129805922e-05,
      "loss": 1.3348990678787231,
      "step": 5302
    },
    {
      "epoch": 0.4342761211600078,
      "grad_norm": 0.17758409678936005,
      "learning_rate": 1.781928351193657e-05,
      "loss": 0.8599410653114319,
      "step": 5303
    },
    {
      "epoch": 0.4343580136965267,
      "grad_norm": 0.16102300584316254,
      "learning_rate": 1.7817095472883887e-05,
      "loss": 0.8594273328781128,
      "step": 5304
    },
    {
      "epoch": 0.4344399062330457,
      "grad_norm": 0.18289902806282043,
      "learning_rate": 1.7814907180996274e-05,
      "loss": 0.8032743334770203,
      "step": 5305
    },
    {
      "epoch": 0.43452179876956465,
      "grad_norm": 0.18315783143043518,
      "learning_rate": 1.781271863636885e-05,
      "loss": 1.0232023000717163,
      "step": 5306
    },
    {
      "epoch": 0.4346036913060836,
      "grad_norm": 0.21194827556610107,
      "learning_rate": 1.7810529839096727e-05,
      "loss": 0.8434994220733643,
      "step": 5307
    },
    {
      "epoch": 0.43468558384260253,
      "grad_norm": 0.15675370395183563,
      "learning_rate": 1.780834078927505e-05,
      "loss": 1.1023461818695068,
      "step": 5308
    },
    {
      "epoch": 0.4347674763791215,
      "grad_norm": 0.17738156020641327,
      "learning_rate": 1.780615148699896e-05,
      "loss": 0.7819367051124573,
      "step": 5309
    },
    {
      "epoch": 0.43484936891564047,
      "grad_norm": 0.19961993396282196,
      "learning_rate": 1.780396193236361e-05,
      "loss": 0.856561541557312,
      "step": 5310
    },
    {
      "epoch": 0.4349312614521594,
      "grad_norm": 0.19377334415912628,
      "learning_rate": 1.780177212546417e-05,
      "loss": 1.0451115369796753,
      "step": 5311
    },
    {
      "epoch": 0.43501315398867835,
      "grad_norm": 0.17293603718280792,
      "learning_rate": 1.7799582066395818e-05,
      "loss": 0.6771226525306702,
      "step": 5312
    },
    {
      "epoch": 0.4350950465251973,
      "grad_norm": 0.19175678491592407,
      "learning_rate": 1.7797391755253745e-05,
      "loss": 0.770937979221344,
      "step": 5313
    },
    {
      "epoch": 0.4351769390617163,
      "grad_norm": 0.2649477422237396,
      "learning_rate": 1.7795201192133146e-05,
      "loss": 0.7696906328201294,
      "step": 5314
    },
    {
      "epoch": 0.4352588315982352,
      "grad_norm": 0.20499393343925476,
      "learning_rate": 1.7793010377129234e-05,
      "loss": 0.5918949246406555,
      "step": 5315
    },
    {
      "epoch": 0.43534072413475416,
      "grad_norm": 0.16641917824745178,
      "learning_rate": 1.7790819310337233e-05,
      "loss": 0.7682873010635376,
      "step": 5316
    },
    {
      "epoch": 0.43542261667127313,
      "grad_norm": 0.1669362485408783,
      "learning_rate": 1.7788627991852374e-05,
      "loss": 0.7710197567939758,
      "step": 5317
    },
    {
      "epoch": 0.4355045092077921,
      "grad_norm": 0.19260668754577637,
      "learning_rate": 1.77864364217699e-05,
      "loss": 0.8183823227882385,
      "step": 5318
    },
    {
      "epoch": 0.435586401744311,
      "grad_norm": 0.17928192019462585,
      "learning_rate": 1.7784244600185065e-05,
      "loss": 0.8545870184898376,
      "step": 5319
    },
    {
      "epoch": 0.43566829428083,
      "grad_norm": 0.2326846420764923,
      "learning_rate": 1.778205252719314e-05,
      "loss": 0.7696012258529663,
      "step": 5320
    },
    {
      "epoch": 0.43575018681734895,
      "grad_norm": 0.18279686570167542,
      "learning_rate": 1.777986020288939e-05,
      "loss": 0.7747607231140137,
      "step": 5321
    },
    {
      "epoch": 0.43583207935386786,
      "grad_norm": 0.17589135468006134,
      "learning_rate": 1.777766762736912e-05,
      "loss": 0.7990815043449402,
      "step": 5322
    },
    {
      "epoch": 0.43591397189038683,
      "grad_norm": 0.18682031333446503,
      "learning_rate": 1.7775474800727613e-05,
      "loss": 1.2742195129394531,
      "step": 5323
    },
    {
      "epoch": 0.4359958644269058,
      "grad_norm": 0.19327767193317413,
      "learning_rate": 1.7773281723060185e-05,
      "loss": 0.8878986239433289,
      "step": 5324
    },
    {
      "epoch": 0.43607775696342477,
      "grad_norm": 0.15939031541347504,
      "learning_rate": 1.7771088394462156e-05,
      "loss": 0.8430910110473633,
      "step": 5325
    },
    {
      "epoch": 0.4361596494999437,
      "grad_norm": 0.21649105846881866,
      "learning_rate": 1.7768894815028853e-05,
      "loss": 0.8194905519485474,
      "step": 5326
    },
    {
      "epoch": 0.43624154203646265,
      "grad_norm": 0.18161314725875854,
      "learning_rate": 1.7766700984855623e-05,
      "loss": 0.7152822613716125,
      "step": 5327
    },
    {
      "epoch": 0.4363234345729816,
      "grad_norm": 0.18267078697681427,
      "learning_rate": 1.7764506904037816e-05,
      "loss": 0.6546453237533569,
      "step": 5328
    },
    {
      "epoch": 0.4364053271095006,
      "grad_norm": 0.23010072112083435,
      "learning_rate": 1.77623125726708e-05,
      "loss": 1.086508870124817,
      "step": 5329
    },
    {
      "epoch": 0.4364872196460195,
      "grad_norm": 0.17149338126182556,
      "learning_rate": 1.7760117990849942e-05,
      "loss": 0.9295289516448975,
      "step": 5330
    },
    {
      "epoch": 0.43656911218253847,
      "grad_norm": 0.20218384265899658,
      "learning_rate": 1.7757923158670638e-05,
      "loss": 0.9519931077957153,
      "step": 5331
    },
    {
      "epoch": 0.43665100471905743,
      "grad_norm": 0.17656703293323517,
      "learning_rate": 1.775572807622827e-05,
      "loss": 0.6793431043624878,
      "step": 5332
    },
    {
      "epoch": 0.43673289725557635,
      "grad_norm": 0.1682281643152237,
      "learning_rate": 1.7753532743618257e-05,
      "loss": 0.6486058831214905,
      "step": 5333
    },
    {
      "epoch": 0.4368147897920953,
      "grad_norm": 0.20148693025112152,
      "learning_rate": 1.7751337160936018e-05,
      "loss": 0.82770836353302,
      "step": 5334
    },
    {
      "epoch": 0.4368966823286143,
      "grad_norm": 0.19946426153182983,
      "learning_rate": 1.774914132827697e-05,
      "loss": 0.576486349105835,
      "step": 5335
    },
    {
      "epoch": 0.43697857486513325,
      "grad_norm": 0.19758914411067963,
      "learning_rate": 1.7746945245736568e-05,
      "loss": 0.6773539185523987,
      "step": 5336
    },
    {
      "epoch": 0.43706046740165216,
      "grad_norm": 0.1711030751466751,
      "learning_rate": 1.7744748913410255e-05,
      "loss": 0.8180829882621765,
      "step": 5337
    },
    {
      "epoch": 0.43714235993817113,
      "grad_norm": 0.18911796808242798,
      "learning_rate": 1.774255233139349e-05,
      "loss": 0.8503258228302002,
      "step": 5338
    },
    {
      "epoch": 0.4372242524746901,
      "grad_norm": 0.2244829684495926,
      "learning_rate": 1.774035549978175e-05,
      "loss": 1.099334478378296,
      "step": 5339
    },
    {
      "epoch": 0.43730614501120907,
      "grad_norm": 0.17577069997787476,
      "learning_rate": 1.7738158418670518e-05,
      "loss": 0.7109920382499695,
      "step": 5340
    },
    {
      "epoch": 0.437388037547728,
      "grad_norm": 0.17887598276138306,
      "learning_rate": 1.7735961088155287e-05,
      "loss": 0.9608672857284546,
      "step": 5341
    },
    {
      "epoch": 0.43746993008424695,
      "grad_norm": 0.21850259602069855,
      "learning_rate": 1.773376350833156e-05,
      "loss": 1.1691210269927979,
      "step": 5342
    },
    {
      "epoch": 0.4375518226207659,
      "grad_norm": 0.17714087665081024,
      "learning_rate": 1.773156567929486e-05,
      "loss": 0.7391451597213745,
      "step": 5343
    },
    {
      "epoch": 0.43763371515728483,
      "grad_norm": 0.15036965906620026,
      "learning_rate": 1.7729367601140705e-05,
      "loss": 0.685166597366333,
      "step": 5344
    },
    {
      "epoch": 0.4377156076938038,
      "grad_norm": 0.17373575270175934,
      "learning_rate": 1.772716927396464e-05,
      "loss": 1.000604510307312,
      "step": 5345
    },
    {
      "epoch": 0.43779750023032277,
      "grad_norm": 0.16224631667137146,
      "learning_rate": 1.7724970697862207e-05,
      "loss": 0.9363677501678467,
      "step": 5346
    },
    {
      "epoch": 0.43787939276684174,
      "grad_norm": 0.18499144911766052,
      "learning_rate": 1.772277187292897e-05,
      "loss": 0.8159418106079102,
      "step": 5347
    },
    {
      "epoch": 0.43796128530336065,
      "grad_norm": 0.15708468854427338,
      "learning_rate": 1.7720572799260497e-05,
      "loss": 0.6539720296859741,
      "step": 5348
    },
    {
      "epoch": 0.4380431778398796,
      "grad_norm": 0.17841537296772003,
      "learning_rate": 1.7718373476952366e-05,
      "loss": 0.7083767056465149,
      "step": 5349
    },
    {
      "epoch": 0.4381250703763986,
      "grad_norm": 0.1555212289094925,
      "learning_rate": 1.7716173906100178e-05,
      "loss": 0.6411643624305725,
      "step": 5350
    },
    {
      "epoch": 0.4382069629129175,
      "grad_norm": 0.23363284766674042,
      "learning_rate": 1.771397408679952e-05,
      "loss": 1.0165244340896606,
      "step": 5351
    },
    {
      "epoch": 0.43828885544943647,
      "grad_norm": 0.16776911914348602,
      "learning_rate": 1.771177401914602e-05,
      "loss": 0.7910541296005249,
      "step": 5352
    },
    {
      "epoch": 0.43837074798595543,
      "grad_norm": 0.19235385954380035,
      "learning_rate": 1.7709573703235295e-05,
      "loss": 0.7911033630371094,
      "step": 5353
    },
    {
      "epoch": 0.4384526405224744,
      "grad_norm": 0.1787407398223877,
      "learning_rate": 1.7707373139162982e-05,
      "loss": 0.7406383156776428,
      "step": 5354
    },
    {
      "epoch": 0.4385345330589933,
      "grad_norm": 0.17045576870441437,
      "learning_rate": 1.7705172327024718e-05,
      "loss": 0.9196993708610535,
      "step": 5355
    },
    {
      "epoch": 0.4386164255955123,
      "grad_norm": 0.18317103385925293,
      "learning_rate": 1.7702971266916176e-05,
      "loss": 0.8093180656433105,
      "step": 5356
    },
    {
      "epoch": 0.43869831813203125,
      "grad_norm": 0.18082581460475922,
      "learning_rate": 1.770076995893301e-05,
      "loss": 0.9671875238418579,
      "step": 5357
    },
    {
      "epoch": 0.4387802106685502,
      "grad_norm": 0.16281995177268982,
      "learning_rate": 1.7698568403170904e-05,
      "loss": 0.7753483057022095,
      "step": 5358
    },
    {
      "epoch": 0.43886210320506913,
      "grad_norm": 0.19645413756370544,
      "learning_rate": 1.769636659972554e-05,
      "loss": 0.899834394454956,
      "step": 5359
    },
    {
      "epoch": 0.4389439957415881,
      "grad_norm": 0.18177786469459534,
      "learning_rate": 1.7694164548692624e-05,
      "loss": 0.7106770873069763,
      "step": 5360
    },
    {
      "epoch": 0.43902588827810707,
      "grad_norm": 0.17229938507080078,
      "learning_rate": 1.7691962250167866e-05,
      "loss": 0.7530112266540527,
      "step": 5361
    },
    {
      "epoch": 0.439107780814626,
      "grad_norm": 0.17109821736812592,
      "learning_rate": 1.7689759704246983e-05,
      "loss": 0.7759050726890564,
      "step": 5362
    },
    {
      "epoch": 0.43918967335114495,
      "grad_norm": 0.2030181735754013,
      "learning_rate": 1.7687556911025707e-05,
      "loss": 0.8982888460159302,
      "step": 5363
    },
    {
      "epoch": 0.4392715658876639,
      "grad_norm": 0.16997826099395752,
      "learning_rate": 1.768535387059979e-05,
      "loss": 0.6584606766700745,
      "step": 5364
    },
    {
      "epoch": 0.4393534584241829,
      "grad_norm": 0.18718430399894714,
      "learning_rate": 1.7683150583064972e-05,
      "loss": 0.6908299326896667,
      "step": 5365
    },
    {
      "epoch": 0.4394353509607018,
      "grad_norm": 0.1918424814939499,
      "learning_rate": 1.768094704851702e-05,
      "loss": 0.8388704061508179,
      "step": 5366
    },
    {
      "epoch": 0.43951724349722077,
      "grad_norm": 0.16878333687782288,
      "learning_rate": 1.767874326705172e-05,
      "loss": 0.7723832130432129,
      "step": 5367
    },
    {
      "epoch": 0.43959913603373973,
      "grad_norm": 0.2200322151184082,
      "learning_rate": 1.7676539238764838e-05,
      "loss": 0.9158124923706055,
      "step": 5368
    },
    {
      "epoch": 0.4396810285702587,
      "grad_norm": 0.19098779559135437,
      "learning_rate": 1.7674334963752187e-05,
      "loss": 0.6910420656204224,
      "step": 5369
    },
    {
      "epoch": 0.4397629211067776,
      "grad_norm": 0.19668938219547272,
      "learning_rate": 1.7672130442109567e-05,
      "loss": 0.6618331670761108,
      "step": 5370
    },
    {
      "epoch": 0.4398448136432966,
      "grad_norm": 0.18889592587947845,
      "learning_rate": 1.7669925673932796e-05,
      "loss": 0.82016921043396,
      "step": 5371
    },
    {
      "epoch": 0.43992670617981555,
      "grad_norm": 0.22577214241027832,
      "learning_rate": 1.76677206593177e-05,
      "loss": 0.8170837163925171,
      "step": 5372
    },
    {
      "epoch": 0.44000859871633446,
      "grad_norm": 0.1758626103401184,
      "learning_rate": 1.7665515398360124e-05,
      "loss": 0.7099708318710327,
      "step": 5373
    },
    {
      "epoch": 0.44009049125285343,
      "grad_norm": 0.18815259635448456,
      "learning_rate": 1.7663309891155915e-05,
      "loss": 0.7074691653251648,
      "step": 5374
    },
    {
      "epoch": 0.4401723837893724,
      "grad_norm": 0.19174188375473022,
      "learning_rate": 1.7661104137800928e-05,
      "loss": 0.6769420504570007,
      "step": 5375
    },
    {
      "epoch": 0.44025427632589137,
      "grad_norm": 0.18055561184883118,
      "learning_rate": 1.765889813839104e-05,
      "loss": 0.7461521029472351,
      "step": 5376
    },
    {
      "epoch": 0.4403361688624103,
      "grad_norm": 0.18693168461322784,
      "learning_rate": 1.7656691893022136e-05,
      "loss": 0.799895167350769,
      "step": 5377
    },
    {
      "epoch": 0.44041806139892925,
      "grad_norm": 0.20090514421463013,
      "learning_rate": 1.76544854017901e-05,
      "loss": 0.7830310463905334,
      "step": 5378
    },
    {
      "epoch": 0.4404999539354482,
      "grad_norm": 0.20648477971553802,
      "learning_rate": 1.7652278664790837e-05,
      "loss": 1.1128127574920654,
      "step": 5379
    },
    {
      "epoch": 0.4405818464719672,
      "grad_norm": 0.16559775173664093,
      "learning_rate": 1.765007168212027e-05,
      "loss": 1.059472918510437,
      "step": 5380
    },
    {
      "epoch": 0.4406637390084861,
      "grad_norm": 0.2078838050365448,
      "learning_rate": 1.7647864453874313e-05,
      "loss": 0.8263754844665527,
      "step": 5381
    },
    {
      "epoch": 0.44074563154500507,
      "grad_norm": 0.17898233234882355,
      "learning_rate": 1.7645656980148904e-05,
      "loss": 0.944334864616394,
      "step": 5382
    },
    {
      "epoch": 0.44082752408152404,
      "grad_norm": 0.1673971712589264,
      "learning_rate": 1.764344926103999e-05,
      "loss": 0.7917730212211609,
      "step": 5383
    },
    {
      "epoch": 0.44090941661804295,
      "grad_norm": 0.1558142751455307,
      "learning_rate": 1.764124129664353e-05,
      "loss": 0.8971219062805176,
      "step": 5384
    },
    {
      "epoch": 0.4409913091545619,
      "grad_norm": 0.17135639488697052,
      "learning_rate": 1.7639033087055487e-05,
      "loss": 1.0078907012939453,
      "step": 5385
    },
    {
      "epoch": 0.4410732016910809,
      "grad_norm": 0.19785155355930328,
      "learning_rate": 1.7636824632371843e-05,
      "loss": 0.7111561894416809,
      "step": 5386
    },
    {
      "epoch": 0.44115509422759985,
      "grad_norm": 0.15664352476596832,
      "learning_rate": 1.763461593268858e-05,
      "loss": 0.6788076758384705,
      "step": 5387
    },
    {
      "epoch": 0.44123698676411877,
      "grad_norm": 0.17624083161354065,
      "learning_rate": 1.7632406988101704e-05,
      "loss": 1.0577363967895508,
      "step": 5388
    },
    {
      "epoch": 0.44131887930063773,
      "grad_norm": 0.15697833895683289,
      "learning_rate": 1.763019779870722e-05,
      "loss": 0.9716111421585083,
      "step": 5389
    },
    {
      "epoch": 0.4414007718371567,
      "grad_norm": 0.19322696328163147,
      "learning_rate": 1.7627988364601153e-05,
      "loss": 0.8862764239311218,
      "step": 5390
    },
    {
      "epoch": 0.44148266437367567,
      "grad_norm": 0.17831338942050934,
      "learning_rate": 1.7625778685879528e-05,
      "loss": 0.606127142906189,
      "step": 5391
    },
    {
      "epoch": 0.4415645569101946,
      "grad_norm": 0.1701916754245758,
      "learning_rate": 1.7623568762638394e-05,
      "loss": 0.8640361428260803,
      "step": 5392
    },
    {
      "epoch": 0.44164644944671355,
      "grad_norm": 0.2113039195537567,
      "learning_rate": 1.76213585949738e-05,
      "loss": 1.2156221866607666,
      "step": 5393
    },
    {
      "epoch": 0.4417283419832325,
      "grad_norm": 0.1668146550655365,
      "learning_rate": 1.761914818298181e-05,
      "loss": 1.0389198064804077,
      "step": 5394
    },
    {
      "epoch": 0.44181023451975143,
      "grad_norm": 0.15076585114002228,
      "learning_rate": 1.761693752675849e-05,
      "loss": 0.7940069437026978,
      "step": 5395
    },
    {
      "epoch": 0.4418921270562704,
      "grad_norm": 0.15243412554264069,
      "learning_rate": 1.7614726626399937e-05,
      "loss": 0.6890599131584167,
      "step": 5396
    },
    {
      "epoch": 0.44197401959278937,
      "grad_norm": 0.16983623802661896,
      "learning_rate": 1.7612515482002237e-05,
      "loss": 0.9388861656188965,
      "step": 5397
    },
    {
      "epoch": 0.44205591212930834,
      "grad_norm": 0.2105804979801178,
      "learning_rate": 1.76103040936615e-05,
      "loss": 0.8224899768829346,
      "step": 5398
    },
    {
      "epoch": 0.44213780466582725,
      "grad_norm": 0.19148819148540497,
      "learning_rate": 1.7608092461473832e-05,
      "loss": 0.6029199957847595,
      "step": 5399
    },
    {
      "epoch": 0.4422196972023462,
      "grad_norm": 0.22187601029872894,
      "learning_rate": 1.7605880585535376e-05,
      "loss": 0.6557669639587402,
      "step": 5400
    },
    {
      "epoch": 0.4423015897388652,
      "grad_norm": 0.1849789023399353,
      "learning_rate": 1.7603668465942258e-05,
      "loss": 0.8905231356620789,
      "step": 5401
    },
    {
      "epoch": 0.44238348227538415,
      "grad_norm": 0.2168682962656021,
      "learning_rate": 1.7601456102790633e-05,
      "loss": 0.831391453742981,
      "step": 5402
    },
    {
      "epoch": 0.44246537481190307,
      "grad_norm": 0.17690175771713257,
      "learning_rate": 1.7599243496176647e-05,
      "loss": 1.0622131824493408,
      "step": 5403
    },
    {
      "epoch": 0.44254726734842204,
      "grad_norm": 0.19221307337284088,
      "learning_rate": 1.759703064619648e-05,
      "loss": 0.6370087265968323,
      "step": 5404
    },
    {
      "epoch": 0.442629159884941,
      "grad_norm": 0.22120662033557892,
      "learning_rate": 1.7594817552946313e-05,
      "loss": 1.1419450044631958,
      "step": 5405
    },
    {
      "epoch": 0.4427110524214599,
      "grad_norm": 0.20749647915363312,
      "learning_rate": 1.759260421652233e-05,
      "loss": 0.8344279527664185,
      "step": 5406
    },
    {
      "epoch": 0.4427929449579789,
      "grad_norm": 0.21582792699337006,
      "learning_rate": 1.759039063702073e-05,
      "loss": 0.8805184960365295,
      "step": 5407
    },
    {
      "epoch": 0.44287483749449785,
      "grad_norm": 0.2117430567741394,
      "learning_rate": 1.758817681453773e-05,
      "loss": 0.9094138741493225,
      "step": 5408
    },
    {
      "epoch": 0.4429567300310168,
      "grad_norm": 0.15981557965278625,
      "learning_rate": 1.758596274916955e-05,
      "loss": 0.8580504059791565,
      "step": 5409
    },
    {
      "epoch": 0.44303862256753573,
      "grad_norm": 0.23026731610298157,
      "learning_rate": 1.758374844101242e-05,
      "loss": 0.9361943602561951,
      "step": 5410
    },
    {
      "epoch": 0.4431205151040547,
      "grad_norm": 0.17902237176895142,
      "learning_rate": 1.758153389016259e-05,
      "loss": 0.7177950739860535,
      "step": 5411
    },
    {
      "epoch": 0.44320240764057367,
      "grad_norm": 0.1778019368648529,
      "learning_rate": 1.75793190967163e-05,
      "loss": 0.8256241083145142,
      "step": 5412
    },
    {
      "epoch": 0.4432843001770926,
      "grad_norm": 0.20447474718093872,
      "learning_rate": 1.7577104060769832e-05,
      "loss": 0.7965594530105591,
      "step": 5413
    },
    {
      "epoch": 0.44336619271361155,
      "grad_norm": 0.19354255497455597,
      "learning_rate": 1.7574888782419445e-05,
      "loss": 0.749048113822937,
      "step": 5414
    },
    {
      "epoch": 0.4434480852501305,
      "grad_norm": 0.15867304801940918,
      "learning_rate": 1.757267326176143e-05,
      "loss": 0.7198055386543274,
      "step": 5415
    },
    {
      "epoch": 0.4435299777866495,
      "grad_norm": 0.19383947551250458,
      "learning_rate": 1.757045749889209e-05,
      "loss": 0.9032157063484192,
      "step": 5416
    },
    {
      "epoch": 0.4436118703231684,
      "grad_norm": 0.1326582431793213,
      "learning_rate": 1.7568241493907715e-05,
      "loss": 0.9449827671051025,
      "step": 5417
    },
    {
      "epoch": 0.44369376285968737,
      "grad_norm": 0.17711769044399261,
      "learning_rate": 1.7566025246904635e-05,
      "loss": 0.7158915996551514,
      "step": 5418
    },
    {
      "epoch": 0.44377565539620634,
      "grad_norm": 0.17684563994407654,
      "learning_rate": 1.7563808757979174e-05,
      "loss": 1.0480120182037354,
      "step": 5419
    },
    {
      "epoch": 0.4438575479327253,
      "grad_norm": 0.18076738715171814,
      "learning_rate": 1.7561592027227664e-05,
      "loss": 0.6801307201385498,
      "step": 5420
    },
    {
      "epoch": 0.4439394404692442,
      "grad_norm": 0.1546599268913269,
      "learning_rate": 1.755937505474646e-05,
      "loss": 0.6293601393699646,
      "step": 5421
    },
    {
      "epoch": 0.4440213330057632,
      "grad_norm": 0.19035924971103668,
      "learning_rate": 1.7557157840631924e-05,
      "loss": 1.0910961627960205,
      "step": 5422
    },
    {
      "epoch": 0.44410322554228215,
      "grad_norm": 0.17091119289398193,
      "learning_rate": 1.755494038498041e-05,
      "loss": 0.6681796908378601,
      "step": 5423
    },
    {
      "epoch": 0.44418511807880107,
      "grad_norm": 0.18414147198200226,
      "learning_rate": 1.7552722687888317e-05,
      "loss": 0.8092550039291382,
      "step": 5424
    },
    {
      "epoch": 0.44426701061532003,
      "grad_norm": 0.18572527170181274,
      "learning_rate": 1.7550504749452023e-05,
      "loss": 1.1223559379577637,
      "step": 5425
    },
    {
      "epoch": 0.444348903151839,
      "grad_norm": 0.15737557411193848,
      "learning_rate": 1.754828656976793e-05,
      "loss": 0.8473818898200989,
      "step": 5426
    },
    {
      "epoch": 0.44443079568835797,
      "grad_norm": 0.15985092520713806,
      "learning_rate": 1.7546068148932455e-05,
      "loss": 0.8165673017501831,
      "step": 5427
    },
    {
      "epoch": 0.4445126882248769,
      "grad_norm": 0.16474726796150208,
      "learning_rate": 1.7543849487042014e-05,
      "loss": 0.8353172540664673,
      "step": 5428
    },
    {
      "epoch": 0.44459458076139585,
      "grad_norm": 0.19072523713111877,
      "learning_rate": 1.7541630584193037e-05,
      "loss": 1.0249396562576294,
      "step": 5429
    },
    {
      "epoch": 0.4446764732979148,
      "grad_norm": 0.17066937685012817,
      "learning_rate": 1.7539411440481975e-05,
      "loss": 1.0296144485473633,
      "step": 5430
    },
    {
      "epoch": 0.4447583658344338,
      "grad_norm": 0.17020459473133087,
      "learning_rate": 1.7537192056005273e-05,
      "loss": 0.8971514701843262,
      "step": 5431
    },
    {
      "epoch": 0.4448402583709527,
      "grad_norm": 0.19096539914608002,
      "learning_rate": 1.7534972430859403e-05,
      "loss": 0.6164766550064087,
      "step": 5432
    },
    {
      "epoch": 0.44492215090747167,
      "grad_norm": 0.17912890017032623,
      "learning_rate": 1.7532752565140834e-05,
      "loss": 0.792995810508728,
      "step": 5433
    },
    {
      "epoch": 0.44500404344399064,
      "grad_norm": 0.1713162660598755,
      "learning_rate": 1.7530532458946044e-05,
      "loss": 0.8437955379486084,
      "step": 5434
    },
    {
      "epoch": 0.44508593598050955,
      "grad_norm": 0.18425898253917694,
      "learning_rate": 1.752831211237154e-05,
      "loss": 0.7909479737281799,
      "step": 5435
    },
    {
      "epoch": 0.4451678285170285,
      "grad_norm": 0.18475206196308136,
      "learning_rate": 1.7526091525513823e-05,
      "loss": 1.0152255296707153,
      "step": 5436
    },
    {
      "epoch": 0.4452497210535475,
      "grad_norm": 0.1779736876487732,
      "learning_rate": 1.7523870698469404e-05,
      "loss": 0.7136324644088745,
      "step": 5437
    },
    {
      "epoch": 0.44533161359006646,
      "grad_norm": 0.1955915242433548,
      "learning_rate": 1.7521649631334817e-05,
      "loss": 0.7692753672599792,
      "step": 5438
    },
    {
      "epoch": 0.44541350612658537,
      "grad_norm": 0.16643686592578888,
      "learning_rate": 1.751942832420659e-05,
      "loss": 0.778534471988678,
      "step": 5439
    },
    {
      "epoch": 0.44549539866310434,
      "grad_norm": 0.16394299268722534,
      "learning_rate": 1.7517206777181283e-05,
      "loss": 0.9160162806510925,
      "step": 5440
    },
    {
      "epoch": 0.4455772911996233,
      "grad_norm": 0.20852093398571014,
      "learning_rate": 1.7514984990355442e-05,
      "loss": 0.9404916763305664,
      "step": 5441
    },
    {
      "epoch": 0.4456591837361423,
      "grad_norm": 0.16706232726573944,
      "learning_rate": 1.751276296382564e-05,
      "loss": 0.6478914618492126,
      "step": 5442
    },
    {
      "epoch": 0.4457410762726612,
      "grad_norm": 0.1882484257221222,
      "learning_rate": 1.751054069768845e-05,
      "loss": 0.9441010355949402,
      "step": 5443
    },
    {
      "epoch": 0.44582296880918015,
      "grad_norm": 0.19298788905143738,
      "learning_rate": 1.7508318192040467e-05,
      "loss": 0.6870664954185486,
      "step": 5444
    },
    {
      "epoch": 0.4459048613456991,
      "grad_norm": 0.17786552011966705,
      "learning_rate": 1.7506095446978288e-05,
      "loss": 0.7045921087265015,
      "step": 5445
    },
    {
      "epoch": 0.44598675388221803,
      "grad_norm": 0.18361935019493103,
      "learning_rate": 1.7503872462598523e-05,
      "loss": 0.8420767188072205,
      "step": 5446
    },
    {
      "epoch": 0.446068646418737,
      "grad_norm": 0.17545048892498016,
      "learning_rate": 1.7501649238997792e-05,
      "loss": 1.03066885471344,
      "step": 5447
    },
    {
      "epoch": 0.44615053895525597,
      "grad_norm": 0.21919608116149902,
      "learning_rate": 1.749942577627273e-05,
      "loss": 0.8753542304039001,
      "step": 5448
    },
    {
      "epoch": 0.44623243149177494,
      "grad_norm": 0.15587112307548523,
      "learning_rate": 1.7497202074519968e-05,
      "loss": 0.7061967253684998,
      "step": 5449
    },
    {
      "epoch": 0.44631432402829385,
      "grad_norm": 0.15762439370155334,
      "learning_rate": 1.7494978133836163e-05,
      "loss": 0.7688755393028259,
      "step": 5450
    },
    {
      "epoch": 0.4463962165648128,
      "grad_norm": 0.16541148722171783,
      "learning_rate": 1.749275395431798e-05,
      "loss": 0.5245738625526428,
      "step": 5451
    },
    {
      "epoch": 0.4464781091013318,
      "grad_norm": 0.24459873139858246,
      "learning_rate": 1.7490529536062084e-05,
      "loss": 0.7518802881240845,
      "step": 5452
    },
    {
      "epoch": 0.44656000163785076,
      "grad_norm": 0.17605768144130707,
      "learning_rate": 1.748830487916516e-05,
      "loss": 0.7318469285964966,
      "step": 5453
    },
    {
      "epoch": 0.44664189417436967,
      "grad_norm": 0.17240917682647705,
      "learning_rate": 1.74860799837239e-05,
      "loss": 0.8843234181404114,
      "step": 5454
    },
    {
      "epoch": 0.44672378671088864,
      "grad_norm": 0.20060975849628448,
      "learning_rate": 1.748385484983501e-05,
      "loss": 0.8388319611549377,
      "step": 5455
    },
    {
      "epoch": 0.4468056792474076,
      "grad_norm": 0.19646620750427246,
      "learning_rate": 1.7481629477595202e-05,
      "loss": 0.9793956279754639,
      "step": 5456
    },
    {
      "epoch": 0.4468875717839265,
      "grad_norm": 0.20142483711242676,
      "learning_rate": 1.7479403867101203e-05,
      "loss": 0.6185078024864197,
      "step": 5457
    },
    {
      "epoch": 0.4469694643204455,
      "grad_norm": 0.18595372140407562,
      "learning_rate": 1.747717801844974e-05,
      "loss": 0.8611507415771484,
      "step": 5458
    },
    {
      "epoch": 0.44705135685696445,
      "grad_norm": 0.20564210414886475,
      "learning_rate": 1.747495193173756e-05,
      "loss": 0.8688211441040039,
      "step": 5459
    },
    {
      "epoch": 0.4471332493934834,
      "grad_norm": 0.15971340239048004,
      "learning_rate": 1.7472725607061427e-05,
      "loss": 0.7364599704742432,
      "step": 5460
    },
    {
      "epoch": 0.44721514193000234,
      "grad_norm": 0.18550996482372284,
      "learning_rate": 1.7470499044518092e-05,
      "loss": 1.013221263885498,
      "step": 5461
    },
    {
      "epoch": 0.4472970344665213,
      "grad_norm": 0.19798415899276733,
      "learning_rate": 1.746827224420434e-05,
      "loss": 1.0060904026031494,
      "step": 5462
    },
    {
      "epoch": 0.44737892700304027,
      "grad_norm": 0.17262700200080872,
      "learning_rate": 1.746604520621695e-05,
      "loss": 0.599404513835907,
      "step": 5463
    },
    {
      "epoch": 0.44746081953955924,
      "grad_norm": 0.1654191017150879,
      "learning_rate": 1.7463817930652724e-05,
      "loss": 1.036536693572998,
      "step": 5464
    },
    {
      "epoch": 0.44754271207607815,
      "grad_norm": 0.19327780604362488,
      "learning_rate": 1.746159041760847e-05,
      "loss": 1.0583151578903198,
      "step": 5465
    },
    {
      "epoch": 0.4476246046125971,
      "grad_norm": 0.18376433849334717,
      "learning_rate": 1.7459362667181e-05,
      "loss": 0.8896461725234985,
      "step": 5466
    },
    {
      "epoch": 0.4477064971491161,
      "grad_norm": 0.1908615082502365,
      "learning_rate": 1.7457134679467142e-05,
      "loss": 0.7980520725250244,
      "step": 5467
    },
    {
      "epoch": 0.447788389685635,
      "grad_norm": 0.17626892030239105,
      "learning_rate": 1.745490645456374e-05,
      "loss": 0.6143047213554382,
      "step": 5468
    },
    {
      "epoch": 0.44787028222215397,
      "grad_norm": 0.2045426070690155,
      "learning_rate": 1.7452677992567628e-05,
      "loss": 0.8465481400489807,
      "step": 5469
    },
    {
      "epoch": 0.44795217475867294,
      "grad_norm": 0.29870834946632385,
      "learning_rate": 1.7450449293575676e-05,
      "loss": 1.0538989305496216,
      "step": 5470
    },
    {
      "epoch": 0.4480340672951919,
      "grad_norm": 0.17277467250823975,
      "learning_rate": 1.7448220357684753e-05,
      "loss": 0.7371391654014587,
      "step": 5471
    },
    {
      "epoch": 0.4481159598317108,
      "grad_norm": 0.17069286108016968,
      "learning_rate": 1.7445991184991725e-05,
      "loss": 0.868395209312439,
      "step": 5472
    },
    {
      "epoch": 0.4481978523682298,
      "grad_norm": 0.190364271402359,
      "learning_rate": 1.7443761775593494e-05,
      "loss": 0.8595293760299683,
      "step": 5473
    },
    {
      "epoch": 0.44827974490474876,
      "grad_norm": 0.18257489800453186,
      "learning_rate": 1.744153212958696e-05,
      "loss": 0.6958825588226318,
      "step": 5474
    },
    {
      "epoch": 0.4483616374412677,
      "grad_norm": 0.17307493090629578,
      "learning_rate": 1.743930224706902e-05,
      "loss": 0.6610575914382935,
      "step": 5475
    },
    {
      "epoch": 0.44844352997778664,
      "grad_norm": 0.17977136373519897,
      "learning_rate": 1.7437072128136606e-05,
      "loss": 0.628240168094635,
      "step": 5476
    },
    {
      "epoch": 0.4485254225143056,
      "grad_norm": 0.16960987448692322,
      "learning_rate": 1.743484177288664e-05,
      "loss": 0.872008740901947,
      "step": 5477
    },
    {
      "epoch": 0.4486073150508246,
      "grad_norm": 0.20718859136104584,
      "learning_rate": 1.743261118141607e-05,
      "loss": 0.9913686513900757,
      "step": 5478
    },
    {
      "epoch": 0.4486892075873435,
      "grad_norm": 0.17606183886528015,
      "learning_rate": 1.7430380353821844e-05,
      "loss": 0.9771310091018677,
      "step": 5479
    },
    {
      "epoch": 0.44877110012386245,
      "grad_norm": 0.1913255751132965,
      "learning_rate": 1.7428149290200915e-05,
      "loss": 0.6810815334320068,
      "step": 5480
    },
    {
      "epoch": 0.4488529926603814,
      "grad_norm": 0.1763249933719635,
      "learning_rate": 1.742591799065027e-05,
      "loss": 1.047243595123291,
      "step": 5481
    },
    {
      "epoch": 0.4489348851969004,
      "grad_norm": 0.17235812544822693,
      "learning_rate": 1.7423686455266876e-05,
      "loss": 0.7062926292419434,
      "step": 5482
    },
    {
      "epoch": 0.4490167777334193,
      "grad_norm": 0.14848990738391876,
      "learning_rate": 1.7421454684147725e-05,
      "loss": 0.6057683229446411,
      "step": 5483
    },
    {
      "epoch": 0.44909867026993827,
      "grad_norm": 0.20214088261127472,
      "learning_rate": 1.7419222677389833e-05,
      "loss": 0.7344767451286316,
      "step": 5484
    },
    {
      "epoch": 0.44918056280645724,
      "grad_norm": 0.19846497476100922,
      "learning_rate": 1.7416990435090202e-05,
      "loss": 1.2643386125564575,
      "step": 5485
    },
    {
      "epoch": 0.44926245534297615,
      "grad_norm": 0.18715107440948486,
      "learning_rate": 1.741475795734585e-05,
      "loss": 0.8277486562728882,
      "step": 5486
    },
    {
      "epoch": 0.4493443478794951,
      "grad_norm": 0.19467340409755707,
      "learning_rate": 1.7412525244253824e-05,
      "loss": 0.7172595858573914,
      "step": 5487
    },
    {
      "epoch": 0.4494262404160141,
      "grad_norm": 0.17791521549224854,
      "learning_rate": 1.741029229591115e-05,
      "loss": 0.9489541053771973,
      "step": 5488
    },
    {
      "epoch": 0.44950813295253306,
      "grad_norm": 0.19396749138832092,
      "learning_rate": 1.7408059112414894e-05,
      "loss": 0.8113859295845032,
      "step": 5489
    },
    {
      "epoch": 0.44959002548905197,
      "grad_norm": 0.17106521129608154,
      "learning_rate": 1.7405825693862113e-05,
      "loss": 0.8235581517219543,
      "step": 5490
    },
    {
      "epoch": 0.44967191802557094,
      "grad_norm": 0.18694078922271729,
      "learning_rate": 1.7403592040349884e-05,
      "loss": 0.8287842869758606,
      "step": 5491
    },
    {
      "epoch": 0.4497538105620899,
      "grad_norm": 0.20243921875953674,
      "learning_rate": 1.740135815197529e-05,
      "loss": 0.7463896870613098,
      "step": 5492
    },
    {
      "epoch": 0.4498357030986089,
      "grad_norm": 0.18900007009506226,
      "learning_rate": 1.739912402883542e-05,
      "loss": 0.5805867910385132,
      "step": 5493
    },
    {
      "epoch": 0.4499175956351278,
      "grad_norm": 0.1829969435930252,
      "learning_rate": 1.739688967102738e-05,
      "loss": 0.800578773021698,
      "step": 5494
    },
    {
      "epoch": 0.44999948817164676,
      "grad_norm": 0.19881930947303772,
      "learning_rate": 1.7394655078648298e-05,
      "loss": 0.780738353729248,
      "step": 5495
    },
    {
      "epoch": 0.4500813807081657,
      "grad_norm": 0.20201973617076874,
      "learning_rate": 1.7392420251795283e-05,
      "loss": 0.6185060739517212,
      "step": 5496
    },
    {
      "epoch": 0.45016327324468464,
      "grad_norm": 0.14701874554157257,
      "learning_rate": 1.7390185190565466e-05,
      "loss": 0.804574728012085,
      "step": 5497
    },
    {
      "epoch": 0.4502451657812036,
      "grad_norm": 0.1929609626531601,
      "learning_rate": 1.7387949895056006e-05,
      "loss": 0.9140502214431763,
      "step": 5498
    },
    {
      "epoch": 0.4503270583177226,
      "grad_norm": 0.19172126054763794,
      "learning_rate": 1.7385714365364054e-05,
      "loss": 0.8545142412185669,
      "step": 5499
    },
    {
      "epoch": 0.45040895085424154,
      "grad_norm": 0.2117965817451477,
      "learning_rate": 1.738347860158677e-05,
      "loss": 0.7460088133811951,
      "step": 5500
    },
    {
      "epoch": 0.45049084339076045,
      "grad_norm": 0.1671985387802124,
      "learning_rate": 1.738124260382134e-05,
      "loss": 0.9331665635108948,
      "step": 5501
    },
    {
      "epoch": 0.4505727359272794,
      "grad_norm": 0.15618084371089935,
      "learning_rate": 1.7379006372164936e-05,
      "loss": 0.5510960221290588,
      "step": 5502
    },
    {
      "epoch": 0.4506546284637984,
      "grad_norm": 0.1842808872461319,
      "learning_rate": 1.737676990671476e-05,
      "loss": 0.8737232685089111,
      "step": 5503
    },
    {
      "epoch": 0.45073652100031736,
      "grad_norm": 0.18284103274345398,
      "learning_rate": 1.737453320756802e-05,
      "loss": 0.6817474365234375,
      "step": 5504
    },
    {
      "epoch": 0.45081841353683627,
      "grad_norm": 0.1913343071937561,
      "learning_rate": 1.7372296274821936e-05,
      "loss": 0.9730151891708374,
      "step": 5505
    },
    {
      "epoch": 0.45090030607335524,
      "grad_norm": 0.19930996000766754,
      "learning_rate": 1.7370059108573723e-05,
      "loss": 0.9446821808815002,
      "step": 5506
    },
    {
      "epoch": 0.4509821986098742,
      "grad_norm": 0.17180867493152618,
      "learning_rate": 1.7367821708920624e-05,
      "loss": 0.9238617420196533,
      "step": 5507
    },
    {
      "epoch": 0.4510640911463931,
      "grad_norm": 0.20669066905975342,
      "learning_rate": 1.7365584075959887e-05,
      "loss": 1.3154327869415283,
      "step": 5508
    },
    {
      "epoch": 0.4511459836829121,
      "grad_norm": 0.18532808125019073,
      "learning_rate": 1.7363346209788766e-05,
      "loss": 0.6891070008277893,
      "step": 5509
    },
    {
      "epoch": 0.45122787621943106,
      "grad_norm": 0.18500258028507233,
      "learning_rate": 1.736110811050453e-05,
      "loss": 0.7747712731361389,
      "step": 5510
    },
    {
      "epoch": 0.45130976875595,
      "grad_norm": 0.18326996266841888,
      "learning_rate": 1.7358869778204452e-05,
      "loss": 0.7281123995780945,
      "step": 5511
    },
    {
      "epoch": 0.45139166129246894,
      "grad_norm": 0.1820676326751709,
      "learning_rate": 1.7356631212985822e-05,
      "loss": 0.813267707824707,
      "step": 5512
    },
    {
      "epoch": 0.4514735538289879,
      "grad_norm": 0.18251243233680725,
      "learning_rate": 1.7354392414945936e-05,
      "loss": 0.6807146668434143,
      "step": 5513
    },
    {
      "epoch": 0.4515554463655069,
      "grad_norm": 0.17788632214069366,
      "learning_rate": 1.7352153384182108e-05,
      "loss": 0.6701048016548157,
      "step": 5514
    },
    {
      "epoch": 0.45163733890202584,
      "grad_norm": 0.20719458162784576,
      "learning_rate": 1.7349914120791648e-05,
      "loss": 0.870582103729248,
      "step": 5515
    },
    {
      "epoch": 0.45171923143854475,
      "grad_norm": 0.15191584825515747,
      "learning_rate": 1.734767462487188e-05,
      "loss": 0.5300018787384033,
      "step": 5516
    },
    {
      "epoch": 0.4518011239750637,
      "grad_norm": 0.15818405151367188,
      "learning_rate": 1.7345434896520154e-05,
      "loss": 1.1534440517425537,
      "step": 5517
    },
    {
      "epoch": 0.4518830165115827,
      "grad_norm": 0.20795802772045135,
      "learning_rate": 1.734319493583381e-05,
      "loss": 1.1220030784606934,
      "step": 5518
    },
    {
      "epoch": 0.4519649090481016,
      "grad_norm": 0.17918804287910461,
      "learning_rate": 1.7340954742910206e-05,
      "loss": 0.7341663241386414,
      "step": 5519
    },
    {
      "epoch": 0.4520468015846206,
      "grad_norm": 0.20755669474601746,
      "learning_rate": 1.733871431784671e-05,
      "loss": 0.7412256002426147,
      "step": 5520
    },
    {
      "epoch": 0.45212869412113954,
      "grad_norm": 0.19410134851932526,
      "learning_rate": 1.7336473660740704e-05,
      "loss": 0.8384029269218445,
      "step": 5521
    },
    {
      "epoch": 0.4522105866576585,
      "grad_norm": 0.17224611341953278,
      "learning_rate": 1.733423277168957e-05,
      "loss": 0.631521463394165,
      "step": 5522
    },
    {
      "epoch": 0.4522924791941774,
      "grad_norm": 0.21063853800296783,
      "learning_rate": 1.7331991650790712e-05,
      "loss": 0.8545302748680115,
      "step": 5523
    },
    {
      "epoch": 0.4523743717306964,
      "grad_norm": 0.20307908952236176,
      "learning_rate": 1.7329750298141535e-05,
      "loss": 1.1925686597824097,
      "step": 5524
    },
    {
      "epoch": 0.45245626426721536,
      "grad_norm": 0.18799248337745667,
      "learning_rate": 1.732750871383946e-05,
      "loss": 0.9018441438674927,
      "step": 5525
    },
    {
      "epoch": 0.4525381568037343,
      "grad_norm": 0.19797302782535553,
      "learning_rate": 1.732526689798191e-05,
      "loss": 1.0830812454223633,
      "step": 5526
    },
    {
      "epoch": 0.45262004934025324,
      "grad_norm": 0.19762778282165527,
      "learning_rate": 1.732302485066633e-05,
      "loss": 1.1677149534225464,
      "step": 5527
    },
    {
      "epoch": 0.4527019418767722,
      "grad_norm": 0.19650253653526306,
      "learning_rate": 1.7320782571990168e-05,
      "loss": 0.7618950605392456,
      "step": 5528
    },
    {
      "epoch": 0.4527838344132912,
      "grad_norm": 0.16138139367103577,
      "learning_rate": 1.7318540062050882e-05,
      "loss": 0.7819635272026062,
      "step": 5529
    },
    {
      "epoch": 0.4528657269498101,
      "grad_norm": 0.17745046317577362,
      "learning_rate": 1.7316297320945934e-05,
      "loss": 0.8661229014396667,
      "step": 5530
    },
    {
      "epoch": 0.45294761948632906,
      "grad_norm": 0.17848540842533112,
      "learning_rate": 1.7314054348772807e-05,
      "loss": 0.6610631942749023,
      "step": 5531
    },
    {
      "epoch": 0.453029512022848,
      "grad_norm": 0.18103136122226715,
      "learning_rate": 1.7311811145628998e-05,
      "loss": 0.5063185095787048,
      "step": 5532
    },
    {
      "epoch": 0.453111404559367,
      "grad_norm": 0.17355968058109283,
      "learning_rate": 1.7309567711611997e-05,
      "loss": 0.7535468935966492,
      "step": 5533
    },
    {
      "epoch": 0.4531932970958859,
      "grad_norm": 0.19797727465629578,
      "learning_rate": 1.730732404681931e-05,
      "loss": 0.8034313917160034,
      "step": 5534
    },
    {
      "epoch": 0.4532751896324049,
      "grad_norm": 0.17570658028125763,
      "learning_rate": 1.7305080151348467e-05,
      "loss": 0.6996821165084839,
      "step": 5535
    },
    {
      "epoch": 0.45335708216892384,
      "grad_norm": 0.17188256978988647,
      "learning_rate": 1.7302836025296986e-05,
      "loss": 0.6018949151039124,
      "step": 5536
    },
    {
      "epoch": 0.4534389747054428,
      "grad_norm": 0.1995725929737091,
      "learning_rate": 1.7300591668762413e-05,
      "loss": 0.7266729474067688,
      "step": 5537
    },
    {
      "epoch": 0.4535208672419617,
      "grad_norm": 0.19049639999866486,
      "learning_rate": 1.7298347081842294e-05,
      "loss": 0.6787266135215759,
      "step": 5538
    },
    {
      "epoch": 0.4536027597784807,
      "grad_norm": 0.1832939237356186,
      "learning_rate": 1.7296102264634188e-05,
      "loss": 0.9585368037223816,
      "step": 5539
    },
    {
      "epoch": 0.45368465231499966,
      "grad_norm": 0.2039623260498047,
      "learning_rate": 1.7293857217235663e-05,
      "loss": 0.7218517065048218,
      "step": 5540
    },
    {
      "epoch": 0.45376654485151857,
      "grad_norm": 0.2135409265756607,
      "learning_rate": 1.7291611939744304e-05,
      "loss": 1.1587527990341187,
      "step": 5541
    },
    {
      "epoch": 0.45384843738803754,
      "grad_norm": 0.18465997278690338,
      "learning_rate": 1.7289366432257696e-05,
      "loss": 1.21940016746521,
      "step": 5542
    },
    {
      "epoch": 0.4539303299245565,
      "grad_norm": 0.1953372210264206,
      "learning_rate": 1.728712069487343e-05,
      "loss": 0.9770494103431702,
      "step": 5543
    },
    {
      "epoch": 0.4540122224610755,
      "grad_norm": 0.16749964654445648,
      "learning_rate": 1.7284874727689125e-05,
      "loss": 0.468506783246994,
      "step": 5544
    },
    {
      "epoch": 0.4540941149975944,
      "grad_norm": 0.32112085819244385,
      "learning_rate": 1.7282628530802405e-05,
      "loss": 0.8787829279899597,
      "step": 5545
    },
    {
      "epoch": 0.45417600753411336,
      "grad_norm": 0.15989165008068085,
      "learning_rate": 1.728038210431088e-05,
      "loss": 0.7138000130653381,
      "step": 5546
    },
    {
      "epoch": 0.4542579000706323,
      "grad_norm": 0.19676348567008972,
      "learning_rate": 1.727813544831221e-05,
      "loss": 1.0367257595062256,
      "step": 5547
    },
    {
      "epoch": 0.45433979260715124,
      "grad_norm": 0.17124904692173004,
      "learning_rate": 1.7275888562904028e-05,
      "loss": 0.49154144525527954,
      "step": 5548
    },
    {
      "epoch": 0.4544216851436702,
      "grad_norm": 0.20187829434871674,
      "learning_rate": 1.7273641448184003e-05,
      "loss": 1.0354478359222412,
      "step": 5549
    },
    {
      "epoch": 0.4545035776801892,
      "grad_norm": 0.19685757160186768,
      "learning_rate": 1.7271394104249802e-05,
      "loss": 1.0775749683380127,
      "step": 5550
    },
    {
      "epoch": 0.45458547021670814,
      "grad_norm": 0.2088436633348465,
      "learning_rate": 1.7269146531199098e-05,
      "loss": 0.9345431327819824,
      "step": 5551
    },
    {
      "epoch": 0.45466736275322706,
      "grad_norm": 0.23142242431640625,
      "learning_rate": 1.7266898729129583e-05,
      "loss": 1.1239397525787354,
      "step": 5552
    },
    {
      "epoch": 0.454749255289746,
      "grad_norm": 0.19443278014659882,
      "learning_rate": 1.7264650698138962e-05,
      "loss": 0.7231912016868591,
      "step": 5553
    },
    {
      "epoch": 0.454831147826265,
      "grad_norm": 0.2733670473098755,
      "learning_rate": 1.7262402438324932e-05,
      "loss": 0.8453428149223328,
      "step": 5554
    },
    {
      "epoch": 0.45491304036278396,
      "grad_norm": 0.18593311309814453,
      "learning_rate": 1.726015394978522e-05,
      "loss": 0.5910945534706116,
      "step": 5555
    },
    {
      "epoch": 0.4549949328993029,
      "grad_norm": 0.17861849069595337,
      "learning_rate": 1.7257905232617555e-05,
      "loss": 0.8169873952865601,
      "step": 5556
    },
    {
      "epoch": 0.45507682543582184,
      "grad_norm": 0.1759481579065323,
      "learning_rate": 1.7255656286919666e-05,
      "loss": 0.9164767861366272,
      "step": 5557
    },
    {
      "epoch": 0.4551587179723408,
      "grad_norm": 0.16161689162254333,
      "learning_rate": 1.7253407112789317e-05,
      "loss": 0.5934125781059265,
      "step": 5558
    },
    {
      "epoch": 0.4552406105088597,
      "grad_norm": 0.1739482581615448,
      "learning_rate": 1.7251157710324253e-05,
      "loss": 1.0104881525039673,
      "step": 5559
    },
    {
      "epoch": 0.4553225030453787,
      "grad_norm": 0.1943245381116867,
      "learning_rate": 1.7248908079622244e-05,
      "loss": 0.9934820532798767,
      "step": 5560
    },
    {
      "epoch": 0.45540439558189766,
      "grad_norm": 0.14801955223083496,
      "learning_rate": 1.724665822078108e-05,
      "loss": 0.9609352946281433,
      "step": 5561
    },
    {
      "epoch": 0.4554862881184166,
      "grad_norm": 0.19583968818187714,
      "learning_rate": 1.724440813389853e-05,
      "loss": 0.9021040797233582,
      "step": 5562
    },
    {
      "epoch": 0.45556818065493554,
      "grad_norm": 0.15777328610420227,
      "learning_rate": 1.7242157819072408e-05,
      "loss": 0.7732074856758118,
      "step": 5563
    },
    {
      "epoch": 0.4556500731914545,
      "grad_norm": 0.18235963582992554,
      "learning_rate": 1.7239907276400518e-05,
      "loss": 0.8612820506095886,
      "step": 5564
    },
    {
      "epoch": 0.4557319657279735,
      "grad_norm": 0.21064628660678864,
      "learning_rate": 1.7237656505980673e-05,
      "loss": 0.9134030938148499,
      "step": 5565
    },
    {
      "epoch": 0.45581385826449244,
      "grad_norm": 0.17954738438129425,
      "learning_rate": 1.7235405507910704e-05,
      "loss": 0.796614944934845,
      "step": 5566
    },
    {
      "epoch": 0.45589575080101136,
      "grad_norm": 0.1477358639240265,
      "learning_rate": 1.7233154282288448e-05,
      "loss": 0.8662889003753662,
      "step": 5567
    },
    {
      "epoch": 0.4559776433375303,
      "grad_norm": 0.18856821954250336,
      "learning_rate": 1.7230902829211754e-05,
      "loss": 0.7465298771858215,
      "step": 5568
    },
    {
      "epoch": 0.4560595358740493,
      "grad_norm": 0.2203916311264038,
      "learning_rate": 1.722865114877848e-05,
      "loss": 0.7384775280952454,
      "step": 5569
    },
    {
      "epoch": 0.4561414284105682,
      "grad_norm": 0.15738825500011444,
      "learning_rate": 1.7226399241086494e-05,
      "loss": 0.8732999563217163,
      "step": 5570
    },
    {
      "epoch": 0.4562233209470872,
      "grad_norm": 0.22765879333019257,
      "learning_rate": 1.7224147106233665e-05,
      "loss": 0.8452392220497131,
      "step": 5571
    },
    {
      "epoch": 0.45630521348360614,
      "grad_norm": 0.1592656821012497,
      "learning_rate": 1.722189474431789e-05,
      "loss": 0.8674609661102295,
      "step": 5572
    },
    {
      "epoch": 0.4563871060201251,
      "grad_norm": 0.19468270242214203,
      "learning_rate": 1.721964215543706e-05,
      "loss": 0.8810535073280334,
      "step": 5573
    },
    {
      "epoch": 0.456468998556644,
      "grad_norm": 0.22282390296459198,
      "learning_rate": 1.7217389339689088e-05,
      "loss": 0.9202957153320312,
      "step": 5574
    },
    {
      "epoch": 0.456550891093163,
      "grad_norm": 0.2271028757095337,
      "learning_rate": 1.721513629717188e-05,
      "loss": 0.8471079468727112,
      "step": 5575
    },
    {
      "epoch": 0.45663278362968196,
      "grad_norm": 0.19008708000183105,
      "learning_rate": 1.7212883027983373e-05,
      "loss": 1.0594571828842163,
      "step": 5576
    },
    {
      "epoch": 0.45671467616620093,
      "grad_norm": 0.21355175971984863,
      "learning_rate": 1.7210629532221497e-05,
      "loss": 0.9036674499511719,
      "step": 5577
    },
    {
      "epoch": 0.45679656870271984,
      "grad_norm": 0.1982448697090149,
      "learning_rate": 1.7208375809984205e-05,
      "loss": 0.7601367831230164,
      "step": 5578
    },
    {
      "epoch": 0.4568784612392388,
      "grad_norm": 0.1790967732667923,
      "learning_rate": 1.7206121861369443e-05,
      "loss": 1.2467598915100098,
      "step": 5579
    },
    {
      "epoch": 0.4569603537757578,
      "grad_norm": 0.2072426825761795,
      "learning_rate": 1.7203867686475185e-05,
      "loss": 0.8452096581459045,
      "step": 5580
    },
    {
      "epoch": 0.4570422463122767,
      "grad_norm": 0.17102275788784027,
      "learning_rate": 1.7201613285399404e-05,
      "loss": 0.7410465478897095,
      "step": 5581
    },
    {
      "epoch": 0.45712413884879566,
      "grad_norm": 0.2763693928718567,
      "learning_rate": 1.719935865824009e-05,
      "loss": 1.0754187107086182,
      "step": 5582
    },
    {
      "epoch": 0.4572060313853146,
      "grad_norm": 0.20731182396411896,
      "learning_rate": 1.7197103805095228e-05,
      "loss": 0.8688687086105347,
      "step": 5583
    },
    {
      "epoch": 0.4572879239218336,
      "grad_norm": 0.18402840197086334,
      "learning_rate": 1.719484872606283e-05,
      "loss": 0.9625867605209351,
      "step": 5584
    },
    {
      "epoch": 0.4573698164583525,
      "grad_norm": 0.17972294986248016,
      "learning_rate": 1.7192593421240912e-05,
      "loss": 0.6844893097877502,
      "step": 5585
    },
    {
      "epoch": 0.4574517089948715,
      "grad_norm": 0.17459367215633392,
      "learning_rate": 1.71903378907275e-05,
      "loss": 0.9083195924758911,
      "step": 5586
    },
    {
      "epoch": 0.45753360153139044,
      "grad_norm": 0.20023025572299957,
      "learning_rate": 1.718808213462062e-05,
      "loss": 0.7861866354942322,
      "step": 5587
    },
    {
      "epoch": 0.4576154940679094,
      "grad_norm": 0.17141178250312805,
      "learning_rate": 1.7185826153018327e-05,
      "loss": 0.8535504341125488,
      "step": 5588
    },
    {
      "epoch": 0.4576973866044283,
      "grad_norm": 0.18199318647384644,
      "learning_rate": 1.7183569946018665e-05,
      "loss": 0.9447701573371887,
      "step": 5589
    },
    {
      "epoch": 0.4577792791409473,
      "grad_norm": 0.1726444810628891,
      "learning_rate": 1.718131351371971e-05,
      "loss": 0.6782296895980835,
      "step": 5590
    },
    {
      "epoch": 0.45786117167746626,
      "grad_norm": 0.1802307367324829,
      "learning_rate": 1.7179056856219527e-05,
      "loss": 0.9088866710662842,
      "step": 5591
    },
    {
      "epoch": 0.4579430642139852,
      "grad_norm": 0.17682746052742004,
      "learning_rate": 1.71767999736162e-05,
      "loss": 1.0102368593215942,
      "step": 5592
    },
    {
      "epoch": 0.45802495675050414,
      "grad_norm": 0.1846659928560257,
      "learning_rate": 1.7174542866007825e-05,
      "loss": 1.1352025270462036,
      "step": 5593
    },
    {
      "epoch": 0.4581068492870231,
      "grad_norm": 0.16514235734939575,
      "learning_rate": 1.7172285533492508e-05,
      "loss": 0.5929916501045227,
      "step": 5594
    },
    {
      "epoch": 0.4581887418235421,
      "grad_norm": 0.2367272675037384,
      "learning_rate": 1.7170027976168355e-05,
      "loss": 1.0209989547729492,
      "step": 5595
    },
    {
      "epoch": 0.458270634360061,
      "grad_norm": 0.20493969321250916,
      "learning_rate": 1.7167770194133494e-05,
      "loss": 0.7963019013404846,
      "step": 5596
    },
    {
      "epoch": 0.45835252689657996,
      "grad_norm": 0.17345423996448517,
      "learning_rate": 1.7165512187486057e-05,
      "loss": 0.6873565316200256,
      "step": 5597
    },
    {
      "epoch": 0.4584344194330989,
      "grad_norm": 0.22013291716575623,
      "learning_rate": 1.7163253956324182e-05,
      "loss": 0.8781787753105164,
      "step": 5598
    },
    {
      "epoch": 0.4585163119696179,
      "grad_norm": 0.18288859724998474,
      "learning_rate": 1.7160995500746028e-05,
      "loss": 0.5488592386245728,
      "step": 5599
    },
    {
      "epoch": 0.4585982045061368,
      "grad_norm": 0.18942083418369293,
      "learning_rate": 1.715873682084975e-05,
      "loss": 0.783503532409668,
      "step": 5600
    },
    {
      "epoch": 0.4586800970426558,
      "grad_norm": 0.17771703004837036,
      "learning_rate": 1.7156477916733525e-05,
      "loss": 0.5279858112335205,
      "step": 5601
    },
    {
      "epoch": 0.45876198957917474,
      "grad_norm": 0.21360014379024506,
      "learning_rate": 1.715421878849553e-05,
      "loss": 1.20832097530365,
      "step": 5602
    },
    {
      "epoch": 0.45884388211569366,
      "grad_norm": 0.23653362691402435,
      "learning_rate": 1.7151959436233957e-05,
      "loss": 0.8957186937332153,
      "step": 5603
    },
    {
      "epoch": 0.4589257746522126,
      "grad_norm": 0.1886117309331894,
      "learning_rate": 1.714969986004701e-05,
      "loss": 0.8974544405937195,
      "step": 5604
    },
    {
      "epoch": 0.4590076671887316,
      "grad_norm": 0.17192848026752472,
      "learning_rate": 1.7147440060032895e-05,
      "loss": 0.7700780630111694,
      "step": 5605
    },
    {
      "epoch": 0.45908955972525056,
      "grad_norm": 0.20000216364860535,
      "learning_rate": 1.7145180036289834e-05,
      "loss": 0.8034606575965881,
      "step": 5606
    },
    {
      "epoch": 0.4591714522617695,
      "grad_norm": 0.14850115776062012,
      "learning_rate": 1.714291978891606e-05,
      "loss": 0.6440001130104065,
      "step": 5607
    },
    {
      "epoch": 0.45925334479828844,
      "grad_norm": 0.19381895661354065,
      "learning_rate": 1.7140659318009807e-05,
      "loss": 1.0006682872772217,
      "step": 5608
    },
    {
      "epoch": 0.4593352373348074,
      "grad_norm": 0.19166319072246552,
      "learning_rate": 1.7138398623669325e-05,
      "loss": 0.8180659413337708,
      "step": 5609
    },
    {
      "epoch": 0.4594171298713264,
      "grad_norm": 0.16670072078704834,
      "learning_rate": 1.7136137705992883e-05,
      "loss": 0.6220978498458862,
      "step": 5610
    },
    {
      "epoch": 0.4594990224078453,
      "grad_norm": 0.1902817338705063,
      "learning_rate": 1.7133876565078735e-05,
      "loss": 1.195241093635559,
      "step": 5611
    },
    {
      "epoch": 0.45958091494436426,
      "grad_norm": 0.16342461109161377,
      "learning_rate": 1.7131615201025165e-05,
      "loss": 0.7688427567481995,
      "step": 5612
    },
    {
      "epoch": 0.45966280748088323,
      "grad_norm": 0.16556203365325928,
      "learning_rate": 1.7129353613930473e-05,
      "loss": 0.8004022240638733,
      "step": 5613
    },
    {
      "epoch": 0.45974470001740214,
      "grad_norm": 0.15656988322734833,
      "learning_rate": 1.712709180389294e-05,
      "loss": 0.6066978573799133,
      "step": 5614
    },
    {
      "epoch": 0.4598265925539211,
      "grad_norm": 0.19695082306861877,
      "learning_rate": 1.7124829771010878e-05,
      "loss": 0.5805996060371399,
      "step": 5615
    },
    {
      "epoch": 0.4599084850904401,
      "grad_norm": 0.17101480066776276,
      "learning_rate": 1.7122567515382612e-05,
      "loss": 0.7877641320228577,
      "step": 5616
    },
    {
      "epoch": 0.45999037762695905,
      "grad_norm": 0.2078593224287033,
      "learning_rate": 1.7120305037106457e-05,
      "loss": 0.965295135974884,
      "step": 5617
    },
    {
      "epoch": 0.46007227016347796,
      "grad_norm": 0.16212765872478485,
      "learning_rate": 1.7118042336280763e-05,
      "loss": 0.6437591910362244,
      "step": 5618
    },
    {
      "epoch": 0.4601541626999969,
      "grad_norm": 0.16204874217510223,
      "learning_rate": 1.7115779413003865e-05,
      "loss": 0.5001888275146484,
      "step": 5619
    },
    {
      "epoch": 0.4602360552365159,
      "grad_norm": 0.18134404718875885,
      "learning_rate": 1.7113516267374124e-05,
      "loss": 0.8329024314880371,
      "step": 5620
    },
    {
      "epoch": 0.4603179477730348,
      "grad_norm": 0.1889573633670807,
      "learning_rate": 1.711125289948991e-05,
      "loss": 0.7223650217056274,
      "step": 5621
    },
    {
      "epoch": 0.4603998403095538,
      "grad_norm": 0.18908809125423431,
      "learning_rate": 1.7108989309449587e-05,
      "loss": 1.00648033618927,
      "step": 5622
    },
    {
      "epoch": 0.46048173284607274,
      "grad_norm": 0.19665831327438354,
      "learning_rate": 1.7106725497351548e-05,
      "loss": 1.105470895767212,
      "step": 5623
    },
    {
      "epoch": 0.4605636253825917,
      "grad_norm": 0.19255506992340088,
      "learning_rate": 1.7104461463294186e-05,
      "loss": 1.028418779373169,
      "step": 5624
    },
    {
      "epoch": 0.4606455179191106,
      "grad_norm": 0.1941685527563095,
      "learning_rate": 1.71021972073759e-05,
      "loss": 0.7433605194091797,
      "step": 5625
    },
    {
      "epoch": 0.4607274104556296,
      "grad_norm": 0.18795400857925415,
      "learning_rate": 1.7099932729695116e-05,
      "loss": 0.7060457468032837,
      "step": 5626
    },
    {
      "epoch": 0.46080930299214856,
      "grad_norm": 0.17160651087760925,
      "learning_rate": 1.7097668030350254e-05,
      "loss": 0.7491191625595093,
      "step": 5627
    },
    {
      "epoch": 0.46089119552866753,
      "grad_norm": 0.23182187974452972,
      "learning_rate": 1.7095403109439734e-05,
      "loss": 0.8939904570579529,
      "step": 5628
    },
    {
      "epoch": 0.46097308806518644,
      "grad_norm": 0.1814865916967392,
      "learning_rate": 1.7093137967062014e-05,
      "loss": 1.0917574167251587,
      "step": 5629
    },
    {
      "epoch": 0.4610549806017054,
      "grad_norm": 0.19662444293498993,
      "learning_rate": 1.709087260331554e-05,
      "loss": 0.9021788835525513,
      "step": 5630
    },
    {
      "epoch": 0.4611368731382244,
      "grad_norm": 0.2308211475610733,
      "learning_rate": 1.708860701829877e-05,
      "loss": 0.8778523206710815,
      "step": 5631
    },
    {
      "epoch": 0.4612187656747433,
      "grad_norm": 0.21362243592739105,
      "learning_rate": 1.708634121211019e-05,
      "loss": 0.9810387492179871,
      "step": 5632
    },
    {
      "epoch": 0.46130065821126226,
      "grad_norm": 0.18830111622810364,
      "learning_rate": 1.7084075184848267e-05,
      "loss": 0.7273983955383301,
      "step": 5633
    },
    {
      "epoch": 0.46138255074778123,
      "grad_norm": 0.18773406744003296,
      "learning_rate": 1.7081808936611494e-05,
      "loss": 0.7527506351470947,
      "step": 5634
    },
    {
      "epoch": 0.4614644432843002,
      "grad_norm": 0.2082962542772293,
      "learning_rate": 1.7079542467498378e-05,
      "loss": 0.8878594040870667,
      "step": 5635
    },
    {
      "epoch": 0.4615463358208191,
      "grad_norm": 0.16195878386497498,
      "learning_rate": 1.7077275777607425e-05,
      "loss": 0.5821598768234253,
      "step": 5636
    },
    {
      "epoch": 0.4616282283573381,
      "grad_norm": 0.16819487512111664,
      "learning_rate": 1.7075008867037152e-05,
      "loss": 0.715498685836792,
      "step": 5637
    },
    {
      "epoch": 0.46171012089385705,
      "grad_norm": 0.199930340051651,
      "learning_rate": 1.7072741735886096e-05,
      "loss": 0.9459577202796936,
      "step": 5638
    },
    {
      "epoch": 0.461792013430376,
      "grad_norm": 0.1911356896162033,
      "learning_rate": 1.7070474384252784e-05,
      "loss": 0.6491736173629761,
      "step": 5639
    },
    {
      "epoch": 0.4618739059668949,
      "grad_norm": 0.18258215487003326,
      "learning_rate": 1.706820681223578e-05,
      "loss": 0.8887325525283813,
      "step": 5640
    },
    {
      "epoch": 0.4619557985034139,
      "grad_norm": 0.17445173859596252,
      "learning_rate": 1.7065939019933628e-05,
      "loss": 0.9699308276176453,
      "step": 5641
    },
    {
      "epoch": 0.46203769103993286,
      "grad_norm": 0.1519145667552948,
      "learning_rate": 1.70636710074449e-05,
      "loss": 0.7765964269638062,
      "step": 5642
    },
    {
      "epoch": 0.4621195835764518,
      "grad_norm": 0.19676879048347473,
      "learning_rate": 1.7061402774868177e-05,
      "loss": 0.6092524528503418,
      "step": 5643
    },
    {
      "epoch": 0.46220147611297074,
      "grad_norm": 0.18010647594928741,
      "learning_rate": 1.7059134322302044e-05,
      "loss": 0.8814293742179871,
      "step": 5644
    },
    {
      "epoch": 0.4622833686494897,
      "grad_norm": 0.15104055404663086,
      "learning_rate": 1.7056865649845094e-05,
      "loss": 0.6018142700195312,
      "step": 5645
    },
    {
      "epoch": 0.4623652611860087,
      "grad_norm": 0.1672787368297577,
      "learning_rate": 1.7054596757595938e-05,
      "loss": 0.796288251876831,
      "step": 5646
    },
    {
      "epoch": 0.4624471537225276,
      "grad_norm": 0.16131895780563354,
      "learning_rate": 1.705232764565318e-05,
      "loss": 0.5134752988815308,
      "step": 5647
    },
    {
      "epoch": 0.46252904625904656,
      "grad_norm": 0.17613138258457184,
      "learning_rate": 1.7050058314115466e-05,
      "loss": 0.7516793012619019,
      "step": 5648
    },
    {
      "epoch": 0.46261093879556553,
      "grad_norm": 0.15342296659946442,
      "learning_rate": 1.704778876308141e-05,
      "loss": 0.6599679589271545,
      "step": 5649
    },
    {
      "epoch": 0.4626928313320845,
      "grad_norm": 0.18254119157791138,
      "learning_rate": 1.704551899264967e-05,
      "loss": 0.901342511177063,
      "step": 5650
    },
    {
      "epoch": 0.4627747238686034,
      "grad_norm": 0.20758965611457825,
      "learning_rate": 1.7043249002918886e-05,
      "loss": 0.4898245334625244,
      "step": 5651
    },
    {
      "epoch": 0.4628566164051224,
      "grad_norm": 0.16814488172531128,
      "learning_rate": 1.704097879398773e-05,
      "loss": 0.6414906978607178,
      "step": 5652
    },
    {
      "epoch": 0.46293850894164135,
      "grad_norm": 0.266051322221756,
      "learning_rate": 1.7038708365954878e-05,
      "loss": 0.7771191000938416,
      "step": 5653
    },
    {
      "epoch": 0.46302040147816026,
      "grad_norm": 0.17315742373466492,
      "learning_rate": 1.7036437718919007e-05,
      "loss": 0.5392952561378479,
      "step": 5654
    },
    {
      "epoch": 0.4631022940146792,
      "grad_norm": 0.195817768573761,
      "learning_rate": 1.7034166852978807e-05,
      "loss": 0.5648343563079834,
      "step": 5655
    },
    {
      "epoch": 0.4631841865511982,
      "grad_norm": 0.18534733355045319,
      "learning_rate": 1.703189576823298e-05,
      "loss": 0.9492453336715698,
      "step": 5656
    },
    {
      "epoch": 0.46326607908771716,
      "grad_norm": 0.18725791573524475,
      "learning_rate": 1.702962446478024e-05,
      "loss": 0.7943506836891174,
      "step": 5657
    },
    {
      "epoch": 0.4633479716242361,
      "grad_norm": 0.1605147421360016,
      "learning_rate": 1.7027352942719306e-05,
      "loss": 0.99833744764328,
      "step": 5658
    },
    {
      "epoch": 0.46342986416075505,
      "grad_norm": 0.21395263075828552,
      "learning_rate": 1.7025081202148908e-05,
      "loss": 0.856096625328064,
      "step": 5659
    },
    {
      "epoch": 0.463511756697274,
      "grad_norm": 0.18935652077198029,
      "learning_rate": 1.702280924316778e-05,
      "loss": 0.6701532602310181,
      "step": 5660
    },
    {
      "epoch": 0.463593649233793,
      "grad_norm": 0.17860884964466095,
      "learning_rate": 1.702053706587468e-05,
      "loss": 0.583470344543457,
      "step": 5661
    },
    {
      "epoch": 0.4636755417703119,
      "grad_norm": 0.1616896390914917,
      "learning_rate": 1.7018264670368357e-05,
      "loss": 0.6640766263008118,
      "step": 5662
    },
    {
      "epoch": 0.46375743430683086,
      "grad_norm": 0.17915141582489014,
      "learning_rate": 1.7015992056747585e-05,
      "loss": 0.8074891567230225,
      "step": 5663
    },
    {
      "epoch": 0.46383932684334983,
      "grad_norm": 0.21121381223201752,
      "learning_rate": 1.7013719225111137e-05,
      "loss": 0.6954071521759033,
      "step": 5664
    },
    {
      "epoch": 0.46392121937986874,
      "grad_norm": 0.21897628903388977,
      "learning_rate": 1.70114461755578e-05,
      "loss": 0.7080338001251221,
      "step": 5665
    },
    {
      "epoch": 0.4640031119163877,
      "grad_norm": 0.20881624519824982,
      "learning_rate": 1.7009172908186374e-05,
      "loss": 0.9644142389297485,
      "step": 5666
    },
    {
      "epoch": 0.4640850044529067,
      "grad_norm": 0.19718126952648163,
      "learning_rate": 1.7006899423095663e-05,
      "loss": 0.7552162408828735,
      "step": 5667
    },
    {
      "epoch": 0.46416689698942565,
      "grad_norm": 0.19201596081256866,
      "learning_rate": 1.700462572038448e-05,
      "loss": 0.8494425415992737,
      "step": 5668
    },
    {
      "epoch": 0.46424878952594456,
      "grad_norm": 0.1904914379119873,
      "learning_rate": 1.700235180015165e-05,
      "loss": 0.888371467590332,
      "step": 5669
    },
    {
      "epoch": 0.46433068206246353,
      "grad_norm": 0.21081507205963135,
      "learning_rate": 1.700007766249601e-05,
      "loss": 0.7260885238647461,
      "step": 5670
    },
    {
      "epoch": 0.4644125745989825,
      "grad_norm": 0.1953665018081665,
      "learning_rate": 1.69978033075164e-05,
      "loss": 0.6683377027511597,
      "step": 5671
    },
    {
      "epoch": 0.46449446713550147,
      "grad_norm": 0.1951061487197876,
      "learning_rate": 1.699552873531167e-05,
      "loss": 0.7700653672218323,
      "step": 5672
    },
    {
      "epoch": 0.4645763596720204,
      "grad_norm": 0.15980330109596252,
      "learning_rate": 1.6993253945980694e-05,
      "loss": 0.7222982048988342,
      "step": 5673
    },
    {
      "epoch": 0.46465825220853935,
      "grad_norm": 0.1792273223400116,
      "learning_rate": 1.6990978939622326e-05,
      "loss": 0.9708443880081177,
      "step": 5674
    },
    {
      "epoch": 0.4647401447450583,
      "grad_norm": 0.18301260471343994,
      "learning_rate": 1.6988703716335466e-05,
      "loss": 0.99620521068573,
      "step": 5675
    },
    {
      "epoch": 0.4648220372815772,
      "grad_norm": 0.19610466063022614,
      "learning_rate": 1.6986428276218986e-05,
      "loss": 0.6621044278144836,
      "step": 5676
    },
    {
      "epoch": 0.4649039298180962,
      "grad_norm": 0.21389372646808624,
      "learning_rate": 1.6984152619371804e-05,
      "loss": 1.09792160987854,
      "step": 5677
    },
    {
      "epoch": 0.46498582235461516,
      "grad_norm": 0.16702988743782043,
      "learning_rate": 1.6981876745892817e-05,
      "loss": 0.8485398888587952,
      "step": 5678
    },
    {
      "epoch": 0.46506771489113413,
      "grad_norm": 0.18003708124160767,
      "learning_rate": 1.697960065588095e-05,
      "loss": 0.8555956482887268,
      "step": 5679
    },
    {
      "epoch": 0.46514960742765304,
      "grad_norm": 0.19652913510799408,
      "learning_rate": 1.6977324349435128e-05,
      "loss": 1.2005982398986816,
      "step": 5680
    },
    {
      "epoch": 0.465231499964172,
      "grad_norm": 0.211979478597641,
      "learning_rate": 1.697504782665429e-05,
      "loss": 0.6962770819664001,
      "step": 5681
    },
    {
      "epoch": 0.465313392500691,
      "grad_norm": 0.1556617021560669,
      "learning_rate": 1.697277108763738e-05,
      "loss": 0.7614181041717529,
      "step": 5682
    },
    {
      "epoch": 0.4653952850372099,
      "grad_norm": 0.1847931295633316,
      "learning_rate": 1.6970494132483363e-05,
      "loss": 0.6512038707733154,
      "step": 5683
    },
    {
      "epoch": 0.46547717757372886,
      "grad_norm": 0.19451624155044556,
      "learning_rate": 1.6968216961291195e-05,
      "loss": 1.0497111082077026,
      "step": 5684
    },
    {
      "epoch": 0.46555907011024783,
      "grad_norm": 0.19757571816444397,
      "learning_rate": 1.6965939574159857e-05,
      "loss": 0.7682152986526489,
      "step": 5685
    },
    {
      "epoch": 0.4656409626467668,
      "grad_norm": 0.17299173772335052,
      "learning_rate": 1.6963661971188328e-05,
      "loss": 0.9950641989707947,
      "step": 5686
    },
    {
      "epoch": 0.4657228551832857,
      "grad_norm": 0.191226065158844,
      "learning_rate": 1.696138415247561e-05,
      "loss": 1.1259377002716064,
      "step": 5687
    },
    {
      "epoch": 0.4658047477198047,
      "grad_norm": 0.17694751918315887,
      "learning_rate": 1.6959106118120697e-05,
      "loss": 0.8161852359771729,
      "step": 5688
    },
    {
      "epoch": 0.46588664025632365,
      "grad_norm": 0.1892276257276535,
      "learning_rate": 1.6956827868222615e-05,
      "loss": 0.6322022080421448,
      "step": 5689
    },
    {
      "epoch": 0.4659685327928426,
      "grad_norm": 0.16336172819137573,
      "learning_rate": 1.6954549402880375e-05,
      "loss": 0.6846575736999512,
      "step": 5690
    },
    {
      "epoch": 0.46605042532936153,
      "grad_norm": 0.17966033518314362,
      "learning_rate": 1.6952270722193008e-05,
      "loss": 0.9478905200958252,
      "step": 5691
    },
    {
      "epoch": 0.4661323178658805,
      "grad_norm": 0.21429374814033508,
      "learning_rate": 1.6949991826259567e-05,
      "loss": 0.5821859240531921,
      "step": 5692
    },
    {
      "epoch": 0.46621421040239946,
      "grad_norm": 0.23263147473335266,
      "learning_rate": 1.6947712715179086e-05,
      "loss": 0.950603723526001,
      "step": 5693
    },
    {
      "epoch": 0.4662961029389184,
      "grad_norm": 0.20973782241344452,
      "learning_rate": 1.694543338905063e-05,
      "loss": 0.8668786883354187,
      "step": 5694
    },
    {
      "epoch": 0.46637799547543735,
      "grad_norm": 0.2305840700864792,
      "learning_rate": 1.6943153847973278e-05,
      "loss": 0.8857128620147705,
      "step": 5695
    },
    {
      "epoch": 0.4664598880119563,
      "grad_norm": 0.16598130762577057,
      "learning_rate": 1.6940874092046092e-05,
      "loss": 0.8672669529914856,
      "step": 5696
    },
    {
      "epoch": 0.4665417805484753,
      "grad_norm": 0.17495568096637726,
      "learning_rate": 1.6938594121368174e-05,
      "loss": 0.7202290892601013,
      "step": 5697
    },
    {
      "epoch": 0.4666236730849942,
      "grad_norm": 0.18416143953800201,
      "learning_rate": 1.6936313936038615e-05,
      "loss": 0.9972740411758423,
      "step": 5698
    },
    {
      "epoch": 0.46670556562151316,
      "grad_norm": 0.12858416140079498,
      "learning_rate": 1.693403353615652e-05,
      "loss": 0.362326055765152,
      "step": 5699
    },
    {
      "epoch": 0.46678745815803213,
      "grad_norm": 0.20268410444259644,
      "learning_rate": 1.6931752921821008e-05,
      "loss": 1.0605305433273315,
      "step": 5700
    },
    {
      "epoch": 0.4668693506945511,
      "grad_norm": 0.17314472794532776,
      "learning_rate": 1.69294720931312e-05,
      "loss": 0.800609827041626,
      "step": 5701
    },
    {
      "epoch": 0.46695124323107,
      "grad_norm": 0.21764276921749115,
      "learning_rate": 1.6927191050186226e-05,
      "loss": 0.7787314653396606,
      "step": 5702
    },
    {
      "epoch": 0.467033135767589,
      "grad_norm": 0.1762946993112564,
      "learning_rate": 1.6924909793085245e-05,
      "loss": 0.8782187700271606,
      "step": 5703
    },
    {
      "epoch": 0.46711502830410795,
      "grad_norm": 0.16525842249393463,
      "learning_rate": 1.6922628321927394e-05,
      "loss": 0.9047735929489136,
      "step": 5704
    },
    {
      "epoch": 0.46719692084062686,
      "grad_norm": 0.17772705852985382,
      "learning_rate": 1.6920346636811842e-05,
      "loss": 0.5702717900276184,
      "step": 5705
    },
    {
      "epoch": 0.46727881337714583,
      "grad_norm": 0.1556197851896286,
      "learning_rate": 1.691806473783776e-05,
      "loss": 0.9687420725822449,
      "step": 5706
    },
    {
      "epoch": 0.4673607059136648,
      "grad_norm": 0.15444719791412354,
      "learning_rate": 1.691578262510433e-05,
      "loss": 0.7855178117752075,
      "step": 5707
    },
    {
      "epoch": 0.46744259845018377,
      "grad_norm": 0.19381767511367798,
      "learning_rate": 1.691350029871074e-05,
      "loss": 0.8316521644592285,
      "step": 5708
    },
    {
      "epoch": 0.4675244909867027,
      "grad_norm": 0.17090266942977905,
      "learning_rate": 1.6911217758756188e-05,
      "loss": 0.9096817970275879,
      "step": 5709
    },
    {
      "epoch": 0.46760638352322165,
      "grad_norm": 0.17258812487125397,
      "learning_rate": 1.6908935005339888e-05,
      "loss": 0.8648214340209961,
      "step": 5710
    },
    {
      "epoch": 0.4676882760597406,
      "grad_norm": 0.19979976117610931,
      "learning_rate": 1.690665203856105e-05,
      "loss": 1.2726770639419556,
      "step": 5711
    },
    {
      "epoch": 0.4677701685962596,
      "grad_norm": 0.16230051219463348,
      "learning_rate": 1.690436885851891e-05,
      "loss": 0.7734116911888123,
      "step": 5712
    },
    {
      "epoch": 0.4678520611327785,
      "grad_norm": 0.21505683660507202,
      "learning_rate": 1.6902085465312692e-05,
      "loss": 0.830539882183075,
      "step": 5713
    },
    {
      "epoch": 0.46793395366929746,
      "grad_norm": 0.1595895141363144,
      "learning_rate": 1.689980185904166e-05,
      "loss": 0.5812140703201294,
      "step": 5714
    },
    {
      "epoch": 0.46801584620581643,
      "grad_norm": 0.21550410985946655,
      "learning_rate": 1.689751803980505e-05,
      "loss": 1.1381564140319824,
      "step": 5715
    },
    {
      "epoch": 0.46809773874233535,
      "grad_norm": 0.21947431564331055,
      "learning_rate": 1.6895234007702135e-05,
      "loss": 0.8519974946975708,
      "step": 5716
    },
    {
      "epoch": 0.4681796312788543,
      "grad_norm": 0.1863469034433365,
      "learning_rate": 1.6892949762832193e-05,
      "loss": 0.9654135704040527,
      "step": 5717
    },
    {
      "epoch": 0.4682615238153733,
      "grad_norm": 0.16923269629478455,
      "learning_rate": 1.68906653052945e-05,
      "loss": 1.1130108833312988,
      "step": 5718
    },
    {
      "epoch": 0.46834341635189225,
      "grad_norm": 0.17674696445465088,
      "learning_rate": 1.688838063518835e-05,
      "loss": 0.6391989588737488,
      "step": 5719
    },
    {
      "epoch": 0.46842530888841116,
      "grad_norm": 0.14777280390262604,
      "learning_rate": 1.6886095752613043e-05,
      "loss": 0.776767373085022,
      "step": 5720
    },
    {
      "epoch": 0.46850720142493013,
      "grad_norm": 0.21793873608112335,
      "learning_rate": 1.688381065766789e-05,
      "loss": 1.19598388671875,
      "step": 5721
    },
    {
      "epoch": 0.4685890939614491,
      "grad_norm": 0.16271845996379852,
      "learning_rate": 1.688152535045221e-05,
      "loss": 0.8221518397331238,
      "step": 5722
    },
    {
      "epoch": 0.46867098649796807,
      "grad_norm": 0.16399644315242767,
      "learning_rate": 1.6879239831065332e-05,
      "loss": 0.6849920749664307,
      "step": 5723
    },
    {
      "epoch": 0.468752879034487,
      "grad_norm": 0.17152543365955353,
      "learning_rate": 1.68769540996066e-05,
      "loss": 0.8478953242301941,
      "step": 5724
    },
    {
      "epoch": 0.46883477157100595,
      "grad_norm": 0.2091539055109024,
      "learning_rate": 1.687466815617535e-05,
      "loss": 1.0456154346466064,
      "step": 5725
    },
    {
      "epoch": 0.4689166641075249,
      "grad_norm": 0.18918949365615845,
      "learning_rate": 1.687238200087095e-05,
      "loss": 0.692163348197937,
      "step": 5726
    },
    {
      "epoch": 0.46899855664404383,
      "grad_norm": 0.18937453627586365,
      "learning_rate": 1.6870095633792758e-05,
      "loss": 0.8134610056877136,
      "step": 5727
    },
    {
      "epoch": 0.4690804491805628,
      "grad_norm": 0.17877361178398132,
      "learning_rate": 1.686780905504015e-05,
      "loss": 0.9310188293457031,
      "step": 5728
    },
    {
      "epoch": 0.46916234171708177,
      "grad_norm": 0.1685734987258911,
      "learning_rate": 1.686552226471251e-05,
      "loss": 0.940948486328125,
      "step": 5729
    },
    {
      "epoch": 0.46924423425360073,
      "grad_norm": 0.2175823301076889,
      "learning_rate": 1.6863235262909236e-05,
      "loss": 0.9955415725708008,
      "step": 5730
    },
    {
      "epoch": 0.46932612679011965,
      "grad_norm": 0.17485475540161133,
      "learning_rate": 1.6860948049729726e-05,
      "loss": 0.9868797659873962,
      "step": 5731
    },
    {
      "epoch": 0.4694080193266386,
      "grad_norm": 0.19364407658576965,
      "learning_rate": 1.6858660625273395e-05,
      "loss": 0.7460997104644775,
      "step": 5732
    },
    {
      "epoch": 0.4694899118631576,
      "grad_norm": 0.1785038411617279,
      "learning_rate": 1.6856372989639662e-05,
      "loss": 0.9369797706604004,
      "step": 5733
    },
    {
      "epoch": 0.46957180439967655,
      "grad_norm": 0.1772373914718628,
      "learning_rate": 1.6854085142927954e-05,
      "loss": 0.7768234610557556,
      "step": 5734
    },
    {
      "epoch": 0.46965369693619546,
      "grad_norm": 0.19427774846553802,
      "learning_rate": 1.6851797085237717e-05,
      "loss": 0.7026744484901428,
      "step": 5735
    },
    {
      "epoch": 0.46973558947271443,
      "grad_norm": 0.20748481154441833,
      "learning_rate": 1.6849508816668392e-05,
      "loss": 1.211838722229004,
      "step": 5736
    },
    {
      "epoch": 0.4698174820092334,
      "grad_norm": 0.20712335407733917,
      "learning_rate": 1.6847220337319438e-05,
      "loss": 0.8668811321258545,
      "step": 5737
    },
    {
      "epoch": 0.4698993745457523,
      "grad_norm": 0.21387356519699097,
      "learning_rate": 1.6844931647290327e-05,
      "loss": 1.122694492340088,
      "step": 5738
    },
    {
      "epoch": 0.4699812670822713,
      "grad_norm": 0.19650539755821228,
      "learning_rate": 1.684264274668053e-05,
      "loss": 1.053371787071228,
      "step": 5739
    },
    {
      "epoch": 0.47006315961879025,
      "grad_norm": 0.17744415998458862,
      "learning_rate": 1.684035363558954e-05,
      "loss": 0.8609519004821777,
      "step": 5740
    },
    {
      "epoch": 0.4701450521553092,
      "grad_norm": 0.16463683545589447,
      "learning_rate": 1.683806431411684e-05,
      "loss": 0.5769405364990234,
      "step": 5741
    },
    {
      "epoch": 0.47022694469182813,
      "grad_norm": 0.2522983253002167,
      "learning_rate": 1.683577478236194e-05,
      "loss": 0.9162659645080566,
      "step": 5742
    },
    {
      "epoch": 0.4703088372283471,
      "grad_norm": 0.21955548226833344,
      "learning_rate": 1.683348504042435e-05,
      "loss": 0.8730915784835815,
      "step": 5743
    },
    {
      "epoch": 0.47039072976486607,
      "grad_norm": 0.20450858771800995,
      "learning_rate": 1.6831195088403594e-05,
      "loss": 1.0019282102584839,
      "step": 5744
    },
    {
      "epoch": 0.47047262230138504,
      "grad_norm": 0.20428945124149323,
      "learning_rate": 1.6828904926399198e-05,
      "loss": 0.8124568462371826,
      "step": 5745
    },
    {
      "epoch": 0.47055451483790395,
      "grad_norm": 0.1979932337999344,
      "learning_rate": 1.6826614554510706e-05,
      "loss": 1.0529217720031738,
      "step": 5746
    },
    {
      "epoch": 0.4706364073744229,
      "grad_norm": 0.1792546957731247,
      "learning_rate": 1.6824323972837673e-05,
      "loss": 0.7455668449401855,
      "step": 5747
    },
    {
      "epoch": 0.4707182999109419,
      "grad_norm": 0.15432637929916382,
      "learning_rate": 1.6822033181479638e-05,
      "loss": 0.9391785860061646,
      "step": 5748
    },
    {
      "epoch": 0.4708001924474608,
      "grad_norm": 0.2170325368642807,
      "learning_rate": 1.6819742180536188e-05,
      "loss": 0.7375652194023132,
      "step": 5749
    },
    {
      "epoch": 0.47088208498397977,
      "grad_norm": 0.1708453744649887,
      "learning_rate": 1.6817450970106895e-05,
      "loss": 0.6354907751083374,
      "step": 5750
    },
    {
      "epoch": 0.47096397752049873,
      "grad_norm": 0.1957627683877945,
      "learning_rate": 1.681515955029133e-05,
      "loss": 0.7261526584625244,
      "step": 5751
    },
    {
      "epoch": 0.4710458700570177,
      "grad_norm": 0.1713513880968094,
      "learning_rate": 1.681286792118911e-05,
      "loss": 1.0553860664367676,
      "step": 5752
    },
    {
      "epoch": 0.4711277625935366,
      "grad_norm": 0.20723366737365723,
      "learning_rate": 1.6810576082899818e-05,
      "loss": 0.8606569170951843,
      "step": 5753
    },
    {
      "epoch": 0.4712096551300556,
      "grad_norm": 0.16217076778411865,
      "learning_rate": 1.6808284035523084e-05,
      "loss": 0.7416989207267761,
      "step": 5754
    },
    {
      "epoch": 0.47129154766657455,
      "grad_norm": 0.20789554715156555,
      "learning_rate": 1.680599177915852e-05,
      "loss": 0.7503676414489746,
      "step": 5755
    },
    {
      "epoch": 0.47137344020309346,
      "grad_norm": 0.17819589376449585,
      "learning_rate": 1.6803699313905755e-05,
      "loss": 0.6370301842689514,
      "step": 5756
    },
    {
      "epoch": 0.47145533273961243,
      "grad_norm": 0.20768415927886963,
      "learning_rate": 1.6801406639864434e-05,
      "loss": 0.9824140071868896,
      "step": 5757
    },
    {
      "epoch": 0.4715372252761314,
      "grad_norm": 0.18644043803215027,
      "learning_rate": 1.6799113757134212e-05,
      "loss": 0.907228946685791,
      "step": 5758
    },
    {
      "epoch": 0.47161911781265037,
      "grad_norm": 0.22746238112449646,
      "learning_rate": 1.679682066581473e-05,
      "loss": 0.6297433376312256,
      "step": 5759
    },
    {
      "epoch": 0.4717010103491693,
      "grad_norm": 0.16416217386722565,
      "learning_rate": 1.679452736600567e-05,
      "loss": 0.8742093443870544,
      "step": 5760
    },
    {
      "epoch": 0.47178290288568825,
      "grad_norm": 0.19041581451892853,
      "learning_rate": 1.6792233857806707e-05,
      "loss": 0.989783763885498,
      "step": 5761
    },
    {
      "epoch": 0.4718647954222072,
      "grad_norm": 0.18986694514751434,
      "learning_rate": 1.6789940141317516e-05,
      "loss": 0.7801942825317383,
      "step": 5762
    },
    {
      "epoch": 0.4719466879587262,
      "grad_norm": 0.15892240405082703,
      "learning_rate": 1.67876462166378e-05,
      "loss": 0.6128737330436707,
      "step": 5763
    },
    {
      "epoch": 0.4720285804952451,
      "grad_norm": 0.21373392641544342,
      "learning_rate": 1.6785352083867262e-05,
      "loss": 0.9701625108718872,
      "step": 5764
    },
    {
      "epoch": 0.47211047303176407,
      "grad_norm": 0.21070343255996704,
      "learning_rate": 1.678305774310561e-05,
      "loss": 1.1207671165466309,
      "step": 5765
    },
    {
      "epoch": 0.47219236556828303,
      "grad_norm": 0.1858547031879425,
      "learning_rate": 1.6780763194452577e-05,
      "loss": 0.6909077167510986,
      "step": 5766
    },
    {
      "epoch": 0.47227425810480195,
      "grad_norm": 0.18076090514659882,
      "learning_rate": 1.6778468438007877e-05,
      "loss": 0.9301475882530212,
      "step": 5767
    },
    {
      "epoch": 0.4723561506413209,
      "grad_norm": 0.17048142850399017,
      "learning_rate": 1.6776173473871257e-05,
      "loss": 0.8735812902450562,
      "step": 5768
    },
    {
      "epoch": 0.4724380431778399,
      "grad_norm": 0.19194333255290985,
      "learning_rate": 1.677387830214247e-05,
      "loss": 1.2091397047042847,
      "step": 5769
    },
    {
      "epoch": 0.47251993571435885,
      "grad_norm": 0.1814650595188141,
      "learning_rate": 1.677158292292127e-05,
      "loss": 0.6549610495567322,
      "step": 5770
    },
    {
      "epoch": 0.47260182825087776,
      "grad_norm": 0.16812385618686676,
      "learning_rate": 1.6769287336307425e-05,
      "loss": 0.7203570604324341,
      "step": 5771
    },
    {
      "epoch": 0.47268372078739673,
      "grad_norm": 0.1686863750219345,
      "learning_rate": 1.6766991542400705e-05,
      "loss": 0.73825603723526,
      "step": 5772
    },
    {
      "epoch": 0.4727656133239157,
      "grad_norm": 0.1793726533651352,
      "learning_rate": 1.67646955413009e-05,
      "loss": 0.5963665246963501,
      "step": 5773
    },
    {
      "epoch": 0.47284750586043467,
      "grad_norm": 0.1557624638080597,
      "learning_rate": 1.6762399333107803e-05,
      "loss": 0.6332284212112427,
      "step": 5774
    },
    {
      "epoch": 0.4729293983969536,
      "grad_norm": 0.19557465612888336,
      "learning_rate": 1.6760102917921217e-05,
      "loss": 0.7497131824493408,
      "step": 5775
    },
    {
      "epoch": 0.47301129093347255,
      "grad_norm": 0.1806601881980896,
      "learning_rate": 1.6757806295840948e-05,
      "loss": 0.675439715385437,
      "step": 5776
    },
    {
      "epoch": 0.4730931834699915,
      "grad_norm": 0.18669982254505157,
      "learning_rate": 1.6755509466966828e-05,
      "loss": 1.1132562160491943,
      "step": 5777
    },
    {
      "epoch": 0.47317507600651043,
      "grad_norm": 0.194931298494339,
      "learning_rate": 1.6753212431398675e-05,
      "loss": 0.9010046124458313,
      "step": 5778
    },
    {
      "epoch": 0.4732569685430294,
      "grad_norm": 0.2079889327287674,
      "learning_rate": 1.6750915189236336e-05,
      "loss": 0.8490778803825378,
      "step": 5779
    },
    {
      "epoch": 0.47333886107954837,
      "grad_norm": 0.20925164222717285,
      "learning_rate": 1.6748617740579653e-05,
      "loss": 1.3072772026062012,
      "step": 5780
    },
    {
      "epoch": 0.47342075361606734,
      "grad_norm": 0.18591450154781342,
      "learning_rate": 1.6746320085528485e-05,
      "loss": 0.7093144655227661,
      "step": 5781
    },
    {
      "epoch": 0.47350264615258625,
      "grad_norm": 0.19523046910762787,
      "learning_rate": 1.67440222241827e-05,
      "loss": 0.8174944519996643,
      "step": 5782
    },
    {
      "epoch": 0.4735845386891052,
      "grad_norm": 0.17547807097434998,
      "learning_rate": 1.6741724156642167e-05,
      "loss": 0.8408262133598328,
      "step": 5783
    },
    {
      "epoch": 0.4736664312256242,
      "grad_norm": 0.18496105074882507,
      "learning_rate": 1.6739425883006775e-05,
      "loss": 1.0224480628967285,
      "step": 5784
    },
    {
      "epoch": 0.47374832376214315,
      "grad_norm": 0.15358321368694305,
      "learning_rate": 1.6737127403376412e-05,
      "loss": 0.6119422912597656,
      "step": 5785
    },
    {
      "epoch": 0.47383021629866207,
      "grad_norm": 0.19539202749729156,
      "learning_rate": 1.673482871785098e-05,
      "loss": 0.880734384059906,
      "step": 5786
    },
    {
      "epoch": 0.47391210883518103,
      "grad_norm": 0.1827968955039978,
      "learning_rate": 1.6732529826530395e-05,
      "loss": 0.4837223291397095,
      "step": 5787
    },
    {
      "epoch": 0.4739940013717,
      "grad_norm": 0.20449529588222504,
      "learning_rate": 1.673023072951457e-05,
      "loss": 0.9290793538093567,
      "step": 5788
    },
    {
      "epoch": 0.4740758939082189,
      "grad_norm": 0.20017297565937042,
      "learning_rate": 1.6727931426903434e-05,
      "loss": 0.9493145942687988,
      "step": 5789
    },
    {
      "epoch": 0.4741577864447379,
      "grad_norm": 0.20536845922470093,
      "learning_rate": 1.6725631918796925e-05,
      "loss": 0.7183416485786438,
      "step": 5790
    },
    {
      "epoch": 0.47423967898125685,
      "grad_norm": 0.12783238291740417,
      "learning_rate": 1.672333220529499e-05,
      "loss": 0.4309667646884918,
      "step": 5791
    },
    {
      "epoch": 0.4743215715177758,
      "grad_norm": 0.17433913052082062,
      "learning_rate": 1.672103228649758e-05,
      "loss": 0.7464715838432312,
      "step": 5792
    },
    {
      "epoch": 0.47440346405429473,
      "grad_norm": 0.19139544665813446,
      "learning_rate": 1.6718732162504667e-05,
      "loss": 0.9179016947746277,
      "step": 5793
    },
    {
      "epoch": 0.4744853565908137,
      "grad_norm": 0.198594868183136,
      "learning_rate": 1.6716431833416218e-05,
      "loss": 0.780087947845459,
      "step": 5794
    },
    {
      "epoch": 0.47456724912733267,
      "grad_norm": 0.17975525557994843,
      "learning_rate": 1.671413129933222e-05,
      "loss": 0.650503396987915,
      "step": 5795
    },
    {
      "epoch": 0.47464914166385164,
      "grad_norm": 0.1643456369638443,
      "learning_rate": 1.671183056035265e-05,
      "loss": 0.8135290741920471,
      "step": 5796
    },
    {
      "epoch": 0.47473103420037055,
      "grad_norm": 0.15091553330421448,
      "learning_rate": 1.6709529616577525e-05,
      "loss": 0.5742231607437134,
      "step": 5797
    },
    {
      "epoch": 0.4748129267368895,
      "grad_norm": 0.17277908325195312,
      "learning_rate": 1.670722846810684e-05,
      "loss": 0.9630283117294312,
      "step": 5798
    },
    {
      "epoch": 0.4748948192734085,
      "grad_norm": 0.2050812989473343,
      "learning_rate": 1.6704927115040627e-05,
      "loss": 0.9857592582702637,
      "step": 5799
    },
    {
      "epoch": 0.4749767118099274,
      "grad_norm": 0.15594175457954407,
      "learning_rate": 1.6702625557478898e-05,
      "loss": 0.9345353841781616,
      "step": 5800
    },
    {
      "epoch": 0.47505860434644637,
      "grad_norm": 0.14824168384075165,
      "learning_rate": 1.6700323795521692e-05,
      "loss": 0.648530900478363,
      "step": 5801
    },
    {
      "epoch": 0.47514049688296534,
      "grad_norm": 0.16983556747436523,
      "learning_rate": 1.6698021829269056e-05,
      "loss": 0.7262153625488281,
      "step": 5802
    },
    {
      "epoch": 0.4752223894194843,
      "grad_norm": 0.14108283817768097,
      "learning_rate": 1.6695719658821043e-05,
      "loss": 0.41172972321510315,
      "step": 5803
    },
    {
      "epoch": 0.4753042819560032,
      "grad_norm": 0.19198612868785858,
      "learning_rate": 1.669341728427771e-05,
      "loss": 0.7691149711608887,
      "step": 5804
    },
    {
      "epoch": 0.4753861744925222,
      "grad_norm": 0.22505192458629608,
      "learning_rate": 1.6691114705739132e-05,
      "loss": 0.998029351234436,
      "step": 5805
    },
    {
      "epoch": 0.47546806702904115,
      "grad_norm": 0.1786547303199768,
      "learning_rate": 1.668881192330539e-05,
      "loss": 0.8285019397735596,
      "step": 5806
    },
    {
      "epoch": 0.4755499595655601,
      "grad_norm": 0.21969948709011078,
      "learning_rate": 1.668650893707657e-05,
      "loss": 0.9418721199035645,
      "step": 5807
    },
    {
      "epoch": 0.47563185210207903,
      "grad_norm": 0.17640012502670288,
      "learning_rate": 1.6684205747152766e-05,
      "loss": 0.6732139587402344,
      "step": 5808
    },
    {
      "epoch": 0.475713744638598,
      "grad_norm": 0.17091907560825348,
      "learning_rate": 1.668190235363409e-05,
      "loss": 0.7789372801780701,
      "step": 5809
    },
    {
      "epoch": 0.47579563717511697,
      "grad_norm": 0.21132904291152954,
      "learning_rate": 1.6679598756620654e-05,
      "loss": 0.6547149419784546,
      "step": 5810
    },
    {
      "epoch": 0.4758775297116359,
      "grad_norm": 0.15671876072883606,
      "learning_rate": 1.6677294956212584e-05,
      "loss": 0.8701981902122498,
      "step": 5811
    },
    {
      "epoch": 0.47595942224815485,
      "grad_norm": 0.17224308848381042,
      "learning_rate": 1.6674990952510008e-05,
      "loss": 0.6987871527671814,
      "step": 5812
    },
    {
      "epoch": 0.4760413147846738,
      "grad_norm": 0.15286237001419067,
      "learning_rate": 1.667268674561307e-05,
      "loss": 0.7890997529029846,
      "step": 5813
    },
    {
      "epoch": 0.4761232073211928,
      "grad_norm": 0.18529218435287476,
      "learning_rate": 1.6670382335621924e-05,
      "loss": 0.9518860578536987,
      "step": 5814
    },
    {
      "epoch": 0.4762050998577117,
      "grad_norm": 0.17764198780059814,
      "learning_rate": 1.6668077722636724e-05,
      "loss": 0.8647974729537964,
      "step": 5815
    },
    {
      "epoch": 0.47628699239423067,
      "grad_norm": 0.1884479820728302,
      "learning_rate": 1.6665772906757636e-05,
      "loss": 0.9043370485305786,
      "step": 5816
    },
    {
      "epoch": 0.47636888493074964,
      "grad_norm": 0.1891394853591919,
      "learning_rate": 1.6663467888084843e-05,
      "loss": 1.0571242570877075,
      "step": 5817
    },
    {
      "epoch": 0.47645077746726855,
      "grad_norm": 0.1889127939939499,
      "learning_rate": 1.6661162666718534e-05,
      "loss": 0.6697296500205994,
      "step": 5818
    },
    {
      "epoch": 0.4765326700037875,
      "grad_norm": 0.1529742181301117,
      "learning_rate": 1.665885724275889e-05,
      "loss": 0.6400282979011536,
      "step": 5819
    },
    {
      "epoch": 0.4766145625403065,
      "grad_norm": 0.1753159463405609,
      "learning_rate": 1.6656551616306124e-05,
      "loss": 0.9439696073532104,
      "step": 5820
    },
    {
      "epoch": 0.47669645507682545,
      "grad_norm": 0.18142856657505035,
      "learning_rate": 1.665424578746045e-05,
      "loss": 0.8795878887176514,
      "step": 5821
    },
    {
      "epoch": 0.47677834761334437,
      "grad_norm": 0.1863868534564972,
      "learning_rate": 1.6651939756322077e-05,
      "loss": 0.8876465559005737,
      "step": 5822
    },
    {
      "epoch": 0.47686024014986333,
      "grad_norm": 0.18236854672431946,
      "learning_rate": 1.6649633522991248e-05,
      "loss": 0.970358669757843,
      "step": 5823
    },
    {
      "epoch": 0.4769421326863823,
      "grad_norm": 0.1758289635181427,
      "learning_rate": 1.6647327087568193e-05,
      "loss": 0.8860161900520325,
      "step": 5824
    },
    {
      "epoch": 0.47702402522290127,
      "grad_norm": 0.21044903993606567,
      "learning_rate": 1.664502045015316e-05,
      "loss": 0.9439597129821777,
      "step": 5825
    },
    {
      "epoch": 0.4771059177594202,
      "grad_norm": 0.18804138898849487,
      "learning_rate": 1.6642713610846414e-05,
      "loss": 0.6427783966064453,
      "step": 5826
    },
    {
      "epoch": 0.47718781029593915,
      "grad_norm": 0.15707151591777802,
      "learning_rate": 1.6640406569748205e-05,
      "loss": 0.9073885679244995,
      "step": 5827
    },
    {
      "epoch": 0.4772697028324581,
      "grad_norm": 0.21776331961154938,
      "learning_rate": 1.663809932695882e-05,
      "loss": 1.0187879800796509,
      "step": 5828
    },
    {
      "epoch": 0.47735159536897703,
      "grad_norm": 0.19416797161102295,
      "learning_rate": 1.6635791882578535e-05,
      "loss": 0.7169744968414307,
      "step": 5829
    },
    {
      "epoch": 0.477433487905496,
      "grad_norm": 0.2072611153125763,
      "learning_rate": 1.6633484236707635e-05,
      "loss": 1.0421209335327148,
      "step": 5830
    },
    {
      "epoch": 0.47751538044201497,
      "grad_norm": 0.21846184134483337,
      "learning_rate": 1.6631176389446432e-05,
      "loss": 0.8848848342895508,
      "step": 5831
    },
    {
      "epoch": 0.47759727297853394,
      "grad_norm": 0.1447957307100296,
      "learning_rate": 1.662886834089523e-05,
      "loss": 0.4993537664413452,
      "step": 5832
    },
    {
      "epoch": 0.47767916551505285,
      "grad_norm": 0.18340003490447998,
      "learning_rate": 1.6626560091154337e-05,
      "loss": 0.6100228428840637,
      "step": 5833
    },
    {
      "epoch": 0.4777610580515718,
      "grad_norm": 0.19292204082012177,
      "learning_rate": 1.6624251640324093e-05,
      "loss": 0.5712056756019592,
      "step": 5834
    },
    {
      "epoch": 0.4778429505880908,
      "grad_norm": 0.16753728687763214,
      "learning_rate": 1.6621942988504826e-05,
      "loss": 0.730812132358551,
      "step": 5835
    },
    {
      "epoch": 0.47792484312460976,
      "grad_norm": 0.1869751363992691,
      "learning_rate": 1.6619634135796877e-05,
      "loss": 0.8702684044837952,
      "step": 5836
    },
    {
      "epoch": 0.47800673566112867,
      "grad_norm": 0.17108774185180664,
      "learning_rate": 1.6617325082300607e-05,
      "loss": 0.6573414206504822,
      "step": 5837
    },
    {
      "epoch": 0.47808862819764764,
      "grad_norm": 0.19015993177890778,
      "learning_rate": 1.661501582811637e-05,
      "loss": 0.8195396065711975,
      "step": 5838
    },
    {
      "epoch": 0.4781705207341666,
      "grad_norm": 0.17832225561141968,
      "learning_rate": 1.6612706373344534e-05,
      "loss": 0.7135940790176392,
      "step": 5839
    },
    {
      "epoch": 0.4782524132706855,
      "grad_norm": 0.17596618831157684,
      "learning_rate": 1.661039671808548e-05,
      "loss": 0.9052462577819824,
      "step": 5840
    },
    {
      "epoch": 0.4783343058072045,
      "grad_norm": 0.21407197415828705,
      "learning_rate": 1.6608086862439602e-05,
      "loss": 0.8048447966575623,
      "step": 5841
    },
    {
      "epoch": 0.47841619834372345,
      "grad_norm": 0.19464482367038727,
      "learning_rate": 1.6605776806507287e-05,
      "loss": 1.0843486785888672,
      "step": 5842
    },
    {
      "epoch": 0.4784980908802424,
      "grad_norm": 0.16307486593723297,
      "learning_rate": 1.6603466550388943e-05,
      "loss": 0.8358691334724426,
      "step": 5843
    },
    {
      "epoch": 0.47857998341676133,
      "grad_norm": 0.1839357316493988,
      "learning_rate": 1.6601156094184977e-05,
      "loss": 1.0746279954910278,
      "step": 5844
    },
    {
      "epoch": 0.4786618759532803,
      "grad_norm": 0.17272637784481049,
      "learning_rate": 1.659884543799582e-05,
      "loss": 0.7889348268508911,
      "step": 5845
    },
    {
      "epoch": 0.47874376848979927,
      "grad_norm": 0.20222406089305878,
      "learning_rate": 1.6596534581921903e-05,
      "loss": 0.8294821977615356,
      "step": 5846
    },
    {
      "epoch": 0.47882566102631824,
      "grad_norm": 0.15591055154800415,
      "learning_rate": 1.6594223526063653e-05,
      "loss": 0.7688875794410706,
      "step": 5847
    },
    {
      "epoch": 0.47890755356283715,
      "grad_norm": 0.18662510812282562,
      "learning_rate": 1.6591912270521536e-05,
      "loss": 0.7065189480781555,
      "step": 5848
    },
    {
      "epoch": 0.4789894460993561,
      "grad_norm": 0.1918013095855713,
      "learning_rate": 1.6589600815395993e-05,
      "loss": 0.7625279426574707,
      "step": 5849
    },
    {
      "epoch": 0.4790713386358751,
      "grad_norm": 0.20438404381275177,
      "learning_rate": 1.6587289160787493e-05,
      "loss": 0.9446308016777039,
      "step": 5850
    },
    {
      "epoch": 0.479153231172394,
      "grad_norm": 0.24744661152362823,
      "learning_rate": 1.6584977306796518e-05,
      "loss": 0.7864745855331421,
      "step": 5851
    },
    {
      "epoch": 0.47923512370891297,
      "grad_norm": 0.18516987562179565,
      "learning_rate": 1.6582665253523545e-05,
      "loss": 1.0437281131744385,
      "step": 5852
    },
    {
      "epoch": 0.47931701624543194,
      "grad_norm": 0.16550351679325104,
      "learning_rate": 1.6580353001069063e-05,
      "loss": 0.8916025757789612,
      "step": 5853
    },
    {
      "epoch": 0.4793989087819509,
      "grad_norm": 0.19893646240234375,
      "learning_rate": 1.6578040549533576e-05,
      "loss": 0.7225375175476074,
      "step": 5854
    },
    {
      "epoch": 0.4794808013184698,
      "grad_norm": 0.20850993692874908,
      "learning_rate": 1.657572789901759e-05,
      "loss": 0.7167333364486694,
      "step": 5855
    },
    {
      "epoch": 0.4795626938549888,
      "grad_norm": 0.1947079598903656,
      "learning_rate": 1.6573415049621623e-05,
      "loss": 0.9935597777366638,
      "step": 5856
    },
    {
      "epoch": 0.47964458639150775,
      "grad_norm": 0.2638479173183441,
      "learning_rate": 1.65711020014462e-05,
      "loss": 0.7906663417816162,
      "step": 5857
    },
    {
      "epoch": 0.4797264789280267,
      "grad_norm": 0.19740994274616241,
      "learning_rate": 1.656878875459186e-05,
      "loss": 0.8551731109619141,
      "step": 5858
    },
    {
      "epoch": 0.47980837146454564,
      "grad_norm": 0.19004499912261963,
      "learning_rate": 1.656647530915914e-05,
      "loss": 0.8458139896392822,
      "step": 5859
    },
    {
      "epoch": 0.4798902640010646,
      "grad_norm": 0.15888270735740662,
      "learning_rate": 1.65641616652486e-05,
      "loss": 0.9409599304199219,
      "step": 5860
    },
    {
      "epoch": 0.47997215653758357,
      "grad_norm": 0.1668064445257187,
      "learning_rate": 1.6561847822960792e-05,
      "loss": 1.0055943727493286,
      "step": 5861
    },
    {
      "epoch": 0.4800540490741025,
      "grad_norm": 0.17890961468219757,
      "learning_rate": 1.6559533782396293e-05,
      "loss": 0.997800886631012,
      "step": 5862
    },
    {
      "epoch": 0.48013594161062145,
      "grad_norm": 0.18552571535110474,
      "learning_rate": 1.6557219543655673e-05,
      "loss": 0.9187594652175903,
      "step": 5863
    },
    {
      "epoch": 0.4802178341471404,
      "grad_norm": 0.1885429322719574,
      "learning_rate": 1.6554905106839518e-05,
      "loss": 0.7258826494216919,
      "step": 5864
    },
    {
      "epoch": 0.4802997266836594,
      "grad_norm": 0.1855742633342743,
      "learning_rate": 1.6552590472048433e-05,
      "loss": 0.7395814061164856,
      "step": 5865
    },
    {
      "epoch": 0.4803816192201783,
      "grad_norm": 0.15087305009365082,
      "learning_rate": 1.655027563938301e-05,
      "loss": 0.761139452457428,
      "step": 5866
    },
    {
      "epoch": 0.48046351175669727,
      "grad_norm": 0.2607874870300293,
      "learning_rate": 1.654796060894387e-05,
      "loss": 0.7329815030097961,
      "step": 5867
    },
    {
      "epoch": 0.48054540429321624,
      "grad_norm": 0.17394457757472992,
      "learning_rate": 1.6545645380831633e-05,
      "loss": 0.8681674003601074,
      "step": 5868
    },
    {
      "epoch": 0.4806272968297352,
      "grad_norm": 0.2048947811126709,
      "learning_rate": 1.6543329955146922e-05,
      "loss": 1.0877737998962402,
      "step": 5869
    },
    {
      "epoch": 0.4807091893662541,
      "grad_norm": 0.21914289891719818,
      "learning_rate": 1.654101433199038e-05,
      "loss": 0.6633110642433167,
      "step": 5870
    },
    {
      "epoch": 0.4807910819027731,
      "grad_norm": 0.1721796989440918,
      "learning_rate": 1.653869851146265e-05,
      "loss": 1.1039314270019531,
      "step": 5871
    },
    {
      "epoch": 0.48087297443929206,
      "grad_norm": 0.22088006138801575,
      "learning_rate": 1.6536382493664392e-05,
      "loss": 0.8358103036880493,
      "step": 5872
    },
    {
      "epoch": 0.48095486697581097,
      "grad_norm": 0.21586157381534576,
      "learning_rate": 1.6534066278696264e-05,
      "loss": 0.7305868864059448,
      "step": 5873
    },
    {
      "epoch": 0.48103675951232994,
      "grad_norm": 0.18070310354232788,
      "learning_rate": 1.6531749866658945e-05,
      "loss": 0.6932225227355957,
      "step": 5874
    },
    {
      "epoch": 0.4811186520488489,
      "grad_norm": 0.21230816841125488,
      "learning_rate": 1.6529433257653106e-05,
      "loss": 0.614865243434906,
      "step": 5875
    },
    {
      "epoch": 0.4812005445853679,
      "grad_norm": 0.1732427477836609,
      "learning_rate": 1.6527116451779446e-05,
      "loss": 0.8121309280395508,
      "step": 5876
    },
    {
      "epoch": 0.4812824371218868,
      "grad_norm": 0.20355747640132904,
      "learning_rate": 1.652479944913866e-05,
      "loss": 1.0225627422332764,
      "step": 5877
    },
    {
      "epoch": 0.48136432965840575,
      "grad_norm": 0.19230307638645172,
      "learning_rate": 1.6522482249831454e-05,
      "loss": 0.5575873255729675,
      "step": 5878
    },
    {
      "epoch": 0.4814462221949247,
      "grad_norm": 0.180848628282547,
      "learning_rate": 1.652016485395854e-05,
      "loss": 0.7529196739196777,
      "step": 5879
    },
    {
      "epoch": 0.4815281147314437,
      "grad_norm": 0.21735979616641998,
      "learning_rate": 1.6517847261620642e-05,
      "loss": 1.1100118160247803,
      "step": 5880
    },
    {
      "epoch": 0.4816100072679626,
      "grad_norm": 0.22147461771965027,
      "learning_rate": 1.65155294729185e-05,
      "loss": 1.0712127685546875,
      "step": 5881
    },
    {
      "epoch": 0.48169189980448157,
      "grad_norm": 0.1745114028453827,
      "learning_rate": 1.6513211487952848e-05,
      "loss": 0.7644980549812317,
      "step": 5882
    },
    {
      "epoch": 0.48177379234100054,
      "grad_norm": 0.18079325556755066,
      "learning_rate": 1.6510893306824433e-05,
      "loss": 0.843749463558197,
      "step": 5883
    },
    {
      "epoch": 0.48185568487751945,
      "grad_norm": 0.19780589640140533,
      "learning_rate": 1.6508574929634017e-05,
      "loss": 0.8462681770324707,
      "step": 5884
    },
    {
      "epoch": 0.4819375774140384,
      "grad_norm": 0.4819859564304352,
      "learning_rate": 1.6506256356482365e-05,
      "loss": 0.9043001532554626,
      "step": 5885
    },
    {
      "epoch": 0.4820194699505574,
      "grad_norm": 0.22115559875965118,
      "learning_rate": 1.6503937587470255e-05,
      "loss": 0.8394163846969604,
      "step": 5886
    },
    {
      "epoch": 0.48210136248707636,
      "grad_norm": 0.1958402693271637,
      "learning_rate": 1.6501618622698464e-05,
      "loss": 1.0369101762771606,
      "step": 5887
    },
    {
      "epoch": 0.48218325502359527,
      "grad_norm": 0.16870145499706268,
      "learning_rate": 1.6499299462267787e-05,
      "loss": 0.6988822221755981,
      "step": 5888
    },
    {
      "epoch": 0.48226514756011424,
      "grad_norm": 0.184292271733284,
      "learning_rate": 1.6496980106279026e-05,
      "loss": 0.884005606174469,
      "step": 5889
    },
    {
      "epoch": 0.4823470400966332,
      "grad_norm": 0.20188480615615845,
      "learning_rate": 1.6494660554832985e-05,
      "loss": 0.6285704374313354,
      "step": 5890
    },
    {
      "epoch": 0.4824289326331521,
      "grad_norm": 0.1911008656024933,
      "learning_rate": 1.6492340808030486e-05,
      "loss": 0.9541065692901611,
      "step": 5891
    },
    {
      "epoch": 0.4825108251696711,
      "grad_norm": 0.21020163595676422,
      "learning_rate": 1.6490020865972354e-05,
      "loss": 0.7494750022888184,
      "step": 5892
    },
    {
      "epoch": 0.48259271770619006,
      "grad_norm": 0.15996553003787994,
      "learning_rate": 1.6487700728759417e-05,
      "loss": 0.8565782308578491,
      "step": 5893
    },
    {
      "epoch": 0.482674610242709,
      "grad_norm": 0.197597473859787,
      "learning_rate": 1.6485380396492526e-05,
      "loss": 0.6839019060134888,
      "step": 5894
    },
    {
      "epoch": 0.48275650277922794,
      "grad_norm": 0.20415787398815155,
      "learning_rate": 1.6483059869272532e-05,
      "loss": 0.912388026714325,
      "step": 5895
    },
    {
      "epoch": 0.4828383953157469,
      "grad_norm": 0.19238954782485962,
      "learning_rate": 1.648073914720029e-05,
      "loss": 0.7359582185745239,
      "step": 5896
    },
    {
      "epoch": 0.4829202878522659,
      "grad_norm": 0.16258898377418518,
      "learning_rate": 1.647841823037667e-05,
      "loss": 0.9700924754142761,
      "step": 5897
    },
    {
      "epoch": 0.48300218038878484,
      "grad_norm": 0.2144785076379776,
      "learning_rate": 1.6476097118902546e-05,
      "loss": 1.0964922904968262,
      "step": 5898
    },
    {
      "epoch": 0.48308407292530375,
      "grad_norm": 0.2112424671649933,
      "learning_rate": 1.6473775812878806e-05,
      "loss": 0.6591051816940308,
      "step": 5899
    },
    {
      "epoch": 0.4831659654618227,
      "grad_norm": 0.20651821792125702,
      "learning_rate": 1.6471454312406348e-05,
      "loss": 0.708734929561615,
      "step": 5900
    },
    {
      "epoch": 0.4832478579983417,
      "grad_norm": 0.19638611376285553,
      "learning_rate": 1.6469132617586063e-05,
      "loss": 1.0208868980407715,
      "step": 5901
    },
    {
      "epoch": 0.4833297505348606,
      "grad_norm": 0.19778789579868317,
      "learning_rate": 1.646681072851887e-05,
      "loss": 0.6759640574455261,
      "step": 5902
    },
    {
      "epoch": 0.48341164307137957,
      "grad_norm": 0.1592661738395691,
      "learning_rate": 1.6464488645305683e-05,
      "loss": 0.716952383518219,
      "step": 5903
    },
    {
      "epoch": 0.48349353560789854,
      "grad_norm": 0.19048267602920532,
      "learning_rate": 1.6462166368047434e-05,
      "loss": 0.9284117221832275,
      "step": 5904
    },
    {
      "epoch": 0.4835754281444175,
      "grad_norm": 0.18087686598300934,
      "learning_rate": 1.645984389684506e-05,
      "loss": 1.0534253120422363,
      "step": 5905
    },
    {
      "epoch": 0.4836573206809364,
      "grad_norm": 0.19676411151885986,
      "learning_rate": 1.6457521231799494e-05,
      "loss": 0.9921011328697205,
      "step": 5906
    },
    {
      "epoch": 0.4837392132174554,
      "grad_norm": 0.17191162705421448,
      "learning_rate": 1.64551983730117e-05,
      "loss": 0.6204790472984314,
      "step": 5907
    },
    {
      "epoch": 0.48382110575397436,
      "grad_norm": 0.18426401913166046,
      "learning_rate": 1.6452875320582636e-05,
      "loss": 0.5094653964042664,
      "step": 5908
    },
    {
      "epoch": 0.4839029982904933,
      "grad_norm": 0.218369260430336,
      "learning_rate": 1.645055207461327e-05,
      "loss": 0.7832961082458496,
      "step": 5909
    },
    {
      "epoch": 0.48398489082701224,
      "grad_norm": 0.15537796914577484,
      "learning_rate": 1.6448228635204576e-05,
      "loss": 0.7167912721633911,
      "step": 5910
    },
    {
      "epoch": 0.4840667833635312,
      "grad_norm": 0.1970459520816803,
      "learning_rate": 1.644590500245755e-05,
      "loss": 0.8766236305236816,
      "step": 5911
    },
    {
      "epoch": 0.4841486759000502,
      "grad_norm": 0.19825111329555511,
      "learning_rate": 1.644358117647318e-05,
      "loss": 0.7676331996917725,
      "step": 5912
    },
    {
      "epoch": 0.4842305684365691,
      "grad_norm": 0.2296890765428543,
      "learning_rate": 1.6441257157352467e-05,
      "loss": 0.8500493764877319,
      "step": 5913
    },
    {
      "epoch": 0.48431246097308805,
      "grad_norm": 0.19898927211761475,
      "learning_rate": 1.643893294519643e-05,
      "loss": 0.8490474224090576,
      "step": 5914
    },
    {
      "epoch": 0.484394353509607,
      "grad_norm": 0.18297028541564941,
      "learning_rate": 1.6436608540106083e-05,
      "loss": 1.0119972229003906,
      "step": 5915
    },
    {
      "epoch": 0.484476246046126,
      "grad_norm": 0.184550940990448,
      "learning_rate": 1.6434283942182455e-05,
      "loss": 0.7259731292724609,
      "step": 5916
    },
    {
      "epoch": 0.4845581385826449,
      "grad_norm": 0.20221969485282898,
      "learning_rate": 1.6431959151526584e-05,
      "loss": 0.7339967489242554,
      "step": 5917
    },
    {
      "epoch": 0.48464003111916387,
      "grad_norm": 0.1757216602563858,
      "learning_rate": 1.6429634168239513e-05,
      "loss": 0.9319389462471008,
      "step": 5918
    },
    {
      "epoch": 0.48472192365568284,
      "grad_norm": 0.17842978239059448,
      "learning_rate": 1.64273089924223e-05,
      "loss": 0.7654030919075012,
      "step": 5919
    },
    {
      "epoch": 0.4848038161922018,
      "grad_norm": 0.1859051138162613,
      "learning_rate": 1.6424983624176e-05,
      "loss": 0.6247894763946533,
      "step": 5920
    },
    {
      "epoch": 0.4848857087287207,
      "grad_norm": 0.16781671345233917,
      "learning_rate": 1.6422658063601688e-05,
      "loss": 0.8171154260635376,
      "step": 5921
    },
    {
      "epoch": 0.4849676012652397,
      "grad_norm": 0.21774910390377045,
      "learning_rate": 1.6420332310800444e-05,
      "loss": 0.9938828349113464,
      "step": 5922
    },
    {
      "epoch": 0.48504949380175866,
      "grad_norm": 0.1769893914461136,
      "learning_rate": 1.6418006365873348e-05,
      "loss": 0.866245687007904,
      "step": 5923
    },
    {
      "epoch": 0.48513138633827757,
      "grad_norm": 0.19895203411579132,
      "learning_rate": 1.6415680228921497e-05,
      "loss": 0.7402900457382202,
      "step": 5924
    },
    {
      "epoch": 0.48521327887479654,
      "grad_norm": 0.14952559769153595,
      "learning_rate": 1.6413353900046e-05,
      "loss": 0.8994572758674622,
      "step": 5925
    },
    {
      "epoch": 0.4852951714113155,
      "grad_norm": 0.16222989559173584,
      "learning_rate": 1.6411027379347964e-05,
      "loss": 1.1332429647445679,
      "step": 5926
    },
    {
      "epoch": 0.4853770639478345,
      "grad_norm": 0.21819794178009033,
      "learning_rate": 1.6408700666928512e-05,
      "loss": 0.8343693614006042,
      "step": 5927
    },
    {
      "epoch": 0.4854589564843534,
      "grad_norm": 0.17140400409698486,
      "learning_rate": 1.640637376288877e-05,
      "loss": 0.7677515149116516,
      "step": 5928
    },
    {
      "epoch": 0.48554084902087236,
      "grad_norm": 0.2273983359336853,
      "learning_rate": 1.6404046667329876e-05,
      "loss": 1.0695781707763672,
      "step": 5929
    },
    {
      "epoch": 0.4856227415573913,
      "grad_norm": 0.18006975948810577,
      "learning_rate": 1.6401719380352973e-05,
      "loss": 1.282604694366455,
      "step": 5930
    },
    {
      "epoch": 0.4857046340939103,
      "grad_norm": 0.17954321205615997,
      "learning_rate": 1.639939190205922e-05,
      "loss": 1.0202034711837769,
      "step": 5931
    },
    {
      "epoch": 0.4857865266304292,
      "grad_norm": 0.15555952489376068,
      "learning_rate": 1.6397064232549773e-05,
      "loss": 0.7616036534309387,
      "step": 5932
    },
    {
      "epoch": 0.4858684191669482,
      "grad_norm": 0.16191786527633667,
      "learning_rate": 1.6394736371925803e-05,
      "loss": 0.546235978603363,
      "step": 5933
    },
    {
      "epoch": 0.48595031170346714,
      "grad_norm": 0.18401175737380981,
      "learning_rate": 1.6392408320288488e-05,
      "loss": 0.7593172788619995,
      "step": 5934
    },
    {
      "epoch": 0.48603220423998605,
      "grad_norm": 0.16298557817935944,
      "learning_rate": 1.639008007773902e-05,
      "loss": 1.0414432287216187,
      "step": 5935
    },
    {
      "epoch": 0.486114096776505,
      "grad_norm": 0.15804295241832733,
      "learning_rate": 1.638775164437859e-05,
      "loss": 0.6644487977027893,
      "step": 5936
    },
    {
      "epoch": 0.486195989313024,
      "grad_norm": 0.17623047530651093,
      "learning_rate": 1.63854230203084e-05,
      "loss": 0.7886113524436951,
      "step": 5937
    },
    {
      "epoch": 0.48627788184954296,
      "grad_norm": 0.17648865282535553,
      "learning_rate": 1.6383094205629665e-05,
      "loss": 0.6907986402511597,
      "step": 5938
    },
    {
      "epoch": 0.48635977438606187,
      "grad_norm": 0.2065678834915161,
      "learning_rate": 1.6380765200443603e-05,
      "loss": 0.9648787975311279,
      "step": 5939
    },
    {
      "epoch": 0.48644166692258084,
      "grad_norm": 0.19891603291034698,
      "learning_rate": 1.6378436004851446e-05,
      "loss": 0.8552531003952026,
      "step": 5940
    },
    {
      "epoch": 0.4865235594590998,
      "grad_norm": 0.1907181739807129,
      "learning_rate": 1.6376106618954422e-05,
      "loss": 0.825438916683197,
      "step": 5941
    },
    {
      "epoch": 0.4866054519956188,
      "grad_norm": 0.17743533849716187,
      "learning_rate": 1.637377704285378e-05,
      "loss": 0.9448757767677307,
      "step": 5942
    },
    {
      "epoch": 0.4866873445321377,
      "grad_norm": 0.21974827349185944,
      "learning_rate": 1.6371447276650775e-05,
      "loss": 0.7243618369102478,
      "step": 5943
    },
    {
      "epoch": 0.48676923706865666,
      "grad_norm": 0.18077319860458374,
      "learning_rate": 1.6369117320446667e-05,
      "loss": 1.0222158432006836,
      "step": 5944
    },
    {
      "epoch": 0.4868511296051756,
      "grad_norm": 0.2170586735010147,
      "learning_rate": 1.6366787174342723e-05,
      "loss": 0.8983923196792603,
      "step": 5945
    },
    {
      "epoch": 0.48693302214169454,
      "grad_norm": 0.16678480803966522,
      "learning_rate": 1.6364456838440227e-05,
      "loss": 0.7031627893447876,
      "step": 5946
    },
    {
      "epoch": 0.4870149146782135,
      "grad_norm": 0.17797419428825378,
      "learning_rate": 1.636212631284046e-05,
      "loss": 0.7951322793960571,
      "step": 5947
    },
    {
      "epoch": 0.4870968072147325,
      "grad_norm": 0.19243201613426208,
      "learning_rate": 1.6359795597644715e-05,
      "loss": 0.9492900967597961,
      "step": 5948
    },
    {
      "epoch": 0.48717869975125144,
      "grad_norm": 0.16273166239261627,
      "learning_rate": 1.63574646929543e-05,
      "loss": 0.8243293166160583,
      "step": 5949
    },
    {
      "epoch": 0.48726059228777036,
      "grad_norm": 0.17069761455059052,
      "learning_rate": 1.6355133598870514e-05,
      "loss": 0.7924271821975708,
      "step": 5950
    },
    {
      "epoch": 0.4873424848242893,
      "grad_norm": 0.17256031930446625,
      "learning_rate": 1.6352802315494693e-05,
      "loss": 0.8493505716323853,
      "step": 5951
    },
    {
      "epoch": 0.4874243773608083,
      "grad_norm": 0.1741335541009903,
      "learning_rate": 1.6350470842928154e-05,
      "loss": 0.755685567855835,
      "step": 5952
    },
    {
      "epoch": 0.4875062698973272,
      "grad_norm": 0.18250291049480438,
      "learning_rate": 1.634813918127223e-05,
      "loss": 0.6969490051269531,
      "step": 5953
    },
    {
      "epoch": 0.4875881624338462,
      "grad_norm": 0.21383148431777954,
      "learning_rate": 1.6345807330628274e-05,
      "loss": 0.8579533100128174,
      "step": 5954
    },
    {
      "epoch": 0.48767005497036514,
      "grad_norm": 0.17530924081802368,
      "learning_rate": 1.6343475291097626e-05,
      "loss": 0.6964641213417053,
      "step": 5955
    },
    {
      "epoch": 0.4877519475068841,
      "grad_norm": 0.16725848615169525,
      "learning_rate": 1.6341143062781658e-05,
      "loss": 0.7544265389442444,
      "step": 5956
    },
    {
      "epoch": 0.487833840043403,
      "grad_norm": 0.19130340218544006,
      "learning_rate": 1.6338810645781728e-05,
      "loss": 0.9070687294006348,
      "step": 5957
    },
    {
      "epoch": 0.487915732579922,
      "grad_norm": 0.1846885085105896,
      "learning_rate": 1.633647804019922e-05,
      "loss": 0.8739230036735535,
      "step": 5958
    },
    {
      "epoch": 0.48799762511644096,
      "grad_norm": 0.2216329425573349,
      "learning_rate": 1.6334145246135516e-05,
      "loss": 1.321541428565979,
      "step": 5959
    },
    {
      "epoch": 0.4880795176529599,
      "grad_norm": 0.16341957449913025,
      "learning_rate": 1.633181226369201e-05,
      "loss": 0.6577097773551941,
      "step": 5960
    },
    {
      "epoch": 0.48816141018947884,
      "grad_norm": 0.19821658730506897,
      "learning_rate": 1.63294790929701e-05,
      "loss": 0.713863730430603,
      "step": 5961
    },
    {
      "epoch": 0.4882433027259978,
      "grad_norm": 0.18818292021751404,
      "learning_rate": 1.632714573407119e-05,
      "loss": 0.9049433469772339,
      "step": 5962
    },
    {
      "epoch": 0.4883251952625168,
      "grad_norm": 0.18383146822452545,
      "learning_rate": 1.6324812187096716e-05,
      "loss": 0.6146726012229919,
      "step": 5963
    },
    {
      "epoch": 0.4884070877990357,
      "grad_norm": 0.1608864665031433,
      "learning_rate": 1.6322478452148083e-05,
      "loss": 0.6927241683006287,
      "step": 5964
    },
    {
      "epoch": 0.48848898033555466,
      "grad_norm": 0.20579831302165985,
      "learning_rate": 1.632014452932674e-05,
      "loss": 0.6864463090896606,
      "step": 5965
    },
    {
      "epoch": 0.4885708728720736,
      "grad_norm": 0.2145933210849762,
      "learning_rate": 1.6317810418734118e-05,
      "loss": 1.2520948648452759,
      "step": 5966
    },
    {
      "epoch": 0.4886527654085926,
      "grad_norm": 0.20621420443058014,
      "learning_rate": 1.631547612047167e-05,
      "loss": 0.7455288171768188,
      "step": 5967
    },
    {
      "epoch": 0.4887346579451115,
      "grad_norm": 0.21172888576984406,
      "learning_rate": 1.6313141634640862e-05,
      "loss": 0.8102445602416992,
      "step": 5968
    },
    {
      "epoch": 0.4888165504816305,
      "grad_norm": 0.17395006120204926,
      "learning_rate": 1.631080696134315e-05,
      "loss": 0.6828479170799255,
      "step": 5969
    },
    {
      "epoch": 0.48889844301814944,
      "grad_norm": 0.20230428874492645,
      "learning_rate": 1.630847210068001e-05,
      "loss": 0.5908797979354858,
      "step": 5970
    },
    {
      "epoch": 0.4889803355546684,
      "grad_norm": 0.1880623996257782,
      "learning_rate": 1.630613705275293e-05,
      "loss": 0.6746910214424133,
      "step": 5971
    },
    {
      "epoch": 0.4890622280911873,
      "grad_norm": 0.14637719094753265,
      "learning_rate": 1.6303801817663402e-05,
      "loss": 0.5278902649879456,
      "step": 5972
    },
    {
      "epoch": 0.4891441206277063,
      "grad_norm": 0.18227972090244293,
      "learning_rate": 1.630146639551291e-05,
      "loss": 0.8327375054359436,
      "step": 5973
    },
    {
      "epoch": 0.48922601316422526,
      "grad_norm": 0.15522220730781555,
      "learning_rate": 1.6299130786402983e-05,
      "loss": 0.7231636047363281,
      "step": 5974
    },
    {
      "epoch": 0.4893079057007442,
      "grad_norm": 0.1990952342748642,
      "learning_rate": 1.629679499043512e-05,
      "loss": 1.0277854204177856,
      "step": 5975
    },
    {
      "epoch": 0.48938979823726314,
      "grad_norm": 0.17304225265979767,
      "learning_rate": 1.6294459007710852e-05,
      "loss": 0.7476375102996826,
      "step": 5976
    },
    {
      "epoch": 0.4894716907737821,
      "grad_norm": 0.18287530541419983,
      "learning_rate": 1.6292122838331706e-05,
      "loss": 0.6991099119186401,
      "step": 5977
    },
    {
      "epoch": 0.4895535833103011,
      "grad_norm": 0.18304510414600372,
      "learning_rate": 1.6289786482399222e-05,
      "loss": 0.9188651442527771,
      "step": 5978
    },
    {
      "epoch": 0.48963547584682,
      "grad_norm": 0.18915781378746033,
      "learning_rate": 1.6287449940014952e-05,
      "loss": 0.8288341164588928,
      "step": 5979
    },
    {
      "epoch": 0.48971736838333896,
      "grad_norm": 0.188652902841568,
      "learning_rate": 1.628511321128045e-05,
      "loss": 0.4175039529800415,
      "step": 5980
    },
    {
      "epoch": 0.4897992609198579,
      "grad_norm": 0.18770867586135864,
      "learning_rate": 1.6282776296297275e-05,
      "loss": 0.7712852954864502,
      "step": 5981
    },
    {
      "epoch": 0.4898811534563769,
      "grad_norm": 0.20129351317882538,
      "learning_rate": 1.6280439195167007e-05,
      "loss": 0.9338968992233276,
      "step": 5982
    },
    {
      "epoch": 0.4899630459928958,
      "grad_norm": 0.22080492973327637,
      "learning_rate": 1.627810190799122e-05,
      "loss": 0.8265432119369507,
      "step": 5983
    },
    {
      "epoch": 0.4900449385294148,
      "grad_norm": 0.20745952427387238,
      "learning_rate": 1.6275764434871498e-05,
      "loss": 0.830575168132782,
      "step": 5984
    },
    {
      "epoch": 0.49012683106593374,
      "grad_norm": 0.18632270395755768,
      "learning_rate": 1.627342677590945e-05,
      "loss": 1.0338287353515625,
      "step": 5985
    },
    {
      "epoch": 0.49020872360245266,
      "grad_norm": 0.22307045757770538,
      "learning_rate": 1.6271088931206675e-05,
      "loss": 0.8989386558532715,
      "step": 5986
    },
    {
      "epoch": 0.4902906161389716,
      "grad_norm": 0.1769031137228012,
      "learning_rate": 1.6268750900864777e-05,
      "loss": 0.8107023239135742,
      "step": 5987
    },
    {
      "epoch": 0.4903725086754906,
      "grad_norm": 0.192116379737854,
      "learning_rate": 1.6266412684985387e-05,
      "loss": 0.6837422251701355,
      "step": 5988
    },
    {
      "epoch": 0.49045440121200956,
      "grad_norm": 0.17722377181053162,
      "learning_rate": 1.626407428367013e-05,
      "loss": 1.1288189888000488,
      "step": 5989
    },
    {
      "epoch": 0.4905362937485285,
      "grad_norm": 0.26542580127716064,
      "learning_rate": 1.626173569702064e-05,
      "loss": 0.7757213711738586,
      "step": 5990
    },
    {
      "epoch": 0.49061818628504744,
      "grad_norm": 0.20512759685516357,
      "learning_rate": 1.6259396925138568e-05,
      "loss": 0.848060131072998,
      "step": 5991
    },
    {
      "epoch": 0.4907000788215664,
      "grad_norm": 0.20065143704414368,
      "learning_rate": 1.6257057968125556e-05,
      "loss": 0.8630199432373047,
      "step": 5992
    },
    {
      "epoch": 0.4907819713580854,
      "grad_norm": 0.19146764278411865,
      "learning_rate": 1.6254718826083276e-05,
      "loss": 0.7795783877372742,
      "step": 5993
    },
    {
      "epoch": 0.4908638638946043,
      "grad_norm": 0.15905913710594177,
      "learning_rate": 1.625237949911339e-05,
      "loss": 0.5010542273521423,
      "step": 5994
    },
    {
      "epoch": 0.49094575643112326,
      "grad_norm": 0.18451066315174103,
      "learning_rate": 1.6250039987317577e-05,
      "loss": 0.7587258219718933,
      "step": 5995
    },
    {
      "epoch": 0.4910276489676422,
      "grad_norm": 0.1795945018529892,
      "learning_rate": 1.6247700290797524e-05,
      "loss": 0.681980311870575,
      "step": 5996
    },
    {
      "epoch": 0.49110954150416114,
      "grad_norm": 0.23591294884681702,
      "learning_rate": 1.6245360409654918e-05,
      "loss": 1.0770446062088013,
      "step": 5997
    },
    {
      "epoch": 0.4911914340406801,
      "grad_norm": 0.16469796001911163,
      "learning_rate": 1.6243020343991466e-05,
      "loss": 1.0984690189361572,
      "step": 5998
    },
    {
      "epoch": 0.4912733265771991,
      "grad_norm": 0.16598966717720032,
      "learning_rate": 1.6240680093908873e-05,
      "loss": 0.7699282169342041,
      "step": 5999
    },
    {
      "epoch": 0.49135521911371804,
      "grad_norm": 0.1616978943347931,
      "learning_rate": 1.6238339659508855e-05,
      "loss": 0.7049899101257324,
      "step": 6000
    },
    {
      "epoch": 0.49143711165023696,
      "grad_norm": 0.15934009850025177,
      "learning_rate": 1.623599904089314e-05,
      "loss": 1.1693589687347412,
      "step": 6001
    },
    {
      "epoch": 0.4915190041867559,
      "grad_norm": 0.1892499327659607,
      "learning_rate": 1.6233658238163464e-05,
      "loss": 0.9363872408866882,
      "step": 6002
    },
    {
      "epoch": 0.4916008967232749,
      "grad_norm": 0.19122911989688873,
      "learning_rate": 1.6231317251421556e-05,
      "loss": 1.0013247728347778,
      "step": 6003
    },
    {
      "epoch": 0.49168278925979386,
      "grad_norm": 0.19167572259902954,
      "learning_rate": 1.622897608076918e-05,
      "loss": 0.7464023232460022,
      "step": 6004
    },
    {
      "epoch": 0.4917646817963128,
      "grad_norm": 0.18945056200027466,
      "learning_rate": 1.6226634726308083e-05,
      "loss": 0.9319173097610474,
      "step": 6005
    },
    {
      "epoch": 0.49184657433283174,
      "grad_norm": 0.2046397477388382,
      "learning_rate": 1.622429318814003e-05,
      "loss": 0.6826932430267334,
      "step": 6006
    },
    {
      "epoch": 0.4919284668693507,
      "grad_norm": 0.16344715654850006,
      "learning_rate": 1.6221951466366797e-05,
      "loss": 0.6544229388237,
      "step": 6007
    },
    {
      "epoch": 0.4920103594058696,
      "grad_norm": 0.16131818294525146,
      "learning_rate": 1.621960956109017e-05,
      "loss": 0.7040161490440369,
      "step": 6008
    },
    {
      "epoch": 0.4920922519423886,
      "grad_norm": 0.18988215923309326,
      "learning_rate": 1.6217267472411926e-05,
      "loss": 0.7467964291572571,
      "step": 6009
    },
    {
      "epoch": 0.49217414447890756,
      "grad_norm": 0.1994233876466751,
      "learning_rate": 1.6214925200433872e-05,
      "loss": 0.6349900960922241,
      "step": 6010
    },
    {
      "epoch": 0.49225603701542653,
      "grad_norm": 0.21836267411708832,
      "learning_rate": 1.6212582745257807e-05,
      "loss": 0.8732305765151978,
      "step": 6011
    },
    {
      "epoch": 0.49233792955194544,
      "grad_norm": 0.15567655861377716,
      "learning_rate": 1.6210240106985546e-05,
      "loss": 0.5584937930107117,
      "step": 6012
    },
    {
      "epoch": 0.4924198220884644,
      "grad_norm": 0.17866085469722748,
      "learning_rate": 1.6207897285718907e-05,
      "loss": 0.8120579719543457,
      "step": 6013
    },
    {
      "epoch": 0.4925017146249834,
      "grad_norm": 0.2261458784341812,
      "learning_rate": 1.6205554281559725e-05,
      "loss": 1.1751279830932617,
      "step": 6014
    },
    {
      "epoch": 0.49258360716150235,
      "grad_norm": 0.1887982189655304,
      "learning_rate": 1.620321109460983e-05,
      "loss": 1.096996545791626,
      "step": 6015
    },
    {
      "epoch": 0.49266549969802126,
      "grad_norm": 0.23590917885303497,
      "learning_rate": 1.6200867724971067e-05,
      "loss": 0.8173369765281677,
      "step": 6016
    },
    {
      "epoch": 0.4927473922345402,
      "grad_norm": 0.17784301936626434,
      "learning_rate": 1.619852417274529e-05,
      "loss": 0.6741865873336792,
      "step": 6017
    },
    {
      "epoch": 0.4928292847710592,
      "grad_norm": 0.22401024401187897,
      "learning_rate": 1.6196180438034365e-05,
      "loss": 0.9108845591545105,
      "step": 6018
    },
    {
      "epoch": 0.4929111773075781,
      "grad_norm": 0.18459883332252502,
      "learning_rate": 1.6193836520940156e-05,
      "loss": 0.8867147564888,
      "step": 6019
    },
    {
      "epoch": 0.4929930698440971,
      "grad_norm": 0.1768030822277069,
      "learning_rate": 1.6191492421564538e-05,
      "loss": 0.7810871005058289,
      "step": 6020
    },
    {
      "epoch": 0.49307496238061604,
      "grad_norm": 0.1696414351463318,
      "learning_rate": 1.618914814000939e-05,
      "loss": 1.021082878112793,
      "step": 6021
    },
    {
      "epoch": 0.493156854917135,
      "grad_norm": 0.1930042803287506,
      "learning_rate": 1.6186803676376615e-05,
      "loss": 1.0277472734451294,
      "step": 6022
    },
    {
      "epoch": 0.4932387474536539,
      "grad_norm": 0.2281750589609146,
      "learning_rate": 1.6184459030768108e-05,
      "loss": 1.2067933082580566,
      "step": 6023
    },
    {
      "epoch": 0.4933206399901729,
      "grad_norm": 0.20835436880588531,
      "learning_rate": 1.6182114203285773e-05,
      "loss": 0.739664614200592,
      "step": 6024
    },
    {
      "epoch": 0.49340253252669186,
      "grad_norm": 0.18414336442947388,
      "learning_rate": 1.6179769194031534e-05,
      "loss": 0.885077953338623,
      "step": 6025
    },
    {
      "epoch": 0.4934844250632108,
      "grad_norm": 0.17989236116409302,
      "learning_rate": 1.617742400310731e-05,
      "loss": 0.983352780342102,
      "step": 6026
    },
    {
      "epoch": 0.49356631759972974,
      "grad_norm": 0.20264087617397308,
      "learning_rate": 1.617507863061503e-05,
      "loss": 0.8508093357086182,
      "step": 6027
    },
    {
      "epoch": 0.4936482101362487,
      "grad_norm": 0.19364526867866516,
      "learning_rate": 1.6172733076656643e-05,
      "loss": 0.9638767838478088,
      "step": 6028
    },
    {
      "epoch": 0.4937301026727677,
      "grad_norm": 0.22690469026565552,
      "learning_rate": 1.6170387341334086e-05,
      "loss": 0.705052375793457,
      "step": 6029
    },
    {
      "epoch": 0.4938119952092866,
      "grad_norm": 0.37671831250190735,
      "learning_rate": 1.6168041424749317e-05,
      "loss": 0.7582945823669434,
      "step": 6030
    },
    {
      "epoch": 0.49389388774580556,
      "grad_norm": 0.2004888653755188,
      "learning_rate": 1.6165695327004306e-05,
      "loss": 0.8920019268989563,
      "step": 6031
    },
    {
      "epoch": 0.49397578028232453,
      "grad_norm": 0.18018996715545654,
      "learning_rate": 1.6163349048201012e-05,
      "loss": 0.7740724086761475,
      "step": 6032
    },
    {
      "epoch": 0.4940576728188435,
      "grad_norm": 0.19254964590072632,
      "learning_rate": 1.6161002588441424e-05,
      "loss": 0.8580701351165771,
      "step": 6033
    },
    {
      "epoch": 0.4941395653553624,
      "grad_norm": 0.1991758644580841,
      "learning_rate": 1.615865594782753e-05,
      "loss": 0.9248565435409546,
      "step": 6034
    },
    {
      "epoch": 0.4942214578918814,
      "grad_norm": 0.18530799448490143,
      "learning_rate": 1.6156309126461315e-05,
      "loss": 0.6842769980430603,
      "step": 6035
    },
    {
      "epoch": 0.49430335042840035,
      "grad_norm": 0.15058189630508423,
      "learning_rate": 1.615396212444479e-05,
      "loss": 0.9477306604385376,
      "step": 6036
    },
    {
      "epoch": 0.49438524296491926,
      "grad_norm": 0.17723846435546875,
      "learning_rate": 1.615161494187996e-05,
      "loss": 0.8458277583122253,
      "step": 6037
    },
    {
      "epoch": 0.4944671355014382,
      "grad_norm": 0.206573486328125,
      "learning_rate": 1.6149267578868843e-05,
      "loss": 0.9696130752563477,
      "step": 6038
    },
    {
      "epoch": 0.4945490280379572,
      "grad_norm": 0.16157765686511993,
      "learning_rate": 1.6146920035513473e-05,
      "loss": 0.7796776294708252,
      "step": 6039
    },
    {
      "epoch": 0.49463092057447616,
      "grad_norm": 0.18003542721271515,
      "learning_rate": 1.6144572311915876e-05,
      "loss": 0.8854595422744751,
      "step": 6040
    },
    {
      "epoch": 0.4947128131109951,
      "grad_norm": 0.17743580043315887,
      "learning_rate": 1.6142224408178094e-05,
      "loss": 0.7067782878875732,
      "step": 6041
    },
    {
      "epoch": 0.49479470564751404,
      "grad_norm": 0.17366071045398712,
      "learning_rate": 1.6139876324402183e-05,
      "loss": 0.9901339411735535,
      "step": 6042
    },
    {
      "epoch": 0.494876598184033,
      "grad_norm": 0.18295088410377502,
      "learning_rate": 1.613752806069019e-05,
      "loss": 0.8676550388336182,
      "step": 6043
    },
    {
      "epoch": 0.494958490720552,
      "grad_norm": 0.1947259157896042,
      "learning_rate": 1.613517961714419e-05,
      "loss": 0.6863188147544861,
      "step": 6044
    },
    {
      "epoch": 0.4950403832570709,
      "grad_norm": 0.20279835164546967,
      "learning_rate": 1.6132830993866256e-05,
      "loss": 0.9753381013870239,
      "step": 6045
    },
    {
      "epoch": 0.49512227579358986,
      "grad_norm": 0.19291162490844727,
      "learning_rate": 1.613048219095846e-05,
      "loss": 0.9472445249557495,
      "step": 6046
    },
    {
      "epoch": 0.49520416833010883,
      "grad_norm": 0.20757408440113068,
      "learning_rate": 1.6128133208522892e-05,
      "loss": 0.9282315969467163,
      "step": 6047
    },
    {
      "epoch": 0.49528606086662774,
      "grad_norm": 0.18038591742515564,
      "learning_rate": 1.6125784046661663e-05,
      "loss": 0.5307304859161377,
      "step": 6048
    },
    {
      "epoch": 0.4953679534031467,
      "grad_norm": 0.1814199984073639,
      "learning_rate": 1.6123434705476857e-05,
      "loss": 0.6789615750312805,
      "step": 6049
    },
    {
      "epoch": 0.4954498459396657,
      "grad_norm": 0.17542913556098938,
      "learning_rate": 1.61210851850706e-05,
      "loss": 0.7129001617431641,
      "step": 6050
    },
    {
      "epoch": 0.49553173847618465,
      "grad_norm": 0.25318071246147156,
      "learning_rate": 1.611873548554501e-05,
      "loss": 0.9080148935317993,
      "step": 6051
    },
    {
      "epoch": 0.49561363101270356,
      "grad_norm": 0.1979486644268036,
      "learning_rate": 1.6116385607002204e-05,
      "loss": 1.0141981840133667,
      "step": 6052
    },
    {
      "epoch": 0.4956955235492225,
      "grad_norm": 0.19000868499279022,
      "learning_rate": 1.611403554954433e-05,
      "loss": 0.8312104344367981,
      "step": 6053
    },
    {
      "epoch": 0.4957774160857415,
      "grad_norm": 0.16280408203601837,
      "learning_rate": 1.611168531327353e-05,
      "loss": 0.7266126871109009,
      "step": 6054
    },
    {
      "epoch": 0.49585930862226046,
      "grad_norm": 0.2032080590724945,
      "learning_rate": 1.6109334898291944e-05,
      "loss": 1.0277180671691895,
      "step": 6055
    },
    {
      "epoch": 0.4959412011587794,
      "grad_norm": 0.22894930839538574,
      "learning_rate": 1.6106984304701746e-05,
      "loss": 0.9807442426681519,
      "step": 6056
    },
    {
      "epoch": 0.49602309369529834,
      "grad_norm": 0.17532365024089813,
      "learning_rate": 1.6104633532605086e-05,
      "loss": 0.845416784286499,
      "step": 6057
    },
    {
      "epoch": 0.4961049862318173,
      "grad_norm": 0.18398912250995636,
      "learning_rate": 1.6102282582104156e-05,
      "loss": 0.9250243902206421,
      "step": 6058
    },
    {
      "epoch": 0.4961868787683362,
      "grad_norm": 0.1566864401102066,
      "learning_rate": 1.6099931453301123e-05,
      "loss": 0.9628044962882996,
      "step": 6059
    },
    {
      "epoch": 0.4962687713048552,
      "grad_norm": 0.17723681032657623,
      "learning_rate": 1.6097580146298184e-05,
      "loss": 0.9886734485626221,
      "step": 6060
    },
    {
      "epoch": 0.49635066384137416,
      "grad_norm": 0.20652930438518524,
      "learning_rate": 1.6095228661197535e-05,
      "loss": 0.7692238092422485,
      "step": 6061
    },
    {
      "epoch": 0.49643255637789313,
      "grad_norm": 0.18562577664852142,
      "learning_rate": 1.6092876998101383e-05,
      "loss": 0.7086588144302368,
      "step": 6062
    },
    {
      "epoch": 0.49651444891441204,
      "grad_norm": 0.17545676231384277,
      "learning_rate": 1.6090525157111936e-05,
      "loss": 0.729061484336853,
      "step": 6063
    },
    {
      "epoch": 0.496596341450931,
      "grad_norm": 0.18465320765972137,
      "learning_rate": 1.6088173138331423e-05,
      "loss": 0.8835002183914185,
      "step": 6064
    },
    {
      "epoch": 0.49667823398745,
      "grad_norm": 0.19601015746593475,
      "learning_rate": 1.6085820941862063e-05,
      "loss": 0.9176339507102966,
      "step": 6065
    },
    {
      "epoch": 0.49676012652396895,
      "grad_norm": 0.148952454328537,
      "learning_rate": 1.60834685678061e-05,
      "loss": 0.7268451452255249,
      "step": 6066
    },
    {
      "epoch": 0.49684201906048786,
      "grad_norm": 0.15964043140411377,
      "learning_rate": 1.608111601626577e-05,
      "loss": 0.730918824672699,
      "step": 6067
    },
    {
      "epoch": 0.49692391159700683,
      "grad_norm": 0.14792898297309875,
      "learning_rate": 1.6078763287343332e-05,
      "loss": 0.8943265080451965,
      "step": 6068
    },
    {
      "epoch": 0.4970058041335258,
      "grad_norm": 0.18148761987686157,
      "learning_rate": 1.6076410381141043e-05,
      "loss": 0.8143479228019714,
      "step": 6069
    },
    {
      "epoch": 0.4970876966700447,
      "grad_norm": 0.15110190212726593,
      "learning_rate": 1.6074057297761166e-05,
      "loss": 0.6135390996932983,
      "step": 6070
    },
    {
      "epoch": 0.4971695892065637,
      "grad_norm": 0.18802663683891296,
      "learning_rate": 1.6071704037305982e-05,
      "loss": 0.5304161906242371,
      "step": 6071
    },
    {
      "epoch": 0.49725148174308265,
      "grad_norm": 0.21430660784244537,
      "learning_rate": 1.606935059987777e-05,
      "loss": 1.1035454273223877,
      "step": 6072
    },
    {
      "epoch": 0.4973333742796016,
      "grad_norm": 0.1921183168888092,
      "learning_rate": 1.6066996985578823e-05,
      "loss": 0.8215813040733337,
      "step": 6073
    },
    {
      "epoch": 0.4974152668161205,
      "grad_norm": 0.20075133442878723,
      "learning_rate": 1.6064643194511433e-05,
      "loss": 0.8783615231513977,
      "step": 6074
    },
    {
      "epoch": 0.4974971593526395,
      "grad_norm": 0.18515415489673615,
      "learning_rate": 1.6062289226777907e-05,
      "loss": 0.6204794049263,
      "step": 6075
    },
    {
      "epoch": 0.49757905188915846,
      "grad_norm": 0.16364100575447083,
      "learning_rate": 1.6059935082480563e-05,
      "loss": 1.0888397693634033,
      "step": 6076
    },
    {
      "epoch": 0.49766094442567743,
      "grad_norm": 0.17873060703277588,
      "learning_rate": 1.605758076172172e-05,
      "loss": 0.8292990922927856,
      "step": 6077
    },
    {
      "epoch": 0.49774283696219634,
      "grad_norm": 0.17903932929039001,
      "learning_rate": 1.60552262646037e-05,
      "loss": 1.021253228187561,
      "step": 6078
    },
    {
      "epoch": 0.4978247294987153,
      "grad_norm": 0.19135256111621857,
      "learning_rate": 1.605287159122885e-05,
      "loss": 0.8693904876708984,
      "step": 6079
    },
    {
      "epoch": 0.4979066220352343,
      "grad_norm": 0.16468168795108795,
      "learning_rate": 1.6050516741699503e-05,
      "loss": 0.8124088048934937,
      "step": 6080
    },
    {
      "epoch": 0.4979885145717532,
      "grad_norm": 0.18407782912254333,
      "learning_rate": 1.604816171611802e-05,
      "loss": 0.6734189987182617,
      "step": 6081
    },
    {
      "epoch": 0.49807040710827216,
      "grad_norm": 0.16637873649597168,
      "learning_rate": 1.604580651458675e-05,
      "loss": 0.5516297221183777,
      "step": 6082
    },
    {
      "epoch": 0.49815229964479113,
      "grad_norm": 0.20353186130523682,
      "learning_rate": 1.6043451137208074e-05,
      "loss": 0.8016135096549988,
      "step": 6083
    },
    {
      "epoch": 0.4982341921813101,
      "grad_norm": 0.1805795580148697,
      "learning_rate": 1.604109558408435e-05,
      "loss": 0.6310434937477112,
      "step": 6084
    },
    {
      "epoch": 0.498316084717829,
      "grad_norm": 0.19343794882297516,
      "learning_rate": 1.6038739855317972e-05,
      "loss": 1.0442055463790894,
      "step": 6085
    },
    {
      "epoch": 0.498397977254348,
      "grad_norm": 0.16567005217075348,
      "learning_rate": 1.6036383951011325e-05,
      "loss": 0.7219021916389465,
      "step": 6086
    },
    {
      "epoch": 0.49847986979086695,
      "grad_norm": 0.1630488485097885,
      "learning_rate": 1.6034027871266804e-05,
      "loss": 0.5867639183998108,
      "step": 6087
    },
    {
      "epoch": 0.4985617623273859,
      "grad_norm": 0.1810281127691269,
      "learning_rate": 1.6031671616186818e-05,
      "loss": 0.7500725388526917,
      "step": 6088
    },
    {
      "epoch": 0.49864365486390483,
      "grad_norm": 0.17640671133995056,
      "learning_rate": 1.602931518587378e-05,
      "loss": 0.7010160684585571,
      "step": 6089
    },
    {
      "epoch": 0.4987255474004238,
      "grad_norm": 0.17068596184253693,
      "learning_rate": 1.602695858043011e-05,
      "loss": 0.7685654759407043,
      "step": 6090
    },
    {
      "epoch": 0.49880743993694276,
      "grad_norm": 0.1880425065755844,
      "learning_rate": 1.6024601799958232e-05,
      "loss": 0.7995885014533997,
      "step": 6091
    },
    {
      "epoch": 0.4988893324734617,
      "grad_norm": 0.15210434794425964,
      "learning_rate": 1.602224484456058e-05,
      "loss": 0.7728027105331421,
      "step": 6092
    },
    {
      "epoch": 0.49897122500998065,
      "grad_norm": 0.1873197704553604,
      "learning_rate": 1.6019887714339605e-05,
      "loss": 1.0509437322616577,
      "step": 6093
    },
    {
      "epoch": 0.4990531175464996,
      "grad_norm": 0.16454248130321503,
      "learning_rate": 1.6017530409397758e-05,
      "loss": 0.784302830696106,
      "step": 6094
    },
    {
      "epoch": 0.4991350100830186,
      "grad_norm": 0.19081909954547882,
      "learning_rate": 1.6015172929837484e-05,
      "loss": 0.8579769730567932,
      "step": 6095
    },
    {
      "epoch": 0.4992169026195375,
      "grad_norm": 0.18529193103313446,
      "learning_rate": 1.601281527576126e-05,
      "loss": 0.7277228832244873,
      "step": 6096
    },
    {
      "epoch": 0.49929879515605646,
      "grad_norm": 0.19794096052646637,
      "learning_rate": 1.601045744727156e-05,
      "loss": 0.6129662990570068,
      "step": 6097
    },
    {
      "epoch": 0.49938068769257543,
      "grad_norm": 0.18975374102592468,
      "learning_rate": 1.6008099444470853e-05,
      "loss": 1.054952621459961,
      "step": 6098
    },
    {
      "epoch": 0.49946258022909434,
      "grad_norm": 0.28653237223625183,
      "learning_rate": 1.600574126746164e-05,
      "loss": 0.8132591843605042,
      "step": 6099
    },
    {
      "epoch": 0.4995444727656133,
      "grad_norm": 0.16355295479297638,
      "learning_rate": 1.6003382916346416e-05,
      "loss": 0.8547933101654053,
      "step": 6100
    },
    {
      "epoch": 0.4996263653021323,
      "grad_norm": 0.21802574396133423,
      "learning_rate": 1.600102439122768e-05,
      "loss": 0.9124488234519958,
      "step": 6101
    },
    {
      "epoch": 0.49970825783865125,
      "grad_norm": 0.203663632273674,
      "learning_rate": 1.5998665692207946e-05,
      "loss": 0.9423521161079407,
      "step": 6102
    },
    {
      "epoch": 0.49979015037517016,
      "grad_norm": 0.18875637650489807,
      "learning_rate": 1.5996306819389733e-05,
      "loss": 0.7195621132850647,
      "step": 6103
    },
    {
      "epoch": 0.49987204291168913,
      "grad_norm": 0.183464914560318,
      "learning_rate": 1.599394777287556e-05,
      "loss": 0.8054258227348328,
      "step": 6104
    },
    {
      "epoch": 0.4999539354482081,
      "grad_norm": 0.22302785515785217,
      "learning_rate": 1.5991588552767974e-05,
      "loss": 1.0078012943267822,
      "step": 6105
    },
    {
      "epoch": 0.500035827984727,
      "grad_norm": 0.18861261010169983,
      "learning_rate": 1.59892291591695e-05,
      "loss": 1.0852680206298828,
      "step": 6106
    },
    {
      "epoch": 0.500117720521246,
      "grad_norm": 0.1930169314146042,
      "learning_rate": 1.5986869592182705e-05,
      "loss": 0.7914626002311707,
      "step": 6107
    },
    {
      "epoch": 0.500199613057765,
      "grad_norm": 0.203404039144516,
      "learning_rate": 1.5984509851910136e-05,
      "loss": 0.9395484328269958,
      "step": 6108
    },
    {
      "epoch": 0.5002815055942839,
      "grad_norm": 0.19660070538520813,
      "learning_rate": 1.5982149938454353e-05,
      "loss": 0.9900261163711548,
      "step": 6109
    },
    {
      "epoch": 0.5003633981308029,
      "grad_norm": 0.23250092566013336,
      "learning_rate": 1.5979789851917938e-05,
      "loss": 1.376842737197876,
      "step": 6110
    },
    {
      "epoch": 0.5004452906673218,
      "grad_norm": 0.1906135380268097,
      "learning_rate": 1.5977429592403462e-05,
      "loss": 0.8129264116287231,
      "step": 6111
    },
    {
      "epoch": 0.5005271832038407,
      "grad_norm": 0.1819225698709488,
      "learning_rate": 1.5975069160013513e-05,
      "loss": 0.8175798058509827,
      "step": 6112
    },
    {
      "epoch": 0.5006090757403597,
      "grad_norm": 0.17959755659103394,
      "learning_rate": 1.597270855485069e-05,
      "loss": 0.7669265866279602,
      "step": 6113
    },
    {
      "epoch": 0.5006909682768786,
      "grad_norm": 0.15156373381614685,
      "learning_rate": 1.5970347777017587e-05,
      "loss": 0.8157228231430054,
      "step": 6114
    },
    {
      "epoch": 0.5007728608133977,
      "grad_norm": 0.17712171375751495,
      "learning_rate": 1.5967986826616812e-05,
      "loss": 0.9094599485397339,
      "step": 6115
    },
    {
      "epoch": 0.5008547533499166,
      "grad_norm": 0.1868724524974823,
      "learning_rate": 1.5965625703751e-05,
      "loss": 0.9826311469078064,
      "step": 6116
    },
    {
      "epoch": 0.5009366458864355,
      "grad_norm": 0.1626458317041397,
      "learning_rate": 1.5963264408522752e-05,
      "loss": 0.6231788992881775,
      "step": 6117
    },
    {
      "epoch": 0.5010185384229545,
      "grad_norm": 0.16001787781715393,
      "learning_rate": 1.596090294103471e-05,
      "loss": 0.5353693962097168,
      "step": 6118
    },
    {
      "epoch": 0.5011004309594734,
      "grad_norm": 0.1653135120868683,
      "learning_rate": 1.5958541301389514e-05,
      "loss": 0.7741868495941162,
      "step": 6119
    },
    {
      "epoch": 0.5011823234959923,
      "grad_norm": 0.18304568529129028,
      "learning_rate": 1.595617948968981e-05,
      "loss": 0.8161888718605042,
      "step": 6120
    },
    {
      "epoch": 0.5012642160325114,
      "grad_norm": 0.20770657062530518,
      "learning_rate": 1.5953817506038248e-05,
      "loss": 1.0122888088226318,
      "step": 6121
    },
    {
      "epoch": 0.5013461085690303,
      "grad_norm": 0.16320550441741943,
      "learning_rate": 1.5951455350537495e-05,
      "loss": 0.6598148941993713,
      "step": 6122
    },
    {
      "epoch": 0.5014280011055492,
      "grad_norm": 0.15605279803276062,
      "learning_rate": 1.5949093023290215e-05,
      "loss": 0.6082609295845032,
      "step": 6123
    },
    {
      "epoch": 0.5015098936420682,
      "grad_norm": 0.14467981457710266,
      "learning_rate": 1.5946730524399093e-05,
      "loss": 0.5528988242149353,
      "step": 6124
    },
    {
      "epoch": 0.5015917861785871,
      "grad_norm": 0.20720174908638,
      "learning_rate": 1.5944367853966802e-05,
      "loss": 0.8028239607810974,
      "step": 6125
    },
    {
      "epoch": 0.5016736787151062,
      "grad_norm": 0.17479228973388672,
      "learning_rate": 1.5942005012096038e-05,
      "loss": 0.7395589351654053,
      "step": 6126
    },
    {
      "epoch": 0.5017555712516251,
      "grad_norm": 0.17119121551513672,
      "learning_rate": 1.5939641998889503e-05,
      "loss": 0.805945098400116,
      "step": 6127
    },
    {
      "epoch": 0.501837463788144,
      "grad_norm": 0.19484971463680267,
      "learning_rate": 1.59372788144499e-05,
      "loss": 1.003394603729248,
      "step": 6128
    },
    {
      "epoch": 0.501919356324663,
      "grad_norm": 0.17105597257614136,
      "learning_rate": 1.5934915458879943e-05,
      "loss": 0.9606291055679321,
      "step": 6129
    },
    {
      "epoch": 0.5020012488611819,
      "grad_norm": 0.16642872989177704,
      "learning_rate": 1.5932551932282354e-05,
      "loss": 0.9622793197631836,
      "step": 6130
    },
    {
      "epoch": 0.5020831413977008,
      "grad_norm": 0.19118767976760864,
      "learning_rate": 1.593018823475986e-05,
      "loss": 1.0155919790267944,
      "step": 6131
    },
    {
      "epoch": 0.5021650339342199,
      "grad_norm": 0.2053527981042862,
      "learning_rate": 1.5927824366415197e-05,
      "loss": 0.7105567455291748,
      "step": 6132
    },
    {
      "epoch": 0.5022469264707388,
      "grad_norm": 0.15834151208400726,
      "learning_rate": 1.5925460327351115e-05,
      "loss": 1.0442135334014893,
      "step": 6133
    },
    {
      "epoch": 0.5023288190072577,
      "grad_norm": 0.1950991451740265,
      "learning_rate": 1.592309611767035e-05,
      "loss": 0.7112772464752197,
      "step": 6134
    },
    {
      "epoch": 0.5024107115437767,
      "grad_norm": 0.19311010837554932,
      "learning_rate": 1.5920731737475676e-05,
      "loss": 0.5621709227561951,
      "step": 6135
    },
    {
      "epoch": 0.5024926040802956,
      "grad_norm": 0.17039185762405396,
      "learning_rate": 1.591836718686985e-05,
      "loss": 0.8528304100036621,
      "step": 6136
    },
    {
      "epoch": 0.5025744966168146,
      "grad_norm": 0.19568243622779846,
      "learning_rate": 1.591600246595565e-05,
      "loss": 0.9229130148887634,
      "step": 6137
    },
    {
      "epoch": 0.5026563891533335,
      "grad_norm": 0.19473113119602203,
      "learning_rate": 1.591363757483585e-05,
      "loss": 0.8811407089233398,
      "step": 6138
    },
    {
      "epoch": 0.5027382816898525,
      "grad_norm": 0.1736166924238205,
      "learning_rate": 1.591127251361325e-05,
      "loss": 0.41226258873939514,
      "step": 6139
    },
    {
      "epoch": 0.5028201742263715,
      "grad_norm": 0.16802465915679932,
      "learning_rate": 1.590890728239063e-05,
      "loss": 0.7620505690574646,
      "step": 6140
    },
    {
      "epoch": 0.5029020667628904,
      "grad_norm": 0.16849781572818756,
      "learning_rate": 1.59065418812708e-05,
      "loss": 0.5825456976890564,
      "step": 6141
    },
    {
      "epoch": 0.5029839592994093,
      "grad_norm": 0.17279082536697388,
      "learning_rate": 1.5904176310356572e-05,
      "loss": 0.9386383295059204,
      "step": 6142
    },
    {
      "epoch": 0.5030658518359283,
      "grad_norm": 0.18946796655654907,
      "learning_rate": 1.590181056975076e-05,
      "loss": 0.6538257002830505,
      "step": 6143
    },
    {
      "epoch": 0.5031477443724472,
      "grad_norm": 0.17477157711982727,
      "learning_rate": 1.5899444659556193e-05,
      "loss": 0.5137530565261841,
      "step": 6144
    },
    {
      "epoch": 0.5032296369089662,
      "grad_norm": 0.17897672951221466,
      "learning_rate": 1.58970785798757e-05,
      "loss": 0.9241603016853333,
      "step": 6145
    },
    {
      "epoch": 0.5033115294454852,
      "grad_norm": 0.18752330541610718,
      "learning_rate": 1.5894712330812117e-05,
      "loss": 0.9729989767074585,
      "step": 6146
    },
    {
      "epoch": 0.5033934219820041,
      "grad_norm": 0.1529158353805542,
      "learning_rate": 1.58923459124683e-05,
      "loss": 0.7252431511878967,
      "step": 6147
    },
    {
      "epoch": 0.5034753145185231,
      "grad_norm": 0.18967436254024506,
      "learning_rate": 1.5889979324947095e-05,
      "loss": 0.9941389560699463,
      "step": 6148
    },
    {
      "epoch": 0.503557207055042,
      "grad_norm": 0.20673413574695587,
      "learning_rate": 1.588761256835137e-05,
      "loss": 0.694774329662323,
      "step": 6149
    },
    {
      "epoch": 0.503639099591561,
      "grad_norm": 0.18518875539302826,
      "learning_rate": 1.5885245642783984e-05,
      "loss": 0.9380534291267395,
      "step": 6150
    },
    {
      "epoch": 0.50372099212808,
      "grad_norm": 0.17920972406864166,
      "learning_rate": 1.588287854834783e-05,
      "loss": 0.68118816614151,
      "step": 6151
    },
    {
      "epoch": 0.5038028846645989,
      "grad_norm": 0.18359778821468353,
      "learning_rate": 1.5880511285145776e-05,
      "loss": 1.0915721654891968,
      "step": 6152
    },
    {
      "epoch": 0.5038847772011178,
      "grad_norm": 0.1891583949327469,
      "learning_rate": 1.587814385328072e-05,
      "loss": 0.7668443918228149,
      "step": 6153
    },
    {
      "epoch": 0.5039666697376368,
      "grad_norm": 0.19533923268318176,
      "learning_rate": 1.587577625285556e-05,
      "loss": 0.8623557686805725,
      "step": 6154
    },
    {
      "epoch": 0.5040485622741557,
      "grad_norm": 0.18547093868255615,
      "learning_rate": 1.5873408483973198e-05,
      "loss": 0.8885098695755005,
      "step": 6155
    },
    {
      "epoch": 0.5041304548106746,
      "grad_norm": 0.17513704299926758,
      "learning_rate": 1.5871040546736552e-05,
      "loss": 0.6870269179344177,
      "step": 6156
    },
    {
      "epoch": 0.5042123473471937,
      "grad_norm": 0.22231684625148773,
      "learning_rate": 1.5868672441248542e-05,
      "loss": 0.9237499833106995,
      "step": 6157
    },
    {
      "epoch": 0.5042942398837126,
      "grad_norm": 0.1796477735042572,
      "learning_rate": 1.586630416761209e-05,
      "loss": 0.657782256603241,
      "step": 6158
    },
    {
      "epoch": 0.5043761324202316,
      "grad_norm": 0.19057083129882812,
      "learning_rate": 1.5863935725930138e-05,
      "loss": 0.8185007572174072,
      "step": 6159
    },
    {
      "epoch": 0.5044580249567505,
      "grad_norm": 0.20670312643051147,
      "learning_rate": 1.586156711630562e-05,
      "loss": 0.9608175158500671,
      "step": 6160
    },
    {
      "epoch": 0.5045399174932694,
      "grad_norm": 0.25205540657043457,
      "learning_rate": 1.5859198338841494e-05,
      "loss": 0.7630547881126404,
      "step": 6161
    },
    {
      "epoch": 0.5046218100297885,
      "grad_norm": 0.21782465279102325,
      "learning_rate": 1.5856829393640717e-05,
      "loss": 0.9521205425262451,
      "step": 6162
    },
    {
      "epoch": 0.5047037025663074,
      "grad_norm": 0.18899668753147125,
      "learning_rate": 1.5854460280806244e-05,
      "loss": 1.1117991209030151,
      "step": 6163
    },
    {
      "epoch": 0.5047855951028263,
      "grad_norm": 0.17173950374126434,
      "learning_rate": 1.5852091000441056e-05,
      "loss": 0.8057806491851807,
      "step": 6164
    },
    {
      "epoch": 0.5048674876393453,
      "grad_norm": 0.18325428664684296,
      "learning_rate": 1.5849721552648126e-05,
      "loss": 0.8163033723831177,
      "step": 6165
    },
    {
      "epoch": 0.5049493801758642,
      "grad_norm": 0.1738894134759903,
      "learning_rate": 1.5847351937530438e-05,
      "loss": 0.7633187770843506,
      "step": 6166
    },
    {
      "epoch": 0.5050312727123831,
      "grad_norm": 0.1836252510547638,
      "learning_rate": 1.5844982155190994e-05,
      "loss": 0.7511736154556274,
      "step": 6167
    },
    {
      "epoch": 0.5051131652489022,
      "grad_norm": 0.24098731577396393,
      "learning_rate": 1.5842612205732786e-05,
      "loss": 0.8980438709259033,
      "step": 6168
    },
    {
      "epoch": 0.5051950577854211,
      "grad_norm": 0.18256351351737976,
      "learning_rate": 1.5840242089258824e-05,
      "loss": 0.966129720211029,
      "step": 6169
    },
    {
      "epoch": 0.5052769503219401,
      "grad_norm": 0.20424871146678925,
      "learning_rate": 1.5837871805872127e-05,
      "loss": 1.0273600816726685,
      "step": 6170
    },
    {
      "epoch": 0.505358842858459,
      "grad_norm": 0.19612838327884674,
      "learning_rate": 1.5835501355675715e-05,
      "loss": 0.8313472867012024,
      "step": 6171
    },
    {
      "epoch": 0.5054407353949779,
      "grad_norm": 0.20804281532764435,
      "learning_rate": 1.583313073877261e-05,
      "loss": 0.7504562139511108,
      "step": 6172
    },
    {
      "epoch": 0.5055226279314969,
      "grad_norm": 0.1730232834815979,
      "learning_rate": 1.5830759955265864e-05,
      "loss": 1.065096139907837,
      "step": 6173
    },
    {
      "epoch": 0.5056045204680159,
      "grad_norm": 0.1905190646648407,
      "learning_rate": 1.5828389005258508e-05,
      "loss": 0.5174404978752136,
      "step": 6174
    },
    {
      "epoch": 0.5056864130045348,
      "grad_norm": 0.1818682700395584,
      "learning_rate": 1.5826017888853597e-05,
      "loss": 0.9148801565170288,
      "step": 6175
    },
    {
      "epoch": 0.5057683055410538,
      "grad_norm": 0.18313413858413696,
      "learning_rate": 1.5823646606154193e-05,
      "loss": 0.9049778580665588,
      "step": 6176
    },
    {
      "epoch": 0.5058501980775727,
      "grad_norm": 0.1969725638628006,
      "learning_rate": 1.582127515726336e-05,
      "loss": 1.2798010110855103,
      "step": 6177
    },
    {
      "epoch": 0.5059320906140916,
      "grad_norm": 0.1933370679616928,
      "learning_rate": 1.5818903542284164e-05,
      "loss": 1.0624165534973145,
      "step": 6178
    },
    {
      "epoch": 0.5060139831506106,
      "grad_norm": 0.16406072676181793,
      "learning_rate": 1.58165317613197e-05,
      "loss": 0.6419641375541687,
      "step": 6179
    },
    {
      "epoch": 0.5060958756871295,
      "grad_norm": 0.17887158691883087,
      "learning_rate": 1.5814159814473038e-05,
      "loss": 0.7151022553443909,
      "step": 6180
    },
    {
      "epoch": 0.5061777682236486,
      "grad_norm": 0.1604248285293579,
      "learning_rate": 1.5811787701847286e-05,
      "loss": 0.9685138463973999,
      "step": 6181
    },
    {
      "epoch": 0.5062596607601675,
      "grad_norm": 0.14742308855056763,
      "learning_rate": 1.580941542354554e-05,
      "loss": 0.6921091675758362,
      "step": 6182
    },
    {
      "epoch": 0.5063415532966864,
      "grad_norm": 0.20879127085208893,
      "learning_rate": 1.5807042979670908e-05,
      "loss": 0.9967896938323975,
      "step": 6183
    },
    {
      "epoch": 0.5064234458332054,
      "grad_norm": 0.1870083510875702,
      "learning_rate": 1.580467037032651e-05,
      "loss": 0.8751665949821472,
      "step": 6184
    },
    {
      "epoch": 0.5065053383697243,
      "grad_norm": 0.17805488407611847,
      "learning_rate": 1.5802297595615467e-05,
      "loss": 0.9428232312202454,
      "step": 6185
    },
    {
      "epoch": 0.5065872309062432,
      "grad_norm": 0.20805010199546814,
      "learning_rate": 1.579992465564091e-05,
      "loss": 0.7506993412971497,
      "step": 6186
    },
    {
      "epoch": 0.5066691234427623,
      "grad_norm": 0.1952597200870514,
      "learning_rate": 1.579755155050598e-05,
      "loss": 0.8962850570678711,
      "step": 6187
    },
    {
      "epoch": 0.5067510159792812,
      "grad_norm": 0.1840689331293106,
      "learning_rate": 1.5795178280313815e-05,
      "loss": 0.6820369362831116,
      "step": 6188
    },
    {
      "epoch": 0.5068329085158001,
      "grad_norm": 0.19097687304019928,
      "learning_rate": 1.5792804845167567e-05,
      "loss": 1.0595669746398926,
      "step": 6189
    },
    {
      "epoch": 0.5069148010523191,
      "grad_norm": 0.20877598226070404,
      "learning_rate": 1.5790431245170405e-05,
      "loss": 0.9497317671775818,
      "step": 6190
    },
    {
      "epoch": 0.506996693588838,
      "grad_norm": 0.19021402299404144,
      "learning_rate": 1.578805748042549e-05,
      "loss": 0.7152024507522583,
      "step": 6191
    },
    {
      "epoch": 0.507078586125357,
      "grad_norm": 0.15502318739891052,
      "learning_rate": 1.5785683551035993e-05,
      "loss": 0.6683006882667542,
      "step": 6192
    },
    {
      "epoch": 0.507160478661876,
      "grad_norm": 0.1646219789981842,
      "learning_rate": 1.5783309457105097e-05,
      "loss": 0.8411416411399841,
      "step": 6193
    },
    {
      "epoch": 0.5072423711983949,
      "grad_norm": 0.15915091335773468,
      "learning_rate": 1.578093519873599e-05,
      "loss": 0.8426781296730042,
      "step": 6194
    },
    {
      "epoch": 0.5073242637349139,
      "grad_norm": 0.1753385066986084,
      "learning_rate": 1.5778560776031863e-05,
      "loss": 0.8337968587875366,
      "step": 6195
    },
    {
      "epoch": 0.5074061562714328,
      "grad_norm": 0.17402473092079163,
      "learning_rate": 1.5776186189095926e-05,
      "loss": 1.0381420850753784,
      "step": 6196
    },
    {
      "epoch": 0.5074880488079517,
      "grad_norm": 0.19346854090690613,
      "learning_rate": 1.577381143803138e-05,
      "loss": 0.7190743088722229,
      "step": 6197
    },
    {
      "epoch": 0.5075699413444708,
      "grad_norm": 0.20574545860290527,
      "learning_rate": 1.5771436522941453e-05,
      "loss": 0.9691641926765442,
      "step": 6198
    },
    {
      "epoch": 0.5076518338809897,
      "grad_norm": 0.20631082355976105,
      "learning_rate": 1.5769061443929356e-05,
      "loss": 1.0389184951782227,
      "step": 6199
    },
    {
      "epoch": 0.5077337264175086,
      "grad_norm": 0.1436527967453003,
      "learning_rate": 1.5766686201098325e-05,
      "loss": 0.7563503384590149,
      "step": 6200
    },
    {
      "epoch": 0.5078156189540276,
      "grad_norm": 0.2222694456577301,
      "learning_rate": 1.57643107945516e-05,
      "loss": 0.796898365020752,
      "step": 6201
    },
    {
      "epoch": 0.5078975114905465,
      "grad_norm": 0.2159685492515564,
      "learning_rate": 1.5761935224392422e-05,
      "loss": 1.2416822910308838,
      "step": 6202
    },
    {
      "epoch": 0.5079794040270654,
      "grad_norm": 0.23369912803173065,
      "learning_rate": 1.5759559490724044e-05,
      "loss": 0.5414125323295593,
      "step": 6203
    },
    {
      "epoch": 0.5080612965635845,
      "grad_norm": 0.1802685260772705,
      "learning_rate": 1.5757183593649728e-05,
      "loss": 0.9305001497268677,
      "step": 6204
    },
    {
      "epoch": 0.5081431891001034,
      "grad_norm": 0.16473180055618286,
      "learning_rate": 1.575480753327274e-05,
      "loss": 0.8268470764160156,
      "step": 6205
    },
    {
      "epoch": 0.5082250816366224,
      "grad_norm": 0.2181403934955597,
      "learning_rate": 1.575243130969635e-05,
      "loss": 0.9849451780319214,
      "step": 6206
    },
    {
      "epoch": 0.5083069741731413,
      "grad_norm": 0.1859103888273239,
      "learning_rate": 1.575005492302384e-05,
      "loss": 0.775029718875885,
      "step": 6207
    },
    {
      "epoch": 0.5083888667096602,
      "grad_norm": 0.19782453775405884,
      "learning_rate": 1.5747678373358498e-05,
      "loss": 0.9369567632675171,
      "step": 6208
    },
    {
      "epoch": 0.5084707592461792,
      "grad_norm": 0.19090162217617035,
      "learning_rate": 1.5745301660803615e-05,
      "loss": 0.8635470867156982,
      "step": 6209
    },
    {
      "epoch": 0.5085526517826982,
      "grad_norm": 0.16779080033302307,
      "learning_rate": 1.57429247854625e-05,
      "loss": 1.0011975765228271,
      "step": 6210
    },
    {
      "epoch": 0.5086345443192171,
      "grad_norm": 0.18283754587173462,
      "learning_rate": 1.5740547747438456e-05,
      "loss": 1.3412443399429321,
      "step": 6211
    },
    {
      "epoch": 0.5087164368557361,
      "grad_norm": 0.1835862249135971,
      "learning_rate": 1.5738170546834804e-05,
      "loss": 1.0479602813720703,
      "step": 6212
    },
    {
      "epoch": 0.508798329392255,
      "grad_norm": 0.16380038857460022,
      "learning_rate": 1.573579318375486e-05,
      "loss": 0.5879664421081543,
      "step": 6213
    },
    {
      "epoch": 0.5088802219287739,
      "grad_norm": 0.17093974351882935,
      "learning_rate": 1.5733415658301957e-05,
      "loss": 0.5378586649894714,
      "step": 6214
    },
    {
      "epoch": 0.5089621144652929,
      "grad_norm": 0.18705593049526215,
      "learning_rate": 1.5731037970579435e-05,
      "loss": 0.8711770176887512,
      "step": 6215
    }
  ],
  "logging_steps": 1,
  "max_steps": 15265,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 4580,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.7325972803557335e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}