{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2077,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00048155059290916753,
      "grad_norm": 10.997899055480957,
      "learning_rate": 0.0,
      "loss": 4.058963775634766,
      "step": 1
    },
    {
      "epoch": 0.0009631011858183351,
      "grad_norm": 10.688220024108887,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 2.896183490753174,
      "step": 2
    },
    {
      "epoch": 0.0014446517787275025,
      "grad_norm": 9.711779594421387,
      "learning_rate": 6.349206349206349e-06,
      "loss": 3.116713285446167,
      "step": 3
    },
    {
      "epoch": 0.0019262023716366701,
      "grad_norm": 8.781420707702637,
      "learning_rate": 9.523809523809523e-06,
      "loss": 3.17122745513916,
      "step": 4
    },
    {
      "epoch": 0.0024077529645458377,
      "grad_norm": 9.15394401550293,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 3.0743043422698975,
      "step": 5
    },
    {
      "epoch": 0.002889303557455005,
      "grad_norm": 12.029533386230469,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 2.984696626663208,
      "step": 6
    },
    {
      "epoch": 0.0033708541503641725,
      "grad_norm": 5.122688293457031,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 3.4119904041290283,
      "step": 7
    },
    {
      "epoch": 0.0038524047432733403,
      "grad_norm": 8.040493965148926,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 2.5286026000976562,
      "step": 8
    },
    {
      "epoch": 0.004333955336182508,
      "grad_norm": 3.5578160285949707,
      "learning_rate": 2.5396825396825397e-05,
      "loss": 2.5736966133117676,
      "step": 9
    },
    {
      "epoch": 0.004815505929091675,
      "grad_norm": 6.97898530960083,
      "learning_rate": 2.857142857142857e-05,
      "loss": 2.769012451171875,
      "step": 10
    },
    {
      "epoch": 0.005297056522000842,
      "grad_norm": 11.24144458770752,
      "learning_rate": 3.1746031746031745e-05,
      "loss": 3.7683677673339844,
      "step": 11
    },
    {
      "epoch": 0.00577860711491001,
      "grad_norm": 8.045852661132812,
      "learning_rate": 3.492063492063492e-05,
      "loss": 2.8090014457702637,
      "step": 12
    },
    {
      "epoch": 0.006260157707819178,
      "grad_norm": 5.871156692504883,
      "learning_rate": 3.809523809523809e-05,
      "loss": 2.5981831550598145,
      "step": 13
    },
    {
      "epoch": 0.006741708300728345,
      "grad_norm": 8.531469345092773,
      "learning_rate": 4.126984126984127e-05,
      "loss": 2.1772539615631104,
      "step": 14
    },
    {
      "epoch": 0.007223258893637513,
      "grad_norm": 5.277194499969482,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 3.06411075592041,
      "step": 15
    },
    {
      "epoch": 0.0077048094865466805,
      "grad_norm": 5.788327217102051,
      "learning_rate": 4.761904761904762e-05,
      "loss": 3.181549072265625,
      "step": 16
    },
    {
      "epoch": 0.008186360079455848,
      "grad_norm": 7.921364784240723,
      "learning_rate": 5.0793650793650794e-05,
      "loss": 2.962214946746826,
      "step": 17
    },
    {
      "epoch": 0.008667910672365015,
      "grad_norm": 7.972780227661133,
      "learning_rate": 5.396825396825397e-05,
      "loss": 3.1650211811065674,
      "step": 18
    },
    {
      "epoch": 0.009149461265274182,
      "grad_norm": 6.766388893127441,
      "learning_rate": 5.714285714285714e-05,
      "loss": 3.096041440963745,
      "step": 19
    },
    {
      "epoch": 0.00963101185818335,
      "grad_norm": 6.778908729553223,
      "learning_rate": 6.0317460317460316e-05,
      "loss": 2.4642131328582764,
      "step": 20
    },
    {
      "epoch": 0.010112562451092518,
      "grad_norm": 6.714580535888672,
      "learning_rate": 6.349206349206349e-05,
      "loss": 2.379725217819214,
      "step": 21
    },
    {
      "epoch": 0.010594113044001685,
      "grad_norm": 5.996400833129883,
      "learning_rate": 6.666666666666667e-05,
      "loss": 2.6206259727478027,
      "step": 22
    },
    {
      "epoch": 0.011075663636910853,
      "grad_norm": 18.50982093811035,
      "learning_rate": 6.984126984126984e-05,
      "loss": 2.2393219470977783,
      "step": 23
    },
    {
      "epoch": 0.01155721422982002,
      "grad_norm": 7.456697940826416,
      "learning_rate": 7.301587301587302e-05,
      "loss": 2.3178458213806152,
      "step": 24
    },
    {
      "epoch": 0.012038764822729187,
      "grad_norm": 9.348492622375488,
      "learning_rate": 7.619047619047618e-05,
      "loss": 2.3037354946136475,
      "step": 25
    },
    {
      "epoch": 0.012520315415638356,
      "grad_norm": 8.843018531799316,
      "learning_rate": 7.936507936507937e-05,
      "loss": 2.2980844974517822,
      "step": 26
    },
    {
      "epoch": 0.013001866008547523,
      "grad_norm": 3.3920679092407227,
      "learning_rate": 8.253968253968255e-05,
      "loss": 1.5488524436950684,
      "step": 27
    },
    {
      "epoch": 0.01348341660145669,
      "grad_norm": 4.2281174659729,
      "learning_rate": 8.571428571428571e-05,
      "loss": 2.6754565238952637,
      "step": 28
    },
    {
      "epoch": 0.013964967194365858,
      "grad_norm": 12.229516983032227,
      "learning_rate": 8.888888888888889e-05,
      "loss": 2.1637167930603027,
      "step": 29
    },
    {
      "epoch": 0.014446517787275025,
      "grad_norm": 8.624014854431152,
      "learning_rate": 9.206349206349206e-05,
      "loss": 1.9575073719024658,
      "step": 30
    },
    {
      "epoch": 0.014928068380184192,
      "grad_norm": 10.031938552856445,
      "learning_rate": 9.523809523809524e-05,
      "loss": 2.299067735671997,
      "step": 31
    },
    {
      "epoch": 0.015409618973093361,
      "grad_norm": 4.616157054901123,
      "learning_rate": 9.841269841269841e-05,
      "loss": 2.945591926574707,
      "step": 32
    },
    {
      "epoch": 0.01589116956600253,
      "grad_norm": 6.6234517097473145,
      "learning_rate": 0.00010158730158730159,
      "loss": 1.7406888008117676,
      "step": 33
    },
    {
      "epoch": 0.016372720158911697,
      "grad_norm": 6.0356950759887695,
      "learning_rate": 0.00010476190476190477,
      "loss": 2.107468843460083,
      "step": 34
    },
    {
      "epoch": 0.016854270751820864,
      "grad_norm": 12.4510498046875,
      "learning_rate": 0.00010793650793650794,
      "loss": 1.880216360092163,
      "step": 35
    },
    {
      "epoch": 0.01733582134473003,
      "grad_norm": 15.692533493041992,
      "learning_rate": 0.00011111111111111112,
      "loss": 2.3815994262695312,
      "step": 36
    },
    {
      "epoch": 0.017817371937639197,
      "grad_norm": 5.19246244430542,
      "learning_rate": 0.00011428571428571428,
      "loss": 2.592857837677002,
      "step": 37
    },
    {
      "epoch": 0.018298922530548364,
      "grad_norm": 7.335521221160889,
      "learning_rate": 0.00011746031746031746,
      "loss": 1.9369162321090698,
      "step": 38
    },
    {
      "epoch": 0.018780473123457535,
      "grad_norm": 9.601412773132324,
      "learning_rate": 0.00012063492063492063,
      "loss": 1.4544076919555664,
      "step": 39
    },
    {
      "epoch": 0.0192620237163667,
      "grad_norm": 4.23421573638916,
      "learning_rate": 0.0001238095238095238,
      "loss": 1.6413145065307617,
      "step": 40
    },
    {
      "epoch": 0.01974357430927587,
      "grad_norm": 5.228166580200195,
      "learning_rate": 0.00012698412698412698,
      "loss": 1.3118888139724731,
      "step": 41
    },
    {
      "epoch": 0.020225124902185036,
      "grad_norm": 9.26041316986084,
      "learning_rate": 0.00013015873015873017,
      "loss": 1.4355920553207397,
      "step": 42
    },
    {
      "epoch": 0.020706675495094203,
      "grad_norm": 5.0082292556762695,
      "learning_rate": 0.00013333333333333334,
      "loss": 1.5019367933273315,
      "step": 43
    },
    {
      "epoch": 0.02118822608800337,
      "grad_norm": 7.380497932434082,
      "learning_rate": 0.0001365079365079365,
      "loss": 1.3485157489776611,
      "step": 44
    },
    {
      "epoch": 0.02166977668091254,
      "grad_norm": 6.2628960609436035,
      "learning_rate": 0.00013968253968253967,
      "loss": 1.5242847204208374,
      "step": 45
    },
    {
      "epoch": 0.022151327273821707,
      "grad_norm": 4.957411289215088,
      "learning_rate": 0.00014285714285714287,
      "loss": 1.4377493858337402,
      "step": 46
    },
    {
      "epoch": 0.022632877866730874,
      "grad_norm": 9.408003807067871,
      "learning_rate": 0.00014603174603174603,
      "loss": 1.710463523864746,
      "step": 47
    },
    {
      "epoch": 0.02311442845964004,
      "grad_norm": 8.214313507080078,
      "learning_rate": 0.00014920634920634923,
      "loss": 1.5314478874206543,
      "step": 48
    },
    {
      "epoch": 0.023595979052549208,
      "grad_norm": 3.8327417373657227,
      "learning_rate": 0.00015238095238095237,
      "loss": 1.8021790981292725,
      "step": 49
    },
    {
      "epoch": 0.024077529645458375,
      "grad_norm": 4.845479965209961,
      "learning_rate": 0.00015555555555555556,
      "loss": 1.9649401903152466,
      "step": 50
    },
    {
      "epoch": 0.024559080238367545,
      "grad_norm": 5.591531753540039,
      "learning_rate": 0.00015873015873015873,
      "loss": 1.3942663669586182,
      "step": 51
    },
    {
      "epoch": 0.025040630831276712,
      "grad_norm": 3.141206741333008,
      "learning_rate": 0.00016190476190476192,
      "loss": 1.7071387767791748,
      "step": 52
    },
    {
      "epoch": 0.02552218142418588,
      "grad_norm": 3.5735738277435303,
      "learning_rate": 0.0001650793650793651,
      "loss": 0.8868046402931213,
      "step": 53
    },
    {
      "epoch": 0.026003732017095046,
      "grad_norm": 5.667084693908691,
      "learning_rate": 0.00016825396825396826,
      "loss": 1.401733636856079,
      "step": 54
    },
    {
      "epoch": 0.026485282610004213,
      "grad_norm": 3.472832679748535,
      "learning_rate": 0.00017142857142857143,
      "loss": 1.6547670364379883,
      "step": 55
    },
    {
      "epoch": 0.02696683320291338,
      "grad_norm": 4.890654563903809,
      "learning_rate": 0.00017460317460317462,
      "loss": 1.1937718391418457,
      "step": 56
    },
    {
      "epoch": 0.02744838379582255,
      "grad_norm": 3.820284605026245,
      "learning_rate": 0.00017777777777777779,
      "loss": 0.9786993861198425,
      "step": 57
    },
    {
      "epoch": 0.027929934388731717,
      "grad_norm": 3.7111337184906006,
      "learning_rate": 0.00018095238095238095,
      "loss": 0.9013885259628296,
      "step": 58
    },
    {
      "epoch": 0.028411484981640884,
      "grad_norm": 4.06239652633667,
      "learning_rate": 0.00018412698412698412,
      "loss": 1.0757468938827515,
      "step": 59
    },
    {
      "epoch": 0.02889303557455005,
      "grad_norm": 6.237166404724121,
      "learning_rate": 0.00018730158730158731,
      "loss": 1.3646478652954102,
      "step": 60
    },
    {
      "epoch": 0.029374586167459218,
      "grad_norm": 10.802021980285645,
      "learning_rate": 0.00019047619047619048,
      "loss": 1.0453048944473267,
      "step": 61
    },
    {
      "epoch": 0.029856136760368385,
      "grad_norm": 2.528367757797241,
      "learning_rate": 0.00019365079365079365,
      "loss": 1.1397546529769897,
      "step": 62
    },
    {
      "epoch": 0.030337687353277555,
      "grad_norm": 3.3737621307373047,
      "learning_rate": 0.00019682539682539682,
      "loss": 0.9105052351951599,
      "step": 63
    },
    {
      "epoch": 0.030819237946186722,
      "grad_norm": 3.399404764175415,
      "learning_rate": 0.0002,
      "loss": 1.2921056747436523,
      "step": 64
    },
    {
      "epoch": 0.03130078853909589,
      "grad_norm": 2.4056777954101562,
      "learning_rate": 0.00019999987833918285,
      "loss": 1.2942091226577759,
      "step": 65
    },
    {
      "epoch": 0.03178233913200506,
      "grad_norm": 2.7020366191864014,
      "learning_rate": 0.00019999951335702735,
      "loss": 1.829716682434082,
      "step": 66
    },
    {
      "epoch": 0.03226388972491422,
      "grad_norm": 4.247738361358643,
      "learning_rate": 0.00019999890505442158,
      "loss": 1.6451690196990967,
      "step": 67
    },
    {
      "epoch": 0.03274544031782339,
      "grad_norm": 8.600503921508789,
      "learning_rate": 0.0001999980534328457,
      "loss": 1.1226410865783691,
      "step": 68
    },
    {
      "epoch": 0.03322699091073256,
      "grad_norm": 3.182814121246338,
      "learning_rate": 0.0001999969584943719,
      "loss": 1.1608364582061768,
      "step": 69
    },
    {
      "epoch": 0.03370854150364173,
      "grad_norm": 3.4195096492767334,
      "learning_rate": 0.00019999562024166438,
      "loss": 0.9494245052337646,
      "step": 70
    },
    {
      "epoch": 0.03419009209655089,
      "grad_norm": 4.341580867767334,
      "learning_rate": 0.0001999940386779794,
      "loss": 0.8205292820930481,
      "step": 71
    },
    {
      "epoch": 0.03467164268946006,
      "grad_norm": 8.308517456054688,
      "learning_rate": 0.00019999221380716527,
      "loss": 1.5768499374389648,
      "step": 72
    },
    {
      "epoch": 0.03515319328236923,
      "grad_norm": 4.052701950073242,
      "learning_rate": 0.00019999014563366226,
      "loss": 1.0489567518234253,
      "step": 73
    },
    {
      "epoch": 0.035634743875278395,
      "grad_norm": 4.40444803237915,
      "learning_rate": 0.00019998783416250268,
      "loss": 0.9018352627754211,
      "step": 74
    },
    {
      "epoch": 0.036116294468187565,
      "grad_norm": 4.019842624664307,
      "learning_rate": 0.0001999852793993109,
      "loss": 1.4858777523040771,
      "step": 75
    },
    {
      "epoch": 0.03659784506109673,
      "grad_norm": 4.788472652435303,
      "learning_rate": 0.00019998248135030315,
      "loss": 1.061518907546997,
      "step": 76
    },
    {
      "epoch": 0.0370793956540059,
      "grad_norm": 4.773765563964844,
      "learning_rate": 0.00019997944002228774,
      "loss": 1.026103138923645,
      "step": 77
    },
    {
      "epoch": 0.03756094624691507,
      "grad_norm": 1.9739551544189453,
      "learning_rate": 0.00019997615542266482,
      "loss": 1.0412124395370483,
      "step": 78
    },
    {
      "epoch": 0.03804249683982423,
      "grad_norm": 4.075804710388184,
      "learning_rate": 0.00019997262755942655,
      "loss": 0.7491932511329651,
      "step": 79
    },
    {
      "epoch": 0.0385240474327334,
      "grad_norm": 5.227173328399658,
      "learning_rate": 0.000199968856441157,
      "loss": 0.9914160966873169,
      "step": 80
    },
    {
      "epoch": 0.03900559802564257,
      "grad_norm": 8.932019233703613,
      "learning_rate": 0.0001999648420770321,
      "loss": 1.7557719945907593,
      "step": 81
    },
    {
      "epoch": 0.03948714861855174,
      "grad_norm": 4.7886247634887695,
      "learning_rate": 0.0001999605844768197,
      "loss": 0.9102742671966553,
      "step": 82
    },
    {
      "epoch": 0.0399686992114609,
      "grad_norm": 2.3350067138671875,
      "learning_rate": 0.00019995608365087946,
      "loss": 1.2817673683166504,
      "step": 83
    },
    {
      "epoch": 0.04045024980437007,
      "grad_norm": 3.4654324054718018,
      "learning_rate": 0.0001999513396101628,
      "loss": 1.0967198610305786,
      "step": 84
    },
    {
      "epoch": 0.04093180039727924,
      "grad_norm": 5.859123706817627,
      "learning_rate": 0.00019994635236621306,
      "loss": 1.1669429540634155,
      "step": 85
    },
    {
      "epoch": 0.041413350990188405,
      "grad_norm": 3.7878470420837402,
      "learning_rate": 0.00019994112193116528,
      "loss": 1.2522754669189453,
      "step": 86
    },
    {
      "epoch": 0.041894901583097575,
      "grad_norm": 3.6203784942626953,
      "learning_rate": 0.00019993564831774618,
      "loss": 1.3948314189910889,
      "step": 87
    },
    {
      "epoch": 0.04237645217600674,
      "grad_norm": 4.7783308029174805,
      "learning_rate": 0.00019992993153927432,
      "loss": 1.559377908706665,
      "step": 88
    },
    {
      "epoch": 0.04285800276891591,
      "grad_norm": 5.1839189529418945,
      "learning_rate": 0.00019992397160965982,
      "loss": 1.271559238433838,
      "step": 89
    },
    {
      "epoch": 0.04333955336182508,
      "grad_norm": 4.8023529052734375,
      "learning_rate": 0.0001999177685434045,
      "loss": 1.2806529998779297,
      "step": 90
    },
    {
      "epoch": 0.04382110395473424,
      "grad_norm": 3.678349733352661,
      "learning_rate": 0.00019991132235560176,
      "loss": 1.2652122974395752,
      "step": 91
    },
    {
      "epoch": 0.044302654547643414,
      "grad_norm": 3.402353048324585,
      "learning_rate": 0.00019990463306193652,
      "loss": 0.9679451584815979,
      "step": 92
    },
    {
      "epoch": 0.04478420514055258,
      "grad_norm": 4.682291030883789,
      "learning_rate": 0.00019989770067868533,
      "loss": 1.259103536605835,
      "step": 93
    },
    {
      "epoch": 0.04526575573346175,
      "grad_norm": 2.8263933658599854,
      "learning_rate": 0.00019989052522271622,
      "loss": 1.4767948389053345,
      "step": 94
    },
    {
      "epoch": 0.04574730632637091,
      "grad_norm": 3.2325599193573,
      "learning_rate": 0.00019988310671148848,
      "loss": 0.77791428565979,
      "step": 95
    },
    {
      "epoch": 0.04622885691928008,
      "grad_norm": 3.4700870513916016,
      "learning_rate": 0.00019987544516305311,
      "loss": 1.060290813446045,
      "step": 96
    },
    {
      "epoch": 0.04671040751218925,
      "grad_norm": 6.077664375305176,
      "learning_rate": 0.00019986754059605222,
      "loss": 0.9726014137268066,
      "step": 97
    },
    {
      "epoch": 0.047191958105098415,
      "grad_norm": 2.2690978050231934,
      "learning_rate": 0.00019985939302971938,
      "loss": 1.2441374063491821,
      "step": 98
    },
    {
      "epoch": 0.047673508698007586,
      "grad_norm": 5.227272987365723,
      "learning_rate": 0.00019985100248387933,
      "loss": 1.0087199211120605,
      "step": 99
    },
    {
      "epoch": 0.04815505929091675,
      "grad_norm": 4.621099948883057,
      "learning_rate": 0.00019984236897894816,
      "loss": 1.142281413078308,
      "step": 100
    },
    {
      "epoch": 0.04863660988382592,
      "grad_norm": 3.417550802230835,
      "learning_rate": 0.000199833492535933,
      "loss": 0.8646891117095947,
      "step": 101
    },
    {
      "epoch": 0.04911816047673509,
      "grad_norm": 5.096126079559326,
      "learning_rate": 0.00019982437317643217,
      "loss": 1.3684113025665283,
      "step": 102
    },
    {
      "epoch": 0.04959971106964425,
      "grad_norm": 2.903071880340576,
      "learning_rate": 0.00019981501092263503,
      "loss": 1.005616545677185,
      "step": 103
    },
    {
      "epoch": 0.050081261662553424,
      "grad_norm": 3.348477363586426,
      "learning_rate": 0.00019980540579732196,
      "loss": 0.8674540519714355,
      "step": 104
    },
    {
      "epoch": 0.05056281225546259,
      "grad_norm": 1.5354300737380981,
      "learning_rate": 0.00019979555782386434,
      "loss": 0.903405487537384,
      "step": 105
    },
    {
      "epoch": 0.05104436284837176,
      "grad_norm": 3.0561022758483887,
      "learning_rate": 0.00019978546702622443,
      "loss": 1.3945115804672241,
      "step": 106
    },
    {
      "epoch": 0.05152591344128093,
      "grad_norm": 2.3886148929595947,
      "learning_rate": 0.00019977513342895532,
      "loss": 0.9068644642829895,
      "step": 107
    },
    {
      "epoch": 0.05200746403419009,
      "grad_norm": 4.122828960418701,
      "learning_rate": 0.00019976455705720083,
      "loss": 1.2185044288635254,
      "step": 108
    },
    {
      "epoch": 0.05248901462709926,
      "grad_norm": 2.908116340637207,
      "learning_rate": 0.0001997537379366956,
      "loss": 1.3574851751327515,
      "step": 109
    },
    {
      "epoch": 0.052970565220008425,
      "grad_norm": 2.2332262992858887,
      "learning_rate": 0.00019974267609376494,
      "loss": 1.0104408264160156,
      "step": 110
    },
    {
      "epoch": 0.053452115812917596,
      "grad_norm": 5.804801940917969,
      "learning_rate": 0.00019973137155532462,
      "loss": 1.0233808755874634,
      "step": 111
    },
    {
      "epoch": 0.05393366640582676,
      "grad_norm": 3.6912407875061035,
      "learning_rate": 0.00019971982434888107,
      "loss": 0.8569536209106445,
      "step": 112
    },
    {
      "epoch": 0.05441521699873593,
      "grad_norm": 4.011188983917236,
      "learning_rate": 0.00019970803450253114,
      "loss": 1.111314296722412,
      "step": 113
    },
    {
      "epoch": 0.0548967675916451,
      "grad_norm": 2.921574115753174,
      "learning_rate": 0.0001996960020449621,
      "loss": 1.1245489120483398,
      "step": 114
    },
    {
      "epoch": 0.055378318184554264,
      "grad_norm": 4.868406295776367,
      "learning_rate": 0.00019968372700545145,
      "loss": 1.1733529567718506,
      "step": 115
    },
    {
      "epoch": 0.055859868777463434,
      "grad_norm": 4.288796901702881,
      "learning_rate": 0.00019967120941386709,
      "loss": 0.8365505933761597,
      "step": 116
    },
    {
      "epoch": 0.0563414193703726,
      "grad_norm": 5.655908584594727,
      "learning_rate": 0.000199658449300667,
      "loss": 0.9552968740463257,
      "step": 117
    },
    {
      "epoch": 0.05682296996328177,
      "grad_norm": 1.9871619939804077,
      "learning_rate": 0.00019964544669689928,
      "loss": 1.1413183212280273,
      "step": 118
    },
    {
      "epoch": 0.05730452055619094,
      "grad_norm": 4.410016059875488,
      "learning_rate": 0.00019963220163420214,
      "loss": 1.7936046123504639,
      "step": 119
    },
    {
      "epoch": 0.0577860711491001,
      "grad_norm": 3.0000863075256348,
      "learning_rate": 0.0001996187141448036,
      "loss": 1.0092610120773315,
      "step": 120
    },
    {
      "epoch": 0.05826762174200927,
      "grad_norm": 2.5734429359436035,
      "learning_rate": 0.0001996049842615217,
      "loss": 0.927717924118042,
      "step": 121
    },
    {
      "epoch": 0.058749172334918436,
      "grad_norm": 1.9729766845703125,
      "learning_rate": 0.0001995910120177642,
      "loss": 1.0680543184280396,
      "step": 122
    },
    {
      "epoch": 0.059230722927827606,
      "grad_norm": 2.597019672393799,
      "learning_rate": 0.00019957679744752859,
      "loss": 0.5943896770477295,
      "step": 123
    },
    {
      "epoch": 0.05971227352073677,
      "grad_norm": 1.3411080837249756,
      "learning_rate": 0.00019956234058540195,
      "loss": 0.7752348184585571,
      "step": 124
    },
    {
      "epoch": 0.06019382411364594,
      "grad_norm": 2.2941012382507324,
      "learning_rate": 0.00019954764146656105,
      "loss": 1.2813199758529663,
      "step": 125
    },
    {
      "epoch": 0.06067537470655511,
      "grad_norm": 2.6868927478790283,
      "learning_rate": 0.00019953270012677195,
      "loss": 0.9544681310653687,
      "step": 126
    },
    {
      "epoch": 0.061156925299464274,
      "grad_norm": 4.2525458335876465,
      "learning_rate": 0.00019951751660239015,
      "loss": 0.6531223058700562,
      "step": 127
    },
    {
      "epoch": 0.061638475892373444,
      "grad_norm": 7.185410022735596,
      "learning_rate": 0.00019950209093036052,
      "loss": 1.0067213773727417,
      "step": 128
    },
    {
      "epoch": 0.06212002648528261,
      "grad_norm": 4.1806817054748535,
      "learning_rate": 0.000199486423148217,
      "loss": 1.6722239255905151,
      "step": 129
    },
    {
      "epoch": 0.06260157707819178,
      "grad_norm": 2.0638201236724854,
      "learning_rate": 0.00019947051329408276,
      "loss": 1.2204283475875854,
      "step": 130
    },
    {
      "epoch": 0.06308312767110094,
      "grad_norm": 2.3359546661376953,
      "learning_rate": 0.00019945436140666981,
      "loss": 0.7802423238754272,
      "step": 131
    },
    {
      "epoch": 0.06356467826401012,
      "grad_norm": 2.9181530475616455,
      "learning_rate": 0.0001994379675252793,
      "loss": 1.4492213726043701,
      "step": 132
    },
    {
      "epoch": 0.06404622885691928,
      "grad_norm": 4.183376312255859,
      "learning_rate": 0.00019942133168980103,
      "loss": 1.5508232116699219,
      "step": 133
    },
    {
      "epoch": 0.06452777944982845,
      "grad_norm": 5.5420050621032715,
      "learning_rate": 0.00019940445394071355,
      "loss": 1.298860788345337,
      "step": 134
    },
    {
      "epoch": 0.06500933004273761,
      "grad_norm": 3.217740297317505,
      "learning_rate": 0.0001993873343190842,
      "loss": 0.5312100648880005,
      "step": 135
    },
    {
      "epoch": 0.06549088063564679,
      "grad_norm": 3.5055596828460693,
      "learning_rate": 0.00019936997286656855,
      "loss": 1.233045220375061,
      "step": 136
    },
    {
      "epoch": 0.06597243122855595,
      "grad_norm": 6.250275611877441,
      "learning_rate": 0.00019935236962541092,
      "loss": 1.3780754804611206,
      "step": 137
    },
    {
      "epoch": 0.06645398182146511,
      "grad_norm": 2.8800201416015625,
      "learning_rate": 0.00019933452463844376,
      "loss": 1.111133337020874,
      "step": 138
    },
    {
      "epoch": 0.06693553241437429,
      "grad_norm": 1.705559253692627,
      "learning_rate": 0.00019931643794908772,
      "loss": 0.7816795706748962,
      "step": 139
    },
    {
      "epoch": 0.06741708300728345,
      "grad_norm": 2.8540704250335693,
      "learning_rate": 0.00019929810960135172,
      "loss": 1.1823416948318481,
      "step": 140
    },
    {
      "epoch": 0.06789863360019262,
      "grad_norm": 2.656675338745117,
      "learning_rate": 0.00019927953963983254,
      "loss": 0.8718074560165405,
      "step": 141
    },
    {
      "epoch": 0.06838018419310178,
      "grad_norm": 2.6843574047088623,
      "learning_rate": 0.00019926072810971492,
      "loss": 0.7173218727111816,
      "step": 142
    },
    {
      "epoch": 0.06886173478601096,
      "grad_norm": 4.494518756866455,
      "learning_rate": 0.00019924167505677137,
      "loss": 1.9853650331497192,
      "step": 143
    },
    {
      "epoch": 0.06934328537892012,
      "grad_norm": 2.8031065464019775,
      "learning_rate": 0.00019922238052736215,
      "loss": 1.1744643449783325,
      "step": 144
    },
    {
      "epoch": 0.06982483597182929,
      "grad_norm": 2.438189744949341,
      "learning_rate": 0.00019920284456843498,
      "loss": 1.1399515867233276,
      "step": 145
    },
    {
      "epoch": 0.07030638656473846,
      "grad_norm": 2.132323741912842,
      "learning_rate": 0.00019918306722752505,
      "loss": 1.0322271585464478,
      "step": 146
    },
    {
      "epoch": 0.07078793715764763,
      "grad_norm": 2.154754161834717,
      "learning_rate": 0.00019916304855275497,
      "loss": 1.4549391269683838,
      "step": 147
    },
    {
      "epoch": 0.07126948775055679,
      "grad_norm": 3.8711931705474854,
      "learning_rate": 0.00019914278859283445,
      "loss": 0.9161096811294556,
      "step": 148
    },
    {
      "epoch": 0.07175103834346597,
      "grad_norm": 3.9243061542510986,
      "learning_rate": 0.0001991222873970604,
      "loss": 0.9514060020446777,
      "step": 149
    },
    {
      "epoch": 0.07223258893637513,
      "grad_norm": 4.0745110511779785,
      "learning_rate": 0.00019910154501531663,
      "loss": 0.957796573638916,
      "step": 150
    },
    {
      "epoch": 0.0727141395292843,
      "grad_norm": 1.8631442785263062,
      "learning_rate": 0.0001990805614980739,
      "loss": 0.9070965647697449,
      "step": 151
    },
    {
      "epoch": 0.07319569012219346,
      "grad_norm": 2.0105137825012207,
      "learning_rate": 0.00019905933689638955,
      "loss": 0.8033793568611145,
      "step": 152
    },
    {
      "epoch": 0.07367724071510263,
      "grad_norm": 3.0527946949005127,
      "learning_rate": 0.00019903787126190772,
      "loss": 0.7658103108406067,
      "step": 153
    },
    {
      "epoch": 0.0741587913080118,
      "grad_norm": 3.017906904220581,
      "learning_rate": 0.00019901616464685888,
      "loss": 0.8459470272064209,
      "step": 154
    },
    {
      "epoch": 0.07464034190092096,
      "grad_norm": 6.968677043914795,
      "learning_rate": 0.00019899421710405996,
      "loss": 1.189484715461731,
      "step": 155
    },
    {
      "epoch": 0.07512189249383014,
      "grad_norm": 4.721875190734863,
      "learning_rate": 0.00019897202868691407,
      "loss": 1.5837947130203247,
      "step": 156
    },
    {
      "epoch": 0.0756034430867393,
      "grad_norm": 1.931235671043396,
      "learning_rate": 0.00019894959944941038,
      "loss": 1.085717797279358,
      "step": 157
    },
    {
      "epoch": 0.07608499367964847,
      "grad_norm": 2.671604633331299,
      "learning_rate": 0.0001989269294461242,
      "loss": 0.8639560341835022,
      "step": 158
    },
    {
      "epoch": 0.07656654427255763,
      "grad_norm": 4.194220066070557,
      "learning_rate": 0.0001989040187322164,
      "loss": 1.1585806608200073,
      "step": 159
    },
    {
      "epoch": 0.0770480948654668,
      "grad_norm": 1.9198507070541382,
      "learning_rate": 0.00019888086736343384,
      "loss": 0.8374260663986206,
      "step": 160
    },
    {
      "epoch": 0.07752964545837597,
      "grad_norm": 4.862279415130615,
      "learning_rate": 0.0001988574753961087,
      "loss": 1.3817564249038696,
      "step": 161
    },
    {
      "epoch": 0.07801119605128513,
      "grad_norm": 2.366821765899658,
      "learning_rate": 0.00019883384288715874,
      "loss": 0.950917661190033,
      "step": 162
    },
    {
      "epoch": 0.07849274664419431,
      "grad_norm": 3.5843851566314697,
      "learning_rate": 0.000198809969894087,
      "loss": 1.1706701517105103,
      "step": 163
    },
    {
      "epoch": 0.07897429723710347,
      "grad_norm": 2.301211357116699,
      "learning_rate": 0.0001987858564749816,
      "loss": 1.221076250076294,
      "step": 164
    },
    {
      "epoch": 0.07945584783001264,
      "grad_norm": 2.205080270767212,
      "learning_rate": 0.00019876150268851572,
      "loss": 0.9410824179649353,
      "step": 165
    },
    {
      "epoch": 0.0799373984229218,
      "grad_norm": 3.0851192474365234,
      "learning_rate": 0.00019873690859394737,
      "loss": 0.967120885848999,
      "step": 166
    },
    {
      "epoch": 0.08041894901583098,
      "grad_norm": 2.6348469257354736,
      "learning_rate": 0.0001987120742511193,
      "loss": 1.14762282371521,
      "step": 167
    },
    {
      "epoch": 0.08090049960874014,
      "grad_norm": 3.3243017196655273,
      "learning_rate": 0.0001986869997204589,
      "loss": 0.7951797246932983,
      "step": 168
    },
    {
      "epoch": 0.0813820502016493,
      "grad_norm": 6.767729759216309,
      "learning_rate": 0.00019866168506297788,
      "loss": 1.4251643419265747,
      "step": 169
    },
    {
      "epoch": 0.08186360079455848,
      "grad_norm": 4.619649887084961,
      "learning_rate": 0.00019863613034027224,
      "loss": 0.950486421585083,
      "step": 170
    },
    {
      "epoch": 0.08234515138746765,
      "grad_norm": 2.2103381156921387,
      "learning_rate": 0.00019861033561452223,
      "loss": 0.983525812625885,
      "step": 171
    },
    {
      "epoch": 0.08282670198037681,
      "grad_norm": 3.837252140045166,
      "learning_rate": 0.00019858430094849195,
      "loss": 0.9877191185951233,
      "step": 172
    },
    {
      "epoch": 0.08330825257328599,
      "grad_norm": 8.303791999816895,
      "learning_rate": 0.0001985580264055294,
      "loss": 0.8794852495193481,
      "step": 173
    },
    {
      "epoch": 0.08378980316619515,
      "grad_norm": 9.538949012756348,
      "learning_rate": 0.00019853151204956616,
      "loss": 1.2599213123321533,
      "step": 174
    },
    {
      "epoch": 0.08427135375910431,
      "grad_norm": 4.166004180908203,
      "learning_rate": 0.00019850475794511749,
      "loss": 1.0261119604110718,
      "step": 175
    },
    {
      "epoch": 0.08475290435201348,
      "grad_norm": 3.8304171562194824,
      "learning_rate": 0.00019847776415728185,
      "loss": 1.1749240159988403,
      "step": 176
    },
    {
      "epoch": 0.08523445494492266,
      "grad_norm": 2.7093453407287598,
      "learning_rate": 0.000198450530751741,
      "loss": 1.4287409782409668,
      "step": 177
    },
    {
      "epoch": 0.08571600553783182,
      "grad_norm": 4.8279128074646,
      "learning_rate": 0.00019842305779475968,
      "loss": 1.082953691482544,
      "step": 178
    },
    {
      "epoch": 0.08619755613074098,
      "grad_norm": 2.7188494205474854,
      "learning_rate": 0.00019839534535318558,
      "loss": 1.1618900299072266,
      "step": 179
    },
    {
      "epoch": 0.08667910672365016,
      "grad_norm": 2.4408013820648193,
      "learning_rate": 0.00019836739349444899,
      "loss": 0.8602584004402161,
      "step": 180
    },
    {
      "epoch": 0.08716065731655932,
      "grad_norm": 4.936408519744873,
      "learning_rate": 0.00019833920228656292,
      "loss": 0.9554656147956848,
      "step": 181
    },
    {
      "epoch": 0.08764220790946849,
      "grad_norm": 4.548085689544678,
      "learning_rate": 0.0001983107717981226,
      "loss": 0.9557523131370544,
      "step": 182
    },
    {
      "epoch": 0.08812375850237765,
      "grad_norm": 2.5032007694244385,
      "learning_rate": 0.00019828210209830562,
      "loss": 0.6154648661613464,
      "step": 183
    },
    {
      "epoch": 0.08860530909528683,
      "grad_norm": 2.8823978900909424,
      "learning_rate": 0.00019825319325687154,
      "loss": 0.7783450484275818,
      "step": 184
    },
    {
      "epoch": 0.08908685968819599,
      "grad_norm": 4.330085754394531,
      "learning_rate": 0.00019822404534416182,
      "loss": 0.7848780751228333,
      "step": 185
    },
    {
      "epoch": 0.08956841028110515,
      "grad_norm": 2.8750452995300293,
      "learning_rate": 0.00019819465843109963,
      "loss": 0.9348660707473755,
      "step": 186
    },
    {
      "epoch": 0.09004996087401433,
      "grad_norm": 3.339097499847412,
      "learning_rate": 0.00019816503258918969,
      "loss": 0.852605938911438,
      "step": 187
    },
    {
      "epoch": 0.0905315114669235,
      "grad_norm": 4.114784240722656,
      "learning_rate": 0.00019813516789051808,
      "loss": 1.1537458896636963,
      "step": 188
    },
    {
      "epoch": 0.09101306205983266,
      "grad_norm": 1.6025184392929077,
      "learning_rate": 0.0001981050644077521,
      "loss": 1.062227487564087,
      "step": 189
    },
    {
      "epoch": 0.09149461265274182,
      "grad_norm": 3.3087575435638428,
      "learning_rate": 0.00019807472221414002,
      "loss": 0.8522804379463196,
      "step": 190
    },
    {
      "epoch": 0.091976163245651,
      "grad_norm": 4.099035263061523,
      "learning_rate": 0.00019804414138351094,
      "loss": 1.1126275062561035,
      "step": 191
    },
    {
      "epoch": 0.09245771383856016,
      "grad_norm": 3.191251516342163,
      "learning_rate": 0.00019801332199027467,
      "loss": 1.0428874492645264,
      "step": 192
    },
    {
      "epoch": 0.09293926443146933,
      "grad_norm": 1.8017200231552124,
      "learning_rate": 0.00019798226410942146,
      "loss": 0.9217344522476196,
      "step": 193
    },
    {
      "epoch": 0.0934208150243785,
      "grad_norm": 3.0027658939361572,
      "learning_rate": 0.00019795096781652182,
      "loss": 0.9579524397850037,
      "step": 194
    },
    {
      "epoch": 0.09390236561728767,
      "grad_norm": 3.9340720176696777,
      "learning_rate": 0.00019791943318772643,
      "loss": 1.0433151721954346,
      "step": 195
    },
    {
      "epoch": 0.09438391621019683,
      "grad_norm": 4.387567043304443,
      "learning_rate": 0.00019788766029976587,
      "loss": 0.9690942764282227,
      "step": 196
    },
    {
      "epoch": 0.09486546680310601,
      "grad_norm": 2.9441487789154053,
      "learning_rate": 0.0001978556492299504,
      "loss": 0.8219008445739746,
      "step": 197
    },
    {
      "epoch": 0.09534701739601517,
      "grad_norm": 6.007349491119385,
      "learning_rate": 0.00019782340005616996,
      "loss": 0.8440435528755188,
      "step": 198
    },
    {
      "epoch": 0.09582856798892433,
      "grad_norm": 3.545503616333008,
      "learning_rate": 0.0001977909128568937,
      "loss": 0.8135607242584229,
      "step": 199
    },
    {
      "epoch": 0.0963101185818335,
      "grad_norm": 3.472895383834839,
      "learning_rate": 0.00019775818771117,
      "loss": 0.9798284769058228,
      "step": 200
    },
    {
      "epoch": 0.09679166917474268,
      "grad_norm": 4.097575664520264,
      "learning_rate": 0.00019772522469862626,
      "loss": 0.8414688110351562,
      "step": 201
    },
    {
      "epoch": 0.09727321976765184,
      "grad_norm": 5.410140514373779,
      "learning_rate": 0.00019769202389946863,
      "loss": 0.8919621706008911,
      "step": 202
    },
    {
      "epoch": 0.097754770360561,
      "grad_norm": 6.864431858062744,
      "learning_rate": 0.0001976585853944818,
      "loss": 1.0897730588912964,
      "step": 203
    },
    {
      "epoch": 0.09823632095347018,
      "grad_norm": 8.007363319396973,
      "learning_rate": 0.0001976249092650289,
      "loss": 1.319188117980957,
      "step": 204
    },
    {
      "epoch": 0.09871787154637934,
      "grad_norm": 3.0758566856384277,
      "learning_rate": 0.00019759099559305124,
      "loss": 1.063638687133789,
      "step": 205
    },
    {
      "epoch": 0.0991994221392885,
      "grad_norm": 3.2806081771850586,
      "learning_rate": 0.00019755684446106812,
      "loss": 0.9325254559516907,
      "step": 206
    },
    {
      "epoch": 0.09968097273219767,
      "grad_norm": 3.1320528984069824,
      "learning_rate": 0.00019752245595217662,
      "loss": 1.2423100471496582,
      "step": 207
    },
    {
      "epoch": 0.10016252332510685,
      "grad_norm": 2.852522850036621,
      "learning_rate": 0.00019748783015005144,
      "loss": 0.8145530223846436,
      "step": 208
    },
    {
      "epoch": 0.10064407391801601,
      "grad_norm": 5.163227558135986,
      "learning_rate": 0.00019745296713894465,
      "loss": 1.6855522394180298,
      "step": 209
    },
    {
      "epoch": 0.10112562451092517,
      "grad_norm": 2.2417426109313965,
      "learning_rate": 0.00019741786700368548,
      "loss": 0.6451354026794434,
      "step": 210
    },
    {
      "epoch": 0.10160717510383435,
      "grad_norm": 1.9955745935440063,
      "learning_rate": 0.00019738252982968017,
      "loss": 0.8176712989807129,
      "step": 211
    },
    {
      "epoch": 0.10208872569674352,
      "grad_norm": 4.188888072967529,
      "learning_rate": 0.00019734695570291168,
      "loss": 1.0915836095809937,
      "step": 212
    },
    {
      "epoch": 0.10257027628965268,
      "grad_norm": 5.156767845153809,
      "learning_rate": 0.00019731114470993962,
      "loss": 0.8809371590614319,
      "step": 213
    },
    {
      "epoch": 0.10305182688256186,
      "grad_norm": 4.508893013000488,
      "learning_rate": 0.0001972750969378998,
      "loss": 1.1925499439239502,
      "step": 214
    },
    {
      "epoch": 0.10353337747547102,
      "grad_norm": 5.5759100914001465,
      "learning_rate": 0.00019723881247450434,
      "loss": 0.8763666152954102,
      "step": 215
    },
    {
      "epoch": 0.10401492806838018,
      "grad_norm": 4.394690990447998,
      "learning_rate": 0.0001972022914080411,
      "loss": 1.5408443212509155,
      "step": 216
    },
    {
      "epoch": 0.10449647866128935,
      "grad_norm": 4.205685615539551,
      "learning_rate": 0.00019716553382737379,
      "loss": 1.1494263410568237,
      "step": 217
    },
    {
      "epoch": 0.10497802925419852,
      "grad_norm": 3.2098426818847656,
      "learning_rate": 0.00019712853982194152,
      "loss": 1.0792419910430908,
      "step": 218
    },
    {
      "epoch": 0.10545957984710769,
      "grad_norm": 3.500704526901245,
      "learning_rate": 0.00019709130948175876,
      "loss": 1.1637043952941895,
      "step": 219
    },
    {
      "epoch": 0.10594113044001685,
      "grad_norm": 3.043656826019287,
      "learning_rate": 0.0001970538428974149,
      "loss": 1.0555601119995117,
      "step": 220
    },
    {
      "epoch": 0.10642268103292603,
      "grad_norm": 1.6878787279129028,
      "learning_rate": 0.00019701614016007436,
      "loss": 1.065384864807129,
      "step": 221
    },
    {
      "epoch": 0.10690423162583519,
      "grad_norm": 3.139033317565918,
      "learning_rate": 0.00019697820136147597,
      "loss": 0.8590763807296753,
      "step": 222
    },
    {
      "epoch": 0.10738578221874436,
      "grad_norm": 2.716961145401001,
      "learning_rate": 0.00019694002659393305,
      "loss": 1.1652765274047852,
      "step": 223
    },
    {
      "epoch": 0.10786733281165352,
      "grad_norm": 3.1344659328460693,
      "learning_rate": 0.0001969016159503331,
      "loss": 0.9715149402618408,
      "step": 224
    },
    {
      "epoch": 0.1083488834045627,
      "grad_norm": 1.8848605155944824,
      "learning_rate": 0.00019686296952413747,
      "loss": 0.6822095513343811,
      "step": 225
    },
    {
      "epoch": 0.10883043399747186,
      "grad_norm": 2.5366370677948,
      "learning_rate": 0.0001968240874093813,
      "loss": 0.5937865972518921,
      "step": 226
    },
    {
      "epoch": 0.10931198459038102,
      "grad_norm": 4.067916393280029,
      "learning_rate": 0.00019678496970067325,
      "loss": 0.8834372162818909,
      "step": 227
    },
    {
      "epoch": 0.1097935351832902,
      "grad_norm": 1.9713938236236572,
      "learning_rate": 0.0001967456164931951,
      "loss": 0.6779223084449768,
      "step": 228
    },
    {
      "epoch": 0.11027508577619936,
      "grad_norm": 3.140136957168579,
      "learning_rate": 0.0001967060278827017,
      "loss": 0.9885010719299316,
      "step": 229
    },
    {
      "epoch": 0.11075663636910853,
      "grad_norm": 2.9080259799957275,
      "learning_rate": 0.00019666620396552076,
      "loss": 1.0491156578063965,
      "step": 230
    },
    {
      "epoch": 0.11123818696201769,
      "grad_norm": 1.4264732599258423,
      "learning_rate": 0.00019662614483855246,
      "loss": 1.0967994928359985,
      "step": 231
    },
    {
      "epoch": 0.11171973755492687,
      "grad_norm": 4.232779026031494,
      "learning_rate": 0.00019658585059926934,
      "loss": 0.6501812934875488,
      "step": 232
    },
    {
      "epoch": 0.11220128814783603,
      "grad_norm": 4.248823642730713,
      "learning_rate": 0.00019654532134571594,
      "loss": 1.6067156791687012,
      "step": 233
    },
    {
      "epoch": 0.1126828387407452,
      "grad_norm": 2.259688138961792,
      "learning_rate": 0.00019650455717650878,
      "loss": 0.7143255472183228,
      "step": 234
    },
    {
      "epoch": 0.11316438933365437,
      "grad_norm": 2.560673236846924,
      "learning_rate": 0.00019646355819083589,
      "loss": 0.8963885307312012,
      "step": 235
    },
    {
      "epoch": 0.11364593992656354,
      "grad_norm": 4.712884902954102,
      "learning_rate": 0.0001964223244884566,
      "loss": 1.0484883785247803,
      "step": 236
    },
    {
      "epoch": 0.1141274905194727,
      "grad_norm": 3.8194100856781006,
      "learning_rate": 0.00019638085616970153,
      "loss": 1.5355582237243652,
      "step": 237
    },
    {
      "epoch": 0.11460904111238188,
      "grad_norm": 4.143054485321045,
      "learning_rate": 0.00019633915333547202,
      "loss": 0.9635744094848633,
      "step": 238
    },
    {
      "epoch": 0.11509059170529104,
      "grad_norm": 2.48124361038208,
      "learning_rate": 0.00019629721608724004,
      "loss": 0.7980287075042725,
      "step": 239
    },
    {
      "epoch": 0.1155721422982002,
      "grad_norm": 3.2239785194396973,
      "learning_rate": 0.0001962550445270481,
      "loss": 1.3067378997802734,
      "step": 240
    },
    {
      "epoch": 0.11605369289110937,
      "grad_norm": 2.41634202003479,
      "learning_rate": 0.00019621263875750864,
      "loss": 1.3808801174163818,
      "step": 241
    },
    {
      "epoch": 0.11653524348401854,
      "grad_norm": 6.938711643218994,
      "learning_rate": 0.00019616999888180406,
      "loss": 1.2598068714141846,
      "step": 242
    },
    {
      "epoch": 0.11701679407692771,
      "grad_norm": 4.132026195526123,
      "learning_rate": 0.0001961271250036865,
      "loss": 1.1564222574234009,
      "step": 243
    },
    {
      "epoch": 0.11749834466983687,
      "grad_norm": 2.73398756980896,
      "learning_rate": 0.0001960840172274773,
      "loss": 1.0761322975158691,
      "step": 244
    },
    {
      "epoch": 0.11797989526274605,
      "grad_norm": 3.019162893295288,
      "learning_rate": 0.00019604067565806704,
      "loss": 0.8687995076179504,
      "step": 245
    },
    {
      "epoch": 0.11846144585565521,
      "grad_norm": 4.242251873016357,
      "learning_rate": 0.00019599710040091512,
      "loss": 1.123656988143921,
      "step": 246
    },
    {
      "epoch": 0.11894299644856438,
      "grad_norm": 3.6356210708618164,
      "learning_rate": 0.00019595329156204955,
      "loss": 1.0730966329574585,
      "step": 247
    },
    {
      "epoch": 0.11942454704147354,
      "grad_norm": 4.293292045593262,
      "learning_rate": 0.00019590924924806676,
      "loss": 1.356764316558838,
      "step": 248
    },
    {
      "epoch": 0.11990609763438272,
      "grad_norm": 5.343273162841797,
      "learning_rate": 0.0001958649735661312,
      "loss": 1.1163526773452759,
      "step": 249
    },
    {
      "epoch": 0.12038764822729188,
      "grad_norm": 4.327155590057373,
      "learning_rate": 0.00019582046462397515,
      "loss": 1.1132748126983643,
      "step": 250
    },
    {
      "epoch": 0.12086919882020104,
      "grad_norm": 2.669375419616699,
      "learning_rate": 0.00019577572252989854,
      "loss": 1.2773585319519043,
      "step": 251
    },
    {
      "epoch": 0.12135074941311022,
      "grad_norm": 2.766550302505493,
      "learning_rate": 0.00019573074739276858,
      "loss": 1.2460675239562988,
      "step": 252
    },
    {
      "epoch": 0.12183230000601938,
      "grad_norm": 4.023970603942871,
      "learning_rate": 0.00019568553932201947,
      "loss": 0.5884127616882324,
      "step": 253
    },
    {
      "epoch": 0.12231385059892855,
      "grad_norm": 2.5779101848602295,
      "learning_rate": 0.00019564009842765225,
      "loss": 1.1471140384674072,
      "step": 254
    },
    {
      "epoch": 0.12279540119183771,
      "grad_norm": 6.043779373168945,
      "learning_rate": 0.00019559442482023444,
      "loss": 1.1978514194488525,
      "step": 255
    },
    {
      "epoch": 0.12327695178474689,
      "grad_norm": 10.41842269897461,
      "learning_rate": 0.0001955485186108998,
      "loss": 1.285045862197876,
      "step": 256
    },
    {
      "epoch": 0.12375850237765605,
      "grad_norm": 5.805689334869385,
      "learning_rate": 0.00019550237991134805,
      "loss": 1.1859052181243896,
      "step": 257
    },
    {
      "epoch": 0.12424005297056522,
      "grad_norm": 2.6859452724456787,
      "learning_rate": 0.00019545600883384467,
      "loss": 0.6984687447547913,
      "step": 258
    },
    {
      "epoch": 0.12472160356347439,
      "grad_norm": 2.1337859630584717,
      "learning_rate": 0.0001954094054912205,
      "loss": 0.7945561408996582,
      "step": 259
    },
    {
      "epoch": 0.12520315415638356,
      "grad_norm": 2.9161362648010254,
      "learning_rate": 0.00019536256999687157,
      "loss": 0.7539944648742676,
      "step": 260
    },
    {
      "epoch": 0.12568470474929272,
      "grad_norm": 3.417652130126953,
      "learning_rate": 0.00019531550246475876,
      "loss": 0.6809104681015015,
      "step": 261
    },
    {
      "epoch": 0.12616625534220188,
      "grad_norm": 2.307227849960327,
      "learning_rate": 0.00019526820300940756,
      "loss": 0.7057311534881592,
      "step": 262
    },
    {
      "epoch": 0.12664780593511105,
      "grad_norm": 6.535157680511475,
      "learning_rate": 0.00019522067174590778,
      "loss": 1.2811944484710693,
      "step": 263
    },
    {
      "epoch": 0.12712935652802024,
      "grad_norm": 6.469608783721924,
      "learning_rate": 0.00019517290878991324,
      "loss": 0.8942220211029053,
      "step": 264
    },
    {
      "epoch": 0.1276109071209294,
      "grad_norm": 1.8611575365066528,
      "learning_rate": 0.0001951249142576416,
      "loss": 0.9608644843101501,
      "step": 265
    },
    {
      "epoch": 0.12809245771383856,
      "grad_norm": 3.7973973751068115,
      "learning_rate": 0.00019507668826587387,
      "loss": 1.0413118600845337,
      "step": 266
    },
    {
      "epoch": 0.12857400830674773,
      "grad_norm": 3.2934303283691406,
      "learning_rate": 0.0001950282309319544,
      "loss": 1.500727891921997,
      "step": 267
    },
    {
      "epoch": 0.1290555588996569,
      "grad_norm": 3.1601169109344482,
      "learning_rate": 0.0001949795423737903,
      "loss": 1.2729198932647705,
      "step": 268
    },
    {
      "epoch": 0.12953710949256605,
      "grad_norm": 2.145388126373291,
      "learning_rate": 0.00019493062270985144,
      "loss": 0.9249171614646912,
      "step": 269
    },
    {
      "epoch": 0.13001866008547522,
      "grad_norm": 3.4702587127685547,
      "learning_rate": 0.00019488147205916985,
      "loss": 0.7772440910339355,
      "step": 270
    },
    {
      "epoch": 0.1305002106783844,
      "grad_norm": 4.000960826873779,
      "learning_rate": 0.00019483209054133976,
      "loss": 1.1122386455535889,
      "step": 271
    },
    {
      "epoch": 0.13098176127129357,
      "grad_norm": 3.267839193344116,
      "learning_rate": 0.00019478247827651708,
      "loss": 1.021576166152954,
      "step": 272
    },
    {
      "epoch": 0.13146331186420274,
      "grad_norm": 3.086031198501587,
      "learning_rate": 0.00019473263538541914,
      "loss": 0.9281460642814636,
      "step": 273
    },
    {
      "epoch": 0.1319448624571119,
      "grad_norm": 3.7234513759613037,
      "learning_rate": 0.00019468256198932455,
      "loss": 1.3575233221054077,
      "step": 274
    },
    {
      "epoch": 0.13242641305002106,
      "grad_norm": 1.9337539672851562,
      "learning_rate": 0.00019463225821007268,
      "loss": 0.8995951414108276,
      "step": 275
    },
    {
      "epoch": 0.13290796364293023,
      "grad_norm": 2.079463243484497,
      "learning_rate": 0.00019458172417006347,
      "loss": 0.48799100518226624,
      "step": 276
    },
    {
      "epoch": 0.1333895142358394,
      "grad_norm": 4.054724216461182,
      "learning_rate": 0.00019453095999225726,
      "loss": 1.629692554473877,
      "step": 277
    },
    {
      "epoch": 0.13387106482874858,
      "grad_norm": 2.7698373794555664,
      "learning_rate": 0.0001944799658001742,
      "loss": 0.5645720958709717,
      "step": 278
    },
    {
      "epoch": 0.13435261542165775,
      "grad_norm": 3.575896978378296,
      "learning_rate": 0.00019442874171789418,
      "loss": 1.010619878768921,
      "step": 279
    },
    {
      "epoch": 0.1348341660145669,
      "grad_norm": 1.9331309795379639,
      "learning_rate": 0.00019437728787005657,
      "loss": 1.3104228973388672,
      "step": 280
    },
    {
      "epoch": 0.13531571660747607,
      "grad_norm": 4.0163044929504395,
      "learning_rate": 0.00019432560438185963,
      "loss": 1.0282176733016968,
      "step": 281
    },
    {
      "epoch": 0.13579726720038524,
      "grad_norm": 2.375638008117676,
      "learning_rate": 0.00019427369137906046,
      "loss": 0.7544271349906921,
      "step": 282
    },
    {
      "epoch": 0.1362788177932944,
      "grad_norm": 3.7285213470458984,
      "learning_rate": 0.00019422154898797472,
      "loss": 1.2790626287460327,
      "step": 283
    },
    {
      "epoch": 0.13676036838620356,
      "grad_norm": 1.9429062604904175,
      "learning_rate": 0.00019416917733547603,
      "loss": 1.3861905336380005,
      "step": 284
    },
    {
      "epoch": 0.13724191897911275,
      "grad_norm": 3.5537188053131104,
      "learning_rate": 0.00019411657654899597,
      "loss": 1.0798075199127197,
      "step": 285
    },
    {
      "epoch": 0.13772346957202192,
      "grad_norm": 4.190849304199219,
      "learning_rate": 0.0001940637467565237,
      "loss": 1.1922192573547363,
      "step": 286
    },
    {
      "epoch": 0.13820502016493108,
      "grad_norm": 2.8111732006073,
      "learning_rate": 0.00019401068808660546,
      "loss": 1.1688849925994873,
      "step": 287
    },
    {
      "epoch": 0.13868657075784024,
      "grad_norm": 5.3347930908203125,
      "learning_rate": 0.0001939574006683445,
      "loss": 1.1827162504196167,
      "step": 288
    },
    {
      "epoch": 0.1391681213507494,
      "grad_norm": 3.2476818561553955,
      "learning_rate": 0.00019390388463140065,
      "loss": 1.111840009689331,
      "step": 289
    },
    {
      "epoch": 0.13964967194365857,
      "grad_norm": 4.765700340270996,
      "learning_rate": 0.00019385014010598998,
      "loss": 0.9624571800231934,
      "step": 290
    },
    {
      "epoch": 0.14013122253656773,
      "grad_norm": 4.042060375213623,
      "learning_rate": 0.00019379616722288456,
      "loss": 0.6930976510047913,
      "step": 291
    },
    {
      "epoch": 0.14061277312947693,
      "grad_norm": 1.8888696432113647,
      "learning_rate": 0.0001937419661134121,
      "loss": 1.3983337879180908,
      "step": 292
    },
    {
      "epoch": 0.1410943237223861,
      "grad_norm": 3.018704652786255,
      "learning_rate": 0.0001936875369094556,
      "loss": 1.013681411743164,
      "step": 293
    },
    {
      "epoch": 0.14157587431529525,
      "grad_norm": 2.8442327976226807,
      "learning_rate": 0.0001936328797434531,
      "loss": 1.0406699180603027,
      "step": 294
    },
    {
      "epoch": 0.14205742490820442,
      "grad_norm": 2.4000165462493896,
      "learning_rate": 0.00019357799474839735,
      "loss": 1.1291037797927856,
      "step": 295
    },
    {
      "epoch": 0.14253897550111358,
      "grad_norm": 4.966221332550049,
      "learning_rate": 0.00019352288205783536,
      "loss": 1.167711615562439,
      "step": 296
    },
    {
      "epoch": 0.14302052609402274,
      "grad_norm": 4.377453327178955,
      "learning_rate": 0.00019346754180586825,
      "loss": 0.8090459108352661,
      "step": 297
    },
    {
      "epoch": 0.14350207668693193,
      "grad_norm": 3.4018476009368896,
      "learning_rate": 0.00019341197412715082,
      "loss": 1.2430338859558105,
      "step": 298
    },
    {
      "epoch": 0.1439836272798411,
      "grad_norm": 11.827299118041992,
      "learning_rate": 0.00019335617915689128,
      "loss": 0.8407375812530518,
      "step": 299
    },
    {
      "epoch": 0.14446517787275026,
      "grad_norm": 4.312981605529785,
      "learning_rate": 0.00019330015703085082,
      "loss": 0.6932576298713684,
      "step": 300
    },
    {
      "epoch": 0.14494672846565942,
      "grad_norm": 5.305436611175537,
      "learning_rate": 0.00019324390788534343,
      "loss": 1.4131890535354614,
      "step": 301
    },
    {
      "epoch": 0.1454282790585686,
      "grad_norm": 3.1825668811798096,
      "learning_rate": 0.00019318743185723546,
      "loss": 0.8375182151794434,
      "step": 302
    },
    {
      "epoch": 0.14590982965147775,
      "grad_norm": 4.367538928985596,
      "learning_rate": 0.00019313072908394525,
      "loss": 0.7957690954208374,
      "step": 303
    },
    {
      "epoch": 0.14639138024438691,
      "grad_norm": 1.6075003147125244,
      "learning_rate": 0.00019307379970344294,
      "loss": 0.951972246170044,
      "step": 304
    },
    {
      "epoch": 0.1468729308372961,
      "grad_norm": 1.4260181188583374,
      "learning_rate": 0.00019301664385425004,
      "loss": 0.9693677425384521,
      "step": 305
    },
    {
      "epoch": 0.14735448143020527,
      "grad_norm": 3.8832671642303467,
      "learning_rate": 0.0001929592616754391,
      "loss": 0.887086808681488,
      "step": 306
    },
    {
      "epoch": 0.14783603202311443,
      "grad_norm": 2.599985122680664,
      "learning_rate": 0.00019290165330663336,
      "loss": 1.0574231147766113,
      "step": 307
    },
    {
      "epoch": 0.1483175826160236,
      "grad_norm": 3.122762441635132,
      "learning_rate": 0.00019284381888800647,
      "loss": 0.690856397151947,
      "step": 308
    },
    {
      "epoch": 0.14879913320893276,
      "grad_norm": 2.3026108741760254,
      "learning_rate": 0.00019278575856028206,
      "loss": 0.8600311875343323,
      "step": 309
    },
    {
      "epoch": 0.14928068380184192,
      "grad_norm": 1.818914532661438,
      "learning_rate": 0.00019272747246473345,
      "loss": 0.7881981730461121,
      "step": 310
    },
    {
      "epoch": 0.1497622343947511,
      "grad_norm": 3.683953285217285,
      "learning_rate": 0.00019266896074318334,
      "loss": 0.9407275319099426,
      "step": 311
    },
    {
      "epoch": 0.15024378498766028,
      "grad_norm": 3.187880277633667,
      "learning_rate": 0.00019261022353800344,
      "loss": 1.434323787689209,
      "step": 312
    },
    {
      "epoch": 0.15072533558056944,
      "grad_norm": 2.867413282394409,
      "learning_rate": 0.00019255126099211402,
      "loss": 1.2948453426361084,
      "step": 313
    },
    {
      "epoch": 0.1512068861734786,
      "grad_norm": 4.420424461364746,
      "learning_rate": 0.00019249207324898376,
      "loss": 1.2659118175506592,
      "step": 314
    },
    {
      "epoch": 0.15168843676638777,
      "grad_norm": 3.33125376701355,
      "learning_rate": 0.0001924326604526292,
      "loss": 1.0201218128204346,
      "step": 315
    },
    {
      "epoch": 0.15216998735929693,
      "grad_norm": 1.7288691997528076,
      "learning_rate": 0.00019237302274761458,
      "loss": 1.0658215284347534,
      "step": 316
    },
    {
      "epoch": 0.1526515379522061,
      "grad_norm": 3.2253782749176025,
      "learning_rate": 0.0001923131602790513,
      "loss": 1.3618834018707275,
      "step": 317
    },
    {
      "epoch": 0.15313308854511526,
      "grad_norm": 3.4548285007476807,
      "learning_rate": 0.00019225307319259768,
      "loss": 1.2979847192764282,
      "step": 318
    },
    {
      "epoch": 0.15361463913802445,
      "grad_norm": 2.00081729888916,
      "learning_rate": 0.00019219276163445862,
      "loss": 0.5882802605628967,
      "step": 319
    },
    {
      "epoch": 0.1540961897309336,
      "grad_norm": 1.96725332736969,
      "learning_rate": 0.00019213222575138522,
      "loss": 1.1378413438796997,
      "step": 320
    },
    {
      "epoch": 0.15457774032384278,
      "grad_norm": 2.020521640777588,
      "learning_rate": 0.00019207146569067435,
      "loss": 0.7666545510292053,
      "step": 321
    },
    {
      "epoch": 0.15505929091675194,
      "grad_norm": 1.3753012418746948,
      "learning_rate": 0.00019201048160016838,
      "loss": 1.1327338218688965,
      "step": 322
    },
    {
      "epoch": 0.1555408415096611,
      "grad_norm": 2.4989731311798096,
      "learning_rate": 0.00019194927362825478,
      "loss": 0.8492980599403381,
      "step": 323
    },
    {
      "epoch": 0.15602239210257027,
      "grad_norm": 3.0465755462646484,
      "learning_rate": 0.00019188784192386587,
      "loss": 0.7442624568939209,
      "step": 324
    },
    {
      "epoch": 0.15650394269547943,
      "grad_norm": 2.4828391075134277,
      "learning_rate": 0.00019182618663647817,
      "loss": 1.0356868505477905,
      "step": 325
    },
    {
      "epoch": 0.15698549328838862,
      "grad_norm": 3.2842905521392822,
      "learning_rate": 0.0001917643079161124,
      "loss": 0.8602641820907593,
      "step": 326
    },
    {
      "epoch": 0.15746704388129779,
      "grad_norm": 2.0933339595794678,
      "learning_rate": 0.00019170220591333283,
      "loss": 0.6819586753845215,
      "step": 327
    },
    {
      "epoch": 0.15794859447420695,
      "grad_norm": 1.5177057981491089,
      "learning_rate": 0.00019163988077924713,
      "loss": 0.9294080138206482,
      "step": 328
    },
    {
      "epoch": 0.1584301450671161,
      "grad_norm": 2.429011583328247,
      "learning_rate": 0.00019157733266550575,
      "loss": 0.8477257490158081,
      "step": 329
    },
    {
      "epoch": 0.15891169566002528,
      "grad_norm": 2.3468973636627197,
      "learning_rate": 0.00019151456172430183,
      "loss": 0.7324464321136475,
      "step": 330
    },
    {
      "epoch": 0.15939324625293444,
      "grad_norm": 3.0082180500030518,
      "learning_rate": 0.0001914515681083707,
      "loss": 1.1032524108886719,
      "step": 331
    },
    {
      "epoch": 0.1598747968458436,
      "grad_norm": 3.5652754306793213,
      "learning_rate": 0.00019138835197098937,
      "loss": 0.8267630934715271,
      "step": 332
    },
    {
      "epoch": 0.1603563474387528,
      "grad_norm": 5.371067523956299,
      "learning_rate": 0.00019132491346597643,
      "loss": 1.1693214178085327,
      "step": 333
    },
    {
      "epoch": 0.16083789803166196,
      "grad_norm": 4.219338893890381,
      "learning_rate": 0.00019126125274769145,
      "loss": 0.894310712814331,
      "step": 334
    },
    {
      "epoch": 0.16131944862457112,
      "grad_norm": 2.796654462814331,
      "learning_rate": 0.00019119736997103476,
      "loss": 0.9514783620834351,
      "step": 335
    },
    {
      "epoch": 0.16180099921748028,
      "grad_norm": 3.1174535751342773,
      "learning_rate": 0.000191133265291447,
      "loss": 0.5392276644706726,
      "step": 336
    },
    {
      "epoch": 0.16228254981038945,
      "grad_norm": 4.966590881347656,
      "learning_rate": 0.00019106893886490864,
      "loss": 0.9238184094429016,
      "step": 337
    },
    {
      "epoch": 0.1627641004032986,
      "grad_norm": 3.403562068939209,
      "learning_rate": 0.00019100439084793989,
      "loss": 0.5528252124786377,
      "step": 338
    },
    {
      "epoch": 0.1632456509962078,
      "grad_norm": 4.771368980407715,
      "learning_rate": 0.00019093962139759998,
      "loss": 1.440360188484192,
      "step": 339
    },
    {
      "epoch": 0.16372720158911697,
      "grad_norm": 3.6521782875061035,
      "learning_rate": 0.000190874630671487,
      "loss": 1.167242407798767,
      "step": 340
    },
    {
      "epoch": 0.16420875218202613,
      "grad_norm": 2.595914125442505,
      "learning_rate": 0.00019080941882773745,
      "loss": 1.070678472518921,
      "step": 341
    },
    {
      "epoch": 0.1646903027749353,
      "grad_norm": 3.491302490234375,
      "learning_rate": 0.00019074398602502584,
      "loss": 1.02469801902771,
      "step": 342
    },
    {
      "epoch": 0.16517185336784446,
      "grad_norm": 2.4540822505950928,
      "learning_rate": 0.00019067833242256442,
      "loss": 1.1467150449752808,
      "step": 343
    },
    {
      "epoch": 0.16565340396075362,
      "grad_norm": 5.662190914154053,
      "learning_rate": 0.0001906124581801025,
      "loss": 0.649979293346405,
      "step": 344
    },
    {
      "epoch": 0.16613495455366278,
      "grad_norm": 2.3737704753875732,
      "learning_rate": 0.0001905463634579264,
      "loss": 0.6767232418060303,
      "step": 345
    },
    {
      "epoch": 0.16661650514657197,
      "grad_norm": 2.475189685821533,
      "learning_rate": 0.00019048004841685888,
      "loss": 1.279947280883789,
      "step": 346
    },
    {
      "epoch": 0.16709805573948114,
      "grad_norm": 1.7649701833724976,
      "learning_rate": 0.00019041351321825883,
      "loss": 1.043629765510559,
      "step": 347
    },
    {
      "epoch": 0.1675796063323903,
      "grad_norm": 4.496171474456787,
      "learning_rate": 0.00019034675802402068,
      "loss": 0.767302930355072,
      "step": 348
    },
    {
      "epoch": 0.16806115692529947,
      "grad_norm": 3.629096746444702,
      "learning_rate": 0.00019027978299657436,
      "loss": 0.7926378846168518,
      "step": 349
    },
    {
      "epoch": 0.16854270751820863,
      "grad_norm": 2.9156858921051025,
      "learning_rate": 0.00019021258829888456,
      "loss": 0.7250959873199463,
      "step": 350
    },
    {
      "epoch": 0.1690242581111178,
      "grad_norm": 3.121065139770508,
      "learning_rate": 0.00019014517409445052,
      "loss": 1.032824158668518,
      "step": 351
    },
    {
      "epoch": 0.16950580870402696,
      "grad_norm": 3.3483574390411377,
      "learning_rate": 0.00019007754054730554,
      "loss": 1.1369686126708984,
      "step": 352
    },
    {
      "epoch": 0.16998735929693615,
      "grad_norm": 3.044898748397827,
      "learning_rate": 0.00019000968782201675,
      "loss": 1.1780654191970825,
      "step": 353
    },
    {
      "epoch": 0.1704689098898453,
      "grad_norm": 1.7006185054779053,
      "learning_rate": 0.00018994161608368448,
      "loss": 0.9185819625854492,
      "step": 354
    },
    {
      "epoch": 0.17095046048275447,
      "grad_norm": 8.086901664733887,
      "learning_rate": 0.00018987332549794196,
      "loss": 1.2629584074020386,
      "step": 355
    },
    {
      "epoch": 0.17143201107566364,
      "grad_norm": 4.051202297210693,
      "learning_rate": 0.00018980481623095502,
      "loss": 0.7351322770118713,
      "step": 356
    },
    {
      "epoch": 0.1719135616685728,
      "grad_norm": 2.582331657409668,
      "learning_rate": 0.00018973608844942148,
      "loss": 0.8742402195930481,
      "step": 357
    },
    {
      "epoch": 0.17239511226148196,
      "grad_norm": 1.7577705383300781,
      "learning_rate": 0.00018966714232057094,
      "loss": 0.8065980672836304,
      "step": 358
    },
    {
      "epoch": 0.17287666285439113,
      "grad_norm": 3.919654607772827,
      "learning_rate": 0.00018959797801216418,
      "loss": 0.7381652593612671,
      "step": 359
    },
    {
      "epoch": 0.17335821344730032,
      "grad_norm": 2.5199310779571533,
      "learning_rate": 0.000189528595692493,
      "loss": 0.6921659111976624,
      "step": 360
    },
    {
      "epoch": 0.17383976404020948,
      "grad_norm": 1.6235740184783936,
      "learning_rate": 0.00018945899553037956,
      "loss": 0.44717878103256226,
      "step": 361
    },
    {
      "epoch": 0.17432131463311865,
      "grad_norm": 2.755976676940918,
      "learning_rate": 0.00018938917769517613,
      "loss": 0.9652116298675537,
      "step": 362
    },
    {
      "epoch": 0.1748028652260278,
      "grad_norm": 2.8015854358673096,
      "learning_rate": 0.00018931914235676458,
      "loss": 0.9502769708633423,
      "step": 363
    },
    {
      "epoch": 0.17528441581893697,
      "grad_norm": 2.762962579727173,
      "learning_rate": 0.00018924888968555606,
      "loss": 1.1041167974472046,
      "step": 364
    },
    {
      "epoch": 0.17576596641184614,
      "grad_norm": 4.228902816772461,
      "learning_rate": 0.00018917841985249055,
      "loss": 0.9911640882492065,
      "step": 365
    },
    {
      "epoch": 0.1762475170047553,
      "grad_norm": 2.94317889213562,
      "learning_rate": 0.0001891077330290363,
      "loss": 0.9620732069015503,
      "step": 366
    },
    {
      "epoch": 0.1767290675976645,
      "grad_norm": 2.2522852420806885,
      "learning_rate": 0.00018903682938718977,
      "loss": 1.1403533220291138,
      "step": 367
    },
    {
      "epoch": 0.17721061819057365,
      "grad_norm": 3.204599142074585,
      "learning_rate": 0.00018896570909947475,
      "loss": 0.8627252578735352,
      "step": 368
    },
    {
      "epoch": 0.17769216878348282,
      "grad_norm": 1.7018917798995972,
      "learning_rate": 0.00018889437233894234,
      "loss": 0.6692804098129272,
      "step": 369
    },
    {
      "epoch": 0.17817371937639198,
      "grad_norm": 1.8896235227584839,
      "learning_rate": 0.0001888228192791703,
      "loss": 1.1914817094802856,
      "step": 370
    },
    {
      "epoch": 0.17865526996930114,
      "grad_norm": 2.928189277648926,
      "learning_rate": 0.00018875105009426272,
      "loss": 0.9321286678314209,
      "step": 371
    },
    {
      "epoch": 0.1791368205622103,
      "grad_norm": 1.7436562776565552,
      "learning_rate": 0.00018867906495884955,
      "loss": 1.2526813745498657,
      "step": 372
    },
    {
      "epoch": 0.17961837115511947,
      "grad_norm": 1.5223957300186157,
      "learning_rate": 0.0001886068640480862,
      "loss": 1.1435872316360474,
      "step": 373
    },
    {
      "epoch": 0.18009992174802866,
      "grad_norm": 2.2813076972961426,
      "learning_rate": 0.00018853444753765306,
      "loss": 0.8662838339805603,
      "step": 374
    },
    {
      "epoch": 0.18058147234093783,
      "grad_norm": 3.9945993423461914,
      "learning_rate": 0.00018846181560375525,
      "loss": 1.0233099460601807,
      "step": 375
    },
    {
      "epoch": 0.181063022933847,
      "grad_norm": 3.089719533920288,
      "learning_rate": 0.0001883889684231219,
      "loss": 0.8876188397407532,
      "step": 376
    },
    {
      "epoch": 0.18154457352675615,
      "grad_norm": 3.4020583629608154,
      "learning_rate": 0.000188315906173006,
      "loss": 0.9251139163970947,
      "step": 377
    },
    {
      "epoch": 0.18202612411966532,
      "grad_norm": 3.6036579608917236,
      "learning_rate": 0.0001882426290311838,
      "loss": 1.047196626663208,
      "step": 378
    },
    {
      "epoch": 0.18250767471257448,
      "grad_norm": 12.380807876586914,
      "learning_rate": 0.00018816913717595445,
      "loss": 0.9222043752670288,
      "step": 379
    },
    {
      "epoch": 0.18298922530548364,
      "grad_norm": 2.7900030612945557,
      "learning_rate": 0.00018809543078613953,
      "loss": 0.7460460066795349,
      "step": 380
    },
    {
      "epoch": 0.18347077589839283,
      "grad_norm": 5.577341556549072,
      "learning_rate": 0.00018802151004108263,
      "loss": 0.7508425712585449,
      "step": 381
    },
    {
      "epoch": 0.183952326491302,
      "grad_norm": 3.3597304821014404,
      "learning_rate": 0.0001879473751206489,
      "loss": 0.9825782775878906,
      "step": 382
    },
    {
      "epoch": 0.18443387708421116,
      "grad_norm": 2.534180164337158,
      "learning_rate": 0.00018787302620522467,
      "loss": 1.2303452491760254,
      "step": 383
    },
    {
      "epoch": 0.18491542767712033,
      "grad_norm": 5.105559349060059,
      "learning_rate": 0.00018779846347571693,
      "loss": 0.9238927960395813,
      "step": 384
    },
    {
      "epoch": 0.1853969782700295,
      "grad_norm": 3.356008768081665,
      "learning_rate": 0.0001877236871135529,
      "loss": 0.8858113288879395,
      "step": 385
    },
    {
      "epoch": 0.18587852886293865,
      "grad_norm": 4.741906642913818,
      "learning_rate": 0.00018764869730067968,
      "loss": 1.3930648565292358,
      "step": 386
    },
    {
      "epoch": 0.18636007945584784,
      "grad_norm": 6.8427019119262695,
      "learning_rate": 0.0001875734942195637,
      "loss": 0.9347327351570129,
      "step": 387
    },
    {
      "epoch": 0.186841630048757,
      "grad_norm": 2.6932623386383057,
      "learning_rate": 0.0001874980780531903,
      "loss": 0.9895850419998169,
      "step": 388
    },
    {
      "epoch": 0.18732318064166617,
      "grad_norm": 4.764650821685791,
      "learning_rate": 0.00018742244898506337,
      "loss": 0.937369704246521,
      "step": 389
    },
    {
      "epoch": 0.18780473123457533,
      "grad_norm": 7.2434258460998535,
      "learning_rate": 0.00018734660719920475,
      "loss": 1.2479074001312256,
      "step": 390
    },
    {
      "epoch": 0.1882862818274845,
      "grad_norm": 4.188095569610596,
      "learning_rate": 0.00018727055288015397,
      "loss": 1.1375575065612793,
      "step": 391
    },
    {
      "epoch": 0.18876783242039366,
      "grad_norm": 1.2517567873001099,
      "learning_rate": 0.00018719428621296764,
      "loss": 1.0726561546325684,
      "step": 392
    },
    {
      "epoch": 0.18924938301330282,
      "grad_norm": 4.715351104736328,
      "learning_rate": 0.00018711780738321897,
      "loss": 0.7725263833999634,
      "step": 393
    },
    {
      "epoch": 0.18973093360621202,
      "grad_norm": 3.4153072834014893,
      "learning_rate": 0.00018704111657699758,
      "loss": 1.187920331954956,
      "step": 394
    },
    {
      "epoch": 0.19021248419912118,
      "grad_norm": 1.4115175008773804,
      "learning_rate": 0.0001869642139809088,
      "loss": 0.7903649210929871,
      "step": 395
    },
    {
      "epoch": 0.19069403479203034,
      "grad_norm": 1.6498994827270508,
      "learning_rate": 0.00018688709978207323,
      "loss": 1.111178994178772,
      "step": 396
    },
    {
      "epoch": 0.1911755853849395,
      "grad_norm": 1.8135826587677002,
      "learning_rate": 0.00018680977416812644,
      "loss": 0.9596853256225586,
      "step": 397
    },
    {
      "epoch": 0.19165713597784867,
      "grad_norm": 3.5349509716033936,
      "learning_rate": 0.00018673223732721837,
      "loss": 0.4664667844772339,
      "step": 398
    },
    {
      "epoch": 0.19213868657075783,
      "grad_norm": 3.148221015930176,
      "learning_rate": 0.0001866544894480129,
      "loss": 1.0165011882781982,
      "step": 399
    },
    {
      "epoch": 0.192620237163667,
      "grad_norm": 4.141859531402588,
      "learning_rate": 0.00018657653071968747,
      "loss": 0.875876247882843,
      "step": 400
    },
    {
      "epoch": 0.1931017877565762,
      "grad_norm": 4.675681114196777,
      "learning_rate": 0.00018649836133193253,
      "loss": 1.14305579662323,
      "step": 401
    },
    {
      "epoch": 0.19358333834948535,
      "grad_norm": 5.166753768920898,
      "learning_rate": 0.00018641998147495112,
      "loss": 0.9850279092788696,
      "step": 402
    },
    {
      "epoch": 0.19406488894239451,
      "grad_norm": 6.032193183898926,
      "learning_rate": 0.00018634139133945837,
      "loss": 0.9787382483482361,
      "step": 403
    },
    {
      "epoch": 0.19454643953530368,
      "grad_norm": 2.2948343753814697,
      "learning_rate": 0.00018626259111668105,
      "loss": 1.340905785560608,
      "step": 404
    },
    {
      "epoch": 0.19502799012821284,
      "grad_norm": 7.3895745277404785,
      "learning_rate": 0.00018618358099835723,
      "loss": 1.1400458812713623,
      "step": 405
    },
    {
      "epoch": 0.195509540721122,
      "grad_norm": 5.889215469360352,
      "learning_rate": 0.00018610436117673555,
      "loss": 0.8577663898468018,
      "step": 406
    },
    {
      "epoch": 0.19599109131403117,
      "grad_norm": 4.308293342590332,
      "learning_rate": 0.00018602493184457505,
      "loss": 0.786826491355896,
      "step": 407
    },
    {
      "epoch": 0.19647264190694036,
      "grad_norm": 5.4372944831848145,
      "learning_rate": 0.00018594529319514437,
      "loss": 1.1811593770980835,
      "step": 408
    },
    {
      "epoch": 0.19695419249984952,
      "grad_norm": 5.17181921005249,
      "learning_rate": 0.00018586544542222169,
      "loss": 0.925212025642395,
      "step": 409
    },
    {
      "epoch": 0.1974357430927587,
      "grad_norm": 7.996733665466309,
      "learning_rate": 0.00018578538872009384,
      "loss": 0.7122504711151123,
      "step": 410
    },
    {
      "epoch": 0.19791729368566785,
      "grad_norm": 6.0105767250061035,
      "learning_rate": 0.00018570512328355612,
      "loss": 1.1835112571716309,
      "step": 411
    },
    {
      "epoch": 0.198398844278577,
      "grad_norm": 2.4882612228393555,
      "learning_rate": 0.00018562464930791167,
      "loss": 0.8593431115150452,
      "step": 412
    },
    {
      "epoch": 0.19888039487148618,
      "grad_norm": 3.435121774673462,
      "learning_rate": 0.00018554396698897116,
      "loss": 0.8544062972068787,
      "step": 413
    },
    {
      "epoch": 0.19936194546439534,
      "grad_norm": 2.3508243560791016,
      "learning_rate": 0.00018546307652305205,
      "loss": 0.884438157081604,
      "step": 414
    },
    {
      "epoch": 0.19984349605730453,
      "grad_norm": 3.229066848754883,
      "learning_rate": 0.00018538197810697842,
      "loss": 0.8022018671035767,
      "step": 415
    },
    {
      "epoch": 0.2003250466502137,
      "grad_norm": 3.673863172531128,
      "learning_rate": 0.0001853006719380802,
      "loss": 1.3524482250213623,
      "step": 416
    },
    {
      "epoch": 0.20080659724312286,
      "grad_norm": 4.1495041847229,
      "learning_rate": 0.00018521915821419284,
      "loss": 1.0579551458358765,
      "step": 417
    },
    {
      "epoch": 0.20128814783603202,
      "grad_norm": 1.7399744987487793,
      "learning_rate": 0.00018513743713365698,
      "loss": 0.7650774121284485,
      "step": 418
    },
    {
      "epoch": 0.20176969842894119,
      "grad_norm": 5.127481937408447,
      "learning_rate": 0.00018505550889531765,
      "loss": 1.2347795963287354,
      "step": 419
    },
    {
      "epoch": 0.20225124902185035,
      "grad_norm": 2.609527111053467,
      "learning_rate": 0.00018497337369852395,
      "loss": 0.5715798139572144,
      "step": 420
    },
    {
      "epoch": 0.2027327996147595,
      "grad_norm": 5.215476036071777,
      "learning_rate": 0.0001848910317431286,
      "loss": 1.0233427286148071,
      "step": 421
    },
    {
      "epoch": 0.2032143502076687,
      "grad_norm": 1.9264320135116577,
      "learning_rate": 0.00018480848322948739,
      "loss": 0.8668956756591797,
      "step": 422
    },
    {
      "epoch": 0.20369590080057787,
      "grad_norm": 3.321080446243286,
      "learning_rate": 0.00018472572835845873,
      "loss": 0.9470916390419006,
      "step": 423
    },
    {
      "epoch": 0.20417745139348703,
      "grad_norm": 2.84110951423645,
      "learning_rate": 0.00018464276733140306,
      "loss": 1.067541480064392,
      "step": 424
    },
    {
      "epoch": 0.2046590019863962,
      "grad_norm": 3.7342779636383057,
      "learning_rate": 0.0001845596003501826,
      "loss": 0.8175060153007507,
      "step": 425
    },
    {
      "epoch": 0.20514055257930536,
      "grad_norm": 5.291472911834717,
      "learning_rate": 0.00018447622761716057,
      "loss": 0.8132691383361816,
      "step": 426
    },
    {
      "epoch": 0.20562210317221452,
      "grad_norm": 4.226396083831787,
      "learning_rate": 0.00018439264933520084,
      "loss": 0.9199362993240356,
      "step": 427
    },
    {
      "epoch": 0.2061036537651237,
      "grad_norm": 2.670567512512207,
      "learning_rate": 0.00018430886570766747,
      "loss": 1.212700366973877,
      "step": 428
    },
    {
      "epoch": 0.20658520435803288,
      "grad_norm": 2.095062017440796,
      "learning_rate": 0.0001842248769384242,
      "loss": 1.2095063924789429,
      "step": 429
    },
    {
      "epoch": 0.20706675495094204,
      "grad_norm": 3.2593092918395996,
      "learning_rate": 0.00018414068323183375,
      "loss": 1.074259877204895,
      "step": 430
    },
    {
      "epoch": 0.2075483055438512,
      "grad_norm": 1.959027886390686,
      "learning_rate": 0.00018405628479275775,
      "loss": 0.7466518878936768,
      "step": 431
    },
    {
      "epoch": 0.20802985613676037,
      "grad_norm": 3.249171257019043,
      "learning_rate": 0.00018397168182655583,
      "loss": 1.675467848777771,
      "step": 432
    },
    {
      "epoch": 0.20851140672966953,
      "grad_norm": 5.385195255279541,
      "learning_rate": 0.00018388687453908527,
      "loss": 1.0001357793807983,
      "step": 433
    },
    {
      "epoch": 0.2089929573225787,
      "grad_norm": 1.986000657081604,
      "learning_rate": 0.00018380186313670058,
      "loss": 0.8847096562385559,
      "step": 434
    },
    {
      "epoch": 0.20947450791548788,
      "grad_norm": 3.0317087173461914,
      "learning_rate": 0.00018371664782625287,
      "loss": 0.9060726165771484,
      "step": 435
    },
    {
      "epoch": 0.20995605850839705,
      "grad_norm": 0.7143993377685547,
      "learning_rate": 0.00018363122881508945,
      "loss": 0.6094214916229248,
      "step": 436
    },
    {
      "epoch": 0.2104376091013062,
      "grad_norm": 3.502746105194092,
      "learning_rate": 0.00018354560631105328,
      "loss": 0.4233100116252899,
      "step": 437
    },
    {
      "epoch": 0.21091915969421537,
      "grad_norm": 2.000012159347534,
      "learning_rate": 0.00018345978052248233,
      "loss": 0.8414865136146545,
      "step": 438
    },
    {
      "epoch": 0.21140071028712454,
      "grad_norm": 6.736745834350586,
      "learning_rate": 0.00018337375165820944,
      "loss": 0.7155567407608032,
      "step": 439
    },
    {
      "epoch": 0.2118822608800337,
      "grad_norm": 3.919236898422241,
      "learning_rate": 0.00018328751992756137,
      "loss": 0.5850315690040588,
      "step": 440
    },
    {
      "epoch": 0.21236381147294286,
      "grad_norm": 2.2598280906677246,
      "learning_rate": 0.0001832010855403586,
      "loss": 0.8949121236801147,
      "step": 441
    },
    {
      "epoch": 0.21284536206585206,
      "grad_norm": 5.957046985626221,
      "learning_rate": 0.0001831144487069147,
      "loss": 1.146686315536499,
      "step": 442
    },
    {
      "epoch": 0.21332691265876122,
      "grad_norm": 2.0824103355407715,
      "learning_rate": 0.0001830276096380358,
      "loss": 0.5526815056800842,
      "step": 443
    },
    {
      "epoch": 0.21380846325167038,
      "grad_norm": 2.6616811752319336,
      "learning_rate": 0.0001829405685450202,
      "loss": 1.0309299230575562,
      "step": 444
    },
    {
      "epoch": 0.21429001384457955,
      "grad_norm": 4.55696439743042,
      "learning_rate": 0.00018285332563965765,
      "loss": 1.2182782888412476,
      "step": 445
    },
    {
      "epoch": 0.2147715644374887,
      "grad_norm": 1.69126558303833,
      "learning_rate": 0.00018276588113422905,
      "loss": 1.0237442255020142,
      "step": 446
    },
    {
      "epoch": 0.21525311503039787,
      "grad_norm": 2.9522948265075684,
      "learning_rate": 0.00018267823524150575,
      "loss": 0.9743610620498657,
      "step": 447
    },
    {
      "epoch": 0.21573466562330704,
      "grad_norm": 2.191713571548462,
      "learning_rate": 0.00018259038817474923,
      "loss": 0.5438602566719055,
      "step": 448
    },
    {
      "epoch": 0.21621621621621623,
      "grad_norm": 2.9654388427734375,
      "learning_rate": 0.0001825023401477104,
      "loss": 0.7293569445610046,
      "step": 449
    },
    {
      "epoch": 0.2166977668091254,
      "grad_norm": 2.520054817199707,
      "learning_rate": 0.0001824140913746291,
      "loss": 0.7990195155143738,
      "step": 450
    },
    {
      "epoch": 0.21717931740203456,
      "grad_norm": 3.2978196144104004,
      "learning_rate": 0.00018232564207023376,
      "loss": 0.742237389087677,
      "step": 451
    },
    {
      "epoch": 0.21766086799494372,
      "grad_norm": 2.1745803356170654,
      "learning_rate": 0.00018223699244974064,
      "loss": 1.256987452507019,
      "step": 452
    },
    {
      "epoch": 0.21814241858785288,
      "grad_norm": 2.249605894088745,
      "learning_rate": 0.00018214814272885343,
      "loss": 1.2342674732208252,
      "step": 453
    },
    {
      "epoch": 0.21862396918076205,
      "grad_norm": 7.406956672668457,
      "learning_rate": 0.00018205909312376276,
      "loss": 1.2494183778762817,
      "step": 454
    },
    {
      "epoch": 0.2191055197736712,
      "grad_norm": 2.4870083332061768,
      "learning_rate": 0.00018196984385114554,
      "loss": 0.9075199961662292,
      "step": 455
    },
    {
      "epoch": 0.2195870703665804,
      "grad_norm": 3.083622455596924,
      "learning_rate": 0.0001818803951281646,
      "loss": 1.2009893655776978,
      "step": 456
    },
    {
      "epoch": 0.22006862095948956,
      "grad_norm": 2.7718398571014404,
      "learning_rate": 0.000181790747172468,
      "loss": 0.6137069463729858,
      "step": 457
    },
    {
      "epoch": 0.22055017155239873,
      "grad_norm": 4.348511695861816,
      "learning_rate": 0.00018170090020218864,
      "loss": 1.3975932598114014,
      "step": 458
    },
    {
      "epoch": 0.2210317221453079,
      "grad_norm": 3.559659481048584,
      "learning_rate": 0.00018161085443594365,
      "loss": 0.7908211946487427,
      "step": 459
    },
    {
      "epoch": 0.22151327273821705,
      "grad_norm": 4.958198070526123,
      "learning_rate": 0.00018152061009283382,
      "loss": 1.3539707660675049,
      "step": 460
    },
    {
      "epoch": 0.22199482333112622,
      "grad_norm": 4.90092658996582,
      "learning_rate": 0.00018143016739244314,
      "loss": 1.0203726291656494,
      "step": 461
    },
    {
      "epoch": 0.22247637392403538,
      "grad_norm": 5.3405022621154785,
      "learning_rate": 0.0001813395265548383,
      "loss": 1.0012438297271729,
      "step": 462
    },
    {
      "epoch": 0.22295792451694457,
      "grad_norm": 6.6246113777160645,
      "learning_rate": 0.00018124868780056814,
      "loss": 1.1316394805908203,
      "step": 463
    },
    {
      "epoch": 0.22343947510985374,
      "grad_norm": 3.0194599628448486,
      "learning_rate": 0.0001811576513506629,
      "loss": 1.0894339084625244,
      "step": 464
    },
    {
      "epoch": 0.2239210257027629,
      "grad_norm": 2.100529670715332,
      "learning_rate": 0.00018106641742663397,
      "loss": 0.7964451909065247,
      "step": 465
    },
    {
      "epoch": 0.22440257629567206,
      "grad_norm": 2.441472053527832,
      "learning_rate": 0.00018097498625047328,
      "loss": 1.2086806297302246,
      "step": 466
    },
    {
      "epoch": 0.22488412688858123,
      "grad_norm": 3.9360191822052,
      "learning_rate": 0.00018088335804465258,
      "loss": 0.562084436416626,
      "step": 467
    },
    {
      "epoch": 0.2253656774814904,
      "grad_norm": 2.7165207862854004,
      "learning_rate": 0.00018079153303212318,
      "loss": 0.8732426762580872,
      "step": 468
    },
    {
      "epoch": 0.22584722807439955,
      "grad_norm": 2.0756189823150635,
      "learning_rate": 0.0001806995114363152,
      "loss": 1.0946354866027832,
      "step": 469
    },
    {
      "epoch": 0.22632877866730874,
      "grad_norm": 4.449759006500244,
      "learning_rate": 0.00018060729348113707,
      "loss": 1.1055773496627808,
      "step": 470
    },
    {
      "epoch": 0.2268103292602179,
      "grad_norm": 3.1937973499298096,
      "learning_rate": 0.00018051487939097505,
      "loss": 0.9541581869125366,
      "step": 471
    },
    {
      "epoch": 0.22729187985312707,
      "grad_norm": 1.5221508741378784,
      "learning_rate": 0.00018042226939069255,
      "loss": 0.7801059484481812,
      "step": 472
    },
    {
      "epoch": 0.22777343044603623,
      "grad_norm": 4.808314800262451,
      "learning_rate": 0.00018032946370562982,
      "loss": 1.3691374063491821,
      "step": 473
    },
    {
      "epoch": 0.2282549810389454,
      "grad_norm": 1.832607626914978,
      "learning_rate": 0.00018023646256160313,
      "loss": 0.7540324330329895,
      "step": 474
    },
    {
      "epoch": 0.22873653163185456,
      "grad_norm": 2.928666114807129,
      "learning_rate": 0.00018014326618490437,
      "loss": 0.8740776777267456,
      "step": 475
    },
    {
      "epoch": 0.22921808222476375,
      "grad_norm": 2.3927767276763916,
      "learning_rate": 0.0001800498748023005,
      "loss": 0.6442964673042297,
      "step": 476
    },
    {
      "epoch": 0.22969963281767292,
      "grad_norm": 2.4679973125457764,
      "learning_rate": 0.000179956288641033,
      "loss": 0.8856695890426636,
      "step": 477
    },
    {
      "epoch": 0.23018118341058208,
      "grad_norm": 4.338985919952393,
      "learning_rate": 0.00017986250792881718,
      "loss": 1.1790205240249634,
      "step": 478
    },
    {
      "epoch": 0.23066273400349124,
      "grad_norm": 2.226858377456665,
      "learning_rate": 0.00017976853289384184,
      "loss": 0.930794358253479,
      "step": 479
    },
    {
      "epoch": 0.2311442845964004,
      "grad_norm": 3.0856661796569824,
      "learning_rate": 0.00017967436376476855,
      "loss": 1.3886624574661255,
      "step": 480
    },
    {
      "epoch": 0.23162583518930957,
      "grad_norm": 4.037444114685059,
      "learning_rate": 0.0001795800007707312,
      "loss": 0.826001763343811,
      "step": 481
    },
    {
      "epoch": 0.23210738578221873,
      "grad_norm": 1.366046667098999,
      "learning_rate": 0.00017948544414133534,
      "loss": 0.9147120118141174,
      "step": 482
    },
    {
      "epoch": 0.23258893637512792,
      "grad_norm": 4.575778007507324,
      "learning_rate": 0.00017939069410665773,
      "loss": 0.9754892587661743,
      "step": 483
    },
    {
      "epoch": 0.2330704869680371,
      "grad_norm": 2.1202187538146973,
      "learning_rate": 0.0001792957508972457,
      "loss": 0.9992613196372986,
      "step": 484
    },
    {
      "epoch": 0.23355203756094625,
      "grad_norm": 4.352765083312988,
      "learning_rate": 0.00017920061474411658,
      "loss": 1.2638614177703857,
      "step": 485
    },
    {
      "epoch": 0.23403358815385542,
      "grad_norm": 2.24794340133667,
      "learning_rate": 0.00017910528587875729,
      "loss": 1.342864990234375,
      "step": 486
    },
    {
      "epoch": 0.23451513874676458,
      "grad_norm": 1.6007968187332153,
      "learning_rate": 0.00017900976453312352,
      "loss": 0.8669201135635376,
      "step": 487
    },
    {
      "epoch": 0.23499668933967374,
      "grad_norm": 3.301929473876953,
      "learning_rate": 0.00017891405093963938,
      "loss": 1.0607353448867798,
      "step": 488
    },
    {
      "epoch": 0.2354782399325829,
      "grad_norm": 1.9019057750701904,
      "learning_rate": 0.00017881814533119675,
      "loss": 1.0270676612854004,
      "step": 489
    },
    {
      "epoch": 0.2359597905254921,
      "grad_norm": 1.7171289920806885,
      "learning_rate": 0.00017872204794115474,
      "loss": 0.7128303647041321,
      "step": 490
    },
    {
      "epoch": 0.23644134111840126,
      "grad_norm": 1.949903964996338,
      "learning_rate": 0.0001786257590033391,
      "loss": 0.8728585243225098,
      "step": 491
    },
    {
      "epoch": 0.23692289171131042,
      "grad_norm": 2.132519245147705,
      "learning_rate": 0.00017852927875204163,
      "loss": 1.3150029182434082,
      "step": 492
    },
    {
      "epoch": 0.2374044423042196,
      "grad_norm": 2.0490305423736572,
      "learning_rate": 0.00017843260742201963,
      "loss": 1.0350804328918457,
      "step": 493
    },
    {
      "epoch": 0.23788599289712875,
      "grad_norm": 3.5623679161071777,
      "learning_rate": 0.00017833574524849535,
      "loss": 1.2923754453659058,
      "step": 494
    },
    {
      "epoch": 0.23836754349003791,
      "grad_norm": 1.9029139280319214,
      "learning_rate": 0.00017823869246715553,
      "loss": 0.7151404023170471,
      "step": 495
    },
    {
      "epoch": 0.23884909408294708,
      "grad_norm": 4.28120756149292,
      "learning_rate": 0.00017814144931415043,
      "loss": 0.5412241220474243,
      "step": 496
    },
    {
      "epoch": 0.23933064467585627,
      "grad_norm": 2.1949639320373535,
      "learning_rate": 0.0001780440160260938,
      "loss": 0.8384115695953369,
      "step": 497
    },
    {
      "epoch": 0.23981219526876543,
      "grad_norm": 3.5483477115631104,
      "learning_rate": 0.00017794639284006184,
      "loss": 1.1149206161499023,
      "step": 498
    },
    {
      "epoch": 0.2402937458616746,
      "grad_norm": 2.4343881607055664,
      "learning_rate": 0.0001778485799935929,
      "loss": 1.0757761001586914,
      "step": 499
    },
    {
      "epoch": 0.24077529645458376,
      "grad_norm": 2.2507190704345703,
      "learning_rate": 0.00017775057772468679,
      "loss": 0.6096641421318054,
      "step": 500
    },
    {
      "epoch": 0.24125684704749292,
      "grad_norm": 4.638617992401123,
      "learning_rate": 0.00017765238627180424,
      "loss": 1.4337159395217896,
      "step": 501
    },
    {
      "epoch": 0.2417383976404021,
      "grad_norm": 1.8858293294906616,
      "learning_rate": 0.00017755400587386632,
      "loss": 0.8175569772720337,
      "step": 502
    },
    {
      "epoch": 0.24221994823331125,
      "grad_norm": 2.6646251678466797,
      "learning_rate": 0.00017745543677025378,
      "loss": 0.9485897421836853,
      "step": 503
    },
    {
      "epoch": 0.24270149882622044,
      "grad_norm": 2.0465831756591797,
      "learning_rate": 0.00017735667920080661,
      "loss": 0.7836700677871704,
      "step": 504
    },
    {
      "epoch": 0.2431830494191296,
      "grad_norm": 4.2461018562316895,
      "learning_rate": 0.0001772577334058233,
      "loss": 1.0598890781402588,
      "step": 505
    },
    {
      "epoch": 0.24366460001203877,
      "grad_norm": 2.6852498054504395,
      "learning_rate": 0.00017715859962606043,
      "loss": 1.039440631866455,
      "step": 506
    },
    {
      "epoch": 0.24414615060494793,
      "grad_norm": 3.3891656398773193,
      "learning_rate": 0.00017705927810273187,
      "loss": 1.1060519218444824,
      "step": 507
    },
    {
      "epoch": 0.2446277011978571,
      "grad_norm": 3.0936930179595947,
      "learning_rate": 0.00017695976907750844,
      "loss": 1.083831548690796,
      "step": 508
    },
    {
      "epoch": 0.24510925179076626,
      "grad_norm": 3.3679087162017822,
      "learning_rate": 0.00017686007279251706,
      "loss": 0.905409038066864,
      "step": 509
    },
    {
      "epoch": 0.24559080238367542,
      "grad_norm": 2.108914375305176,
      "learning_rate": 0.00017676018949034045,
      "loss": 0.9287945032119751,
      "step": 510
    },
    {
      "epoch": 0.2460723529765846,
      "grad_norm": 6.319317817687988,
      "learning_rate": 0.0001766601194140162,
      "loss": 0.9913032054901123,
      "step": 511
    },
    {
      "epoch": 0.24655390356949378,
      "grad_norm": 2.255915641784668,
      "learning_rate": 0.0001765598628070365,
      "loss": 1.0533208847045898,
      "step": 512
    },
    {
      "epoch": 0.24703545416240294,
      "grad_norm": 2.5953805446624756,
      "learning_rate": 0.00017645941991334732,
      "loss": 0.9860719442367554,
      "step": 513
    },
    {
      "epoch": 0.2475170047553121,
      "grad_norm": 3.232588529586792,
      "learning_rate": 0.00017635879097734804,
      "loss": 0.9936854839324951,
      "step": 514
    },
    {
      "epoch": 0.24799855534822127,
      "grad_norm": 2.25575590133667,
      "learning_rate": 0.00017625797624389055,
      "loss": 0.6766625046730042,
      "step": 515
    },
    {
      "epoch": 0.24848010594113043,
      "grad_norm": 1.6869364976882935,
      "learning_rate": 0.00017615697595827897,
      "loss": 0.6579121947288513,
      "step": 516
    },
    {
      "epoch": 0.2489616565340396,
      "grad_norm": 2.9416613578796387,
      "learning_rate": 0.0001760557903662688,
      "loss": 0.8866895437240601,
      "step": 517
    },
    {
      "epoch": 0.24944320712694878,
      "grad_norm": 2.0137009620666504,
      "learning_rate": 0.00017595441971406648,
      "loss": 0.8526488542556763,
      "step": 518
    },
    {
      "epoch": 0.24992475771985795,
      "grad_norm": 3.513070821762085,
      "learning_rate": 0.00017585286424832874,
      "loss": 0.9100205898284912,
      "step": 519
    },
    {
      "epoch": 0.2504063083127671,
      "grad_norm": 3.3302173614501953,
      "learning_rate": 0.00017575112421616202,
      "loss": 0.9262621402740479,
      "step": 520
    },
    {
      "epoch": 0.2508878589056763,
      "grad_norm": 6.412067890167236,
      "learning_rate": 0.0001756491998651218,
      "loss": 0.7214961051940918,
      "step": 521
    },
    {
      "epoch": 0.25136940949858544,
      "grad_norm": 1.6448354721069336,
      "learning_rate": 0.0001755470914432121,
      "loss": 0.7495888471603394,
      "step": 522
    },
    {
      "epoch": 0.2518509600914946,
      "grad_norm": 5.05918025970459,
      "learning_rate": 0.0001754447991988848,
      "loss": 0.7324005365371704,
      "step": 523
    },
    {
      "epoch": 0.25233251068440377,
      "grad_norm": 1.781410574913025,
      "learning_rate": 0.00017534232338103903,
      "loss": 1.312711238861084,
      "step": 524
    },
    {
      "epoch": 0.25281406127731293,
      "grad_norm": 1.4993844032287598,
      "learning_rate": 0.0001752396642390207,
      "loss": 1.1856259107589722,
      "step": 525
    },
    {
      "epoch": 0.2532956118702221,
      "grad_norm": 1.2658802270889282,
      "learning_rate": 0.00017513682202262163,
      "loss": 0.8794026970863342,
      "step": 526
    },
    {
      "epoch": 0.25377716246313126,
      "grad_norm": 2.454000949859619,
      "learning_rate": 0.00017503379698207918,
      "loss": 1.0167148113250732,
      "step": 527
    },
    {
      "epoch": 0.2542587130560405,
      "grad_norm": 1.565911889076233,
      "learning_rate": 0.00017493058936807562,
      "loss": 0.753157377243042,
      "step": 528
    },
    {
      "epoch": 0.25474026364894964,
      "grad_norm": 3.6118555068969727,
      "learning_rate": 0.00017482719943173739,
      "loss": 1.2428447008132935,
      "step": 529
    },
    {
      "epoch": 0.2552218142418588,
      "grad_norm": 3.4746713638305664,
      "learning_rate": 0.00017472362742463455,
      "loss": 1.241966962814331,
      "step": 530
    },
    {
      "epoch": 0.25570336483476797,
      "grad_norm": 2.8696980476379395,
      "learning_rate": 0.0001746198735987802,
      "loss": 0.9110194444656372,
      "step": 531
    },
    {
      "epoch": 0.25618491542767713,
      "grad_norm": 4.085539817810059,
      "learning_rate": 0.00017451593820662988,
      "loss": 1.1040695905685425,
      "step": 532
    },
    {
      "epoch": 0.2566664660205863,
      "grad_norm": 10.014312744140625,
      "learning_rate": 0.00017441182150108086,
      "loss": 1.4271295070648193,
      "step": 533
    },
    {
      "epoch": 0.25714801661349546,
      "grad_norm": 4.465886116027832,
      "learning_rate": 0.0001743075237354716,
      "loss": 1.784240484237671,
      "step": 534
    },
    {
      "epoch": 0.2576295672064046,
      "grad_norm": 3.7951693534851074,
      "learning_rate": 0.00017420304516358113,
      "loss": 0.6552119255065918,
      "step": 535
    },
    {
      "epoch": 0.2581111177993138,
      "grad_norm": 2.8169782161712646,
      "learning_rate": 0.00017409838603962843,
      "loss": 0.6203606724739075,
      "step": 536
    },
    {
      "epoch": 0.25859266839222295,
      "grad_norm": 1.9252712726593018,
      "learning_rate": 0.00017399354661827178,
      "loss": 0.8554360866546631,
      "step": 537
    },
    {
      "epoch": 0.2590742189851321,
      "grad_norm": 7.761801719665527,
      "learning_rate": 0.00017388852715460819,
      "loss": 0.9287173748016357,
      "step": 538
    },
    {
      "epoch": 0.2595557695780413,
      "grad_norm": 6.011171340942383,
      "learning_rate": 0.00017378332790417273,
      "loss": 1.0620510578155518,
      "step": 539
    },
    {
      "epoch": 0.26003732017095044,
      "grad_norm": 0.9864084720611572,
      "learning_rate": 0.00017367794912293794,
      "loss": 1.0016095638275146,
      "step": 540
    },
    {
      "epoch": 0.26051887076385966,
      "grad_norm": 2.130699872970581,
      "learning_rate": 0.00017357239106731317,
      "loss": 0.5015271902084351,
      "step": 541
    },
    {
      "epoch": 0.2610004213567688,
      "grad_norm": 3.4187889099121094,
      "learning_rate": 0.00017346665399414405,
      "loss": 0.8691253662109375,
      "step": 542
    },
    {
      "epoch": 0.261481971949678,
      "grad_norm": 5.106690406799316,
      "learning_rate": 0.00017336073816071168,
      "loss": 1.043741226196289,
      "step": 543
    },
    {
      "epoch": 0.26196352254258715,
      "grad_norm": 2.392275333404541,
      "learning_rate": 0.00017325464382473226,
      "loss": 0.9952304363250732,
      "step": 544
    },
    {
      "epoch": 0.2624450731354963,
      "grad_norm": 3.087329387664795,
      "learning_rate": 0.00017314837124435622,
      "loss": 1.0388630628585815,
      "step": 545
    },
    {
      "epoch": 0.2629266237284055,
      "grad_norm": 2.757615089416504,
      "learning_rate": 0.00017304192067816782,
      "loss": 1.237568974494934,
      "step": 546
    },
    {
      "epoch": 0.26340817432131464,
      "grad_norm": 2.01505184173584,
      "learning_rate": 0.00017293529238518422,
      "loss": 0.4551701545715332,
      "step": 547
    },
    {
      "epoch": 0.2638897249142238,
      "grad_norm": 0.8415576815605164,
      "learning_rate": 0.0001728284866248552,
      "loss": 1.1316872835159302,
      "step": 548
    },
    {
      "epoch": 0.26437127550713296,
      "grad_norm": 3.1021790504455566,
      "learning_rate": 0.00017272150365706224,
      "loss": 1.084308385848999,
      "step": 549
    },
    {
      "epoch": 0.2648528261000421,
      "grad_norm": 2.9192192554473877,
      "learning_rate": 0.00017261434374211802,
      "loss": 0.5167235732078552,
      "step": 550
    },
    {
      "epoch": 0.2653343766929513,
      "grad_norm": 2.305537223815918,
      "learning_rate": 0.00017250700714076586,
      "loss": 0.6387326717376709,
      "step": 551
    },
    {
      "epoch": 0.26581592728586045,
      "grad_norm": 2.565230131149292,
      "learning_rate": 0.00017239949411417888,
      "loss": 1.1593327522277832,
      "step": 552
    },
    {
      "epoch": 0.2662974778787696,
      "grad_norm": 1.9840683937072754,
      "learning_rate": 0.0001722918049239596,
      "loss": 1.0889431238174438,
      "step": 553
    },
    {
      "epoch": 0.2667790284716788,
      "grad_norm": 3.8681797981262207,
      "learning_rate": 0.00017218393983213902,
      "loss": 0.8686015605926514,
      "step": 554
    },
    {
      "epoch": 0.267260579064588,
      "grad_norm": 2.0458152294158936,
      "learning_rate": 0.00017207589910117634,
      "loss": 0.64968341588974,
      "step": 555
    },
    {
      "epoch": 0.26774212965749716,
      "grad_norm": 3.3043510913848877,
      "learning_rate": 0.00017196768299395797,
      "loss": 0.8026936054229736,
      "step": 556
    },
    {
      "epoch": 0.2682236802504063,
      "grad_norm": 2.270799160003662,
      "learning_rate": 0.00017185929177379714,
      "loss": 0.6929470896720886,
      "step": 557
    },
    {
      "epoch": 0.2687052308433155,
      "grad_norm": 3.7142252922058105,
      "learning_rate": 0.00017175072570443312,
      "loss": 0.9828073978424072,
      "step": 558
    },
    {
      "epoch": 0.26918678143622465,
      "grad_norm": 2.959670066833496,
      "learning_rate": 0.00017164198505003066,
      "loss": 0.7629191279411316,
      "step": 559
    },
    {
      "epoch": 0.2696683320291338,
      "grad_norm": 1.6875691413879395,
      "learning_rate": 0.0001715330700751793,
      "loss": 0.6313432455062866,
      "step": 560
    },
    {
      "epoch": 0.270149882622043,
      "grad_norm": 2.1431081295013428,
      "learning_rate": 0.00017142398104489273,
      "loss": 0.8511427044868469,
      "step": 561
    },
    {
      "epoch": 0.27063143321495214,
      "grad_norm": 3.3747332096099854,
      "learning_rate": 0.00017131471822460814,
      "loss": 0.8130661845207214,
      "step": 562
    },
    {
      "epoch": 0.2711129838078613,
      "grad_norm": 1.629927158355713,
      "learning_rate": 0.00017120528188018565,
      "loss": 0.7882002592086792,
      "step": 563
    },
    {
      "epoch": 0.27159453440077047,
      "grad_norm": 2.728206157684326,
      "learning_rate": 0.00017109567227790754,
      "loss": 0.7358757257461548,
      "step": 564
    },
    {
      "epoch": 0.27207608499367963,
      "grad_norm": 2.5146825313568115,
      "learning_rate": 0.00017098588968447766,
      "loss": 1.142038106918335,
      "step": 565
    },
    {
      "epoch": 0.2725576355865888,
      "grad_norm": 3.4829790592193604,
      "learning_rate": 0.00017087593436702084,
      "loss": 1.349970817565918,
      "step": 566
    },
    {
      "epoch": 0.27303918617949796,
      "grad_norm": 3.340740203857422,
      "learning_rate": 0.00017076580659308222,
      "loss": 1.1955182552337646,
      "step": 567
    },
    {
      "epoch": 0.2735207367724071,
      "grad_norm": 3.2698421478271484,
      "learning_rate": 0.00017065550663062634,
      "loss": 0.9397417902946472,
      "step": 568
    },
    {
      "epoch": 0.27400228736531634,
      "grad_norm": 2.8796544075012207,
      "learning_rate": 0.00017054503474803702,
      "loss": 1.193009376525879,
      "step": 569
    },
    {
      "epoch": 0.2744838379582255,
      "grad_norm": 1.6001664400100708,
      "learning_rate": 0.00017043439121411618,
      "loss": 0.8073057532310486,
      "step": 570
    },
    {
      "epoch": 0.27496538855113467,
      "grad_norm": 3.098938465118408,
      "learning_rate": 0.0001703235762980835,
      "loss": 1.004915475845337,
      "step": 571
    },
    {
      "epoch": 0.27544693914404383,
      "grad_norm": 3.2004833221435547,
      "learning_rate": 0.00017021259026957567,
      "loss": 0.7345186471939087,
      "step": 572
    },
    {
      "epoch": 0.275928489736953,
      "grad_norm": 2.907801389694214,
      "learning_rate": 0.00017010143339864562,
      "loss": 1.1764411926269531,
      "step": 573
    },
    {
      "epoch": 0.27641004032986216,
      "grad_norm": 2.134838104248047,
      "learning_rate": 0.0001699901059557621,
      "loss": 0.9856343865394592,
      "step": 574
    },
    {
      "epoch": 0.2768915909227713,
      "grad_norm": 2.353640079498291,
      "learning_rate": 0.00016987860821180895,
      "loss": 0.8849102854728699,
      "step": 575
    },
    {
      "epoch": 0.2773731415156805,
      "grad_norm": 2.95815372467041,
      "learning_rate": 0.00016976694043808416,
      "loss": 1.000349998474121,
      "step": 576
    },
    {
      "epoch": 0.27785469210858965,
      "grad_norm": 3.7722842693328857,
      "learning_rate": 0.00016965510290629972,
      "loss": 0.989434003829956,
      "step": 577
    },
    {
      "epoch": 0.2783362427014988,
      "grad_norm": 1.2988595962524414,
      "learning_rate": 0.00016954309588858044,
      "loss": 0.7240885496139526,
      "step": 578
    },
    {
      "epoch": 0.278817793294408,
      "grad_norm": 1.5012184381484985,
      "learning_rate": 0.00016943091965746366,
      "loss": 0.644542396068573,
      "step": 579
    },
    {
      "epoch": 0.27929934388731714,
      "grad_norm": 1.3767204284667969,
      "learning_rate": 0.00016931857448589845,
      "loss": 0.9001684784889221,
      "step": 580
    },
    {
      "epoch": 0.2797808944802263,
      "grad_norm": 1.916168451309204,
      "learning_rate": 0.00016920606064724488,
      "loss": 1.1123192310333252,
      "step": 581
    },
    {
      "epoch": 0.28026244507313547,
      "grad_norm": 2.622476100921631,
      "learning_rate": 0.00016909337841527344,
      "loss": 1.3137260675430298,
      "step": 582
    },
    {
      "epoch": 0.2807439956660447,
      "grad_norm": 1.898782730102539,
      "learning_rate": 0.00016898052806416444,
      "loss": 0.824989378452301,
      "step": 583
    },
    {
      "epoch": 0.28122554625895385,
      "grad_norm": 1.9352258443832397,
      "learning_rate": 0.00016886750986850718,
      "loss": 0.6913671493530273,
      "step": 584
    },
    {
      "epoch": 0.281707096851863,
      "grad_norm": 1.5414141416549683,
      "learning_rate": 0.00016875432410329934,
      "loss": 0.7089320421218872,
      "step": 585
    },
    {
      "epoch": 0.2821886474447722,
      "grad_norm": 1.9833709001541138,
      "learning_rate": 0.0001686409710439464,
      "loss": 0.6255727410316467,
      "step": 586
    },
    {
      "epoch": 0.28267019803768134,
      "grad_norm": 1.212068796157837,
      "learning_rate": 0.00016852745096626088,
      "loss": 0.6841402053833008,
      "step": 587
    },
    {
      "epoch": 0.2831517486305905,
      "grad_norm": 2.65910267829895,
      "learning_rate": 0.0001684137641464617,
      "loss": 1.1762453317642212,
      "step": 588
    },
    {
      "epoch": 0.28363329922349967,
      "grad_norm": 2.28123140335083,
      "learning_rate": 0.0001682999108611735,
      "loss": 0.8498421907424927,
      "step": 589
    },
    {
      "epoch": 0.28411484981640883,
      "grad_norm": 2.357839584350586,
      "learning_rate": 0.00016818589138742587,
      "loss": 0.7982231378555298,
      "step": 590
    },
    {
      "epoch": 0.284596400409318,
      "grad_norm": 5.155858993530273,
      "learning_rate": 0.00016807170600265296,
      "loss": 1.6413533687591553,
      "step": 591
    },
    {
      "epoch": 0.28507795100222716,
      "grad_norm": 1.6892406940460205,
      "learning_rate": 0.00016795735498469246,
      "loss": 0.7634009122848511,
      "step": 592
    },
    {
      "epoch": 0.2855595015951363,
      "grad_norm": 2.249336004257202,
      "learning_rate": 0.00016784283861178513,
      "loss": 0.8350369334220886,
      "step": 593
    },
    {
      "epoch": 0.2860410521880455,
      "grad_norm": 2.56176495552063,
      "learning_rate": 0.00016772815716257412,
      "loss": 0.9931415319442749,
      "step": 594
    },
    {
      "epoch": 0.28652260278095465,
      "grad_norm": 2.7509541511535645,
      "learning_rate": 0.00016761331091610416,
      "loss": 1.0643393993377686,
      "step": 595
    },
    {
      "epoch": 0.28700415337386387,
      "grad_norm": 2.5772926807403564,
      "learning_rate": 0.00016749830015182107,
      "loss": 0.5202586650848389,
      "step": 596
    },
    {
      "epoch": 0.28748570396677303,
      "grad_norm": 1.130454421043396,
      "learning_rate": 0.00016738312514957086,
      "loss": 0.6347820162773132,
      "step": 597
    },
    {
      "epoch": 0.2879672545596822,
      "grad_norm": 2.8816230297088623,
      "learning_rate": 0.00016726778618959926,
      "loss": 0.9898222088813782,
      "step": 598
    },
    {
      "epoch": 0.28844880515259136,
      "grad_norm": 2.428179979324341,
      "learning_rate": 0.00016715228355255093,
      "loss": 1.071276307106018,
      "step": 599
    },
    {
      "epoch": 0.2889303557455005,
      "grad_norm": 2.8337578773498535,
      "learning_rate": 0.00016703661751946874,
      "loss": 1.4577418565750122,
      "step": 600
    },
    {
      "epoch": 0.2894119063384097,
      "grad_norm": 3.0683395862579346,
      "learning_rate": 0.00016692078837179318,
      "loss": 1.4638723134994507,
      "step": 601
    },
    {
      "epoch": 0.28989345693131885,
      "grad_norm": 3.20438814163208,
      "learning_rate": 0.00016680479639136163,
      "loss": 1.150207757949829,
      "step": 602
    },
    {
      "epoch": 0.290375007524228,
      "grad_norm": 2.6135637760162354,
      "learning_rate": 0.0001666886418604077,
      "loss": 0.6226499080657959,
      "step": 603
    },
    {
      "epoch": 0.2908565581171372,
      "grad_norm": 4.191404342651367,
      "learning_rate": 0.0001665723250615604,
      "loss": 0.9844186305999756,
      "step": 604
    },
    {
      "epoch": 0.29133810871004634,
      "grad_norm": 3.192216396331787,
      "learning_rate": 0.00016645584627784381,
      "loss": 0.8181465268135071,
      "step": 605
    },
    {
      "epoch": 0.2918196593029555,
      "grad_norm": 3.006706714630127,
      "learning_rate": 0.0001663392057926759,
      "loss": 0.8875517845153809,
      "step": 606
    },
    {
      "epoch": 0.29230120989586467,
      "grad_norm": 3.8680756092071533,
      "learning_rate": 0.00016622240388986824,
      "loss": 0.9057378768920898,
      "step": 607
    },
    {
      "epoch": 0.29278276048877383,
      "grad_norm": 2.380160093307495,
      "learning_rate": 0.0001661054408536251,
      "loss": 1.6692523956298828,
      "step": 608
    },
    {
      "epoch": 0.293264311081683,
      "grad_norm": 2.7110695838928223,
      "learning_rate": 0.00016598831696854288,
      "loss": 0.8386815786361694,
      "step": 609
    },
    {
      "epoch": 0.2937458616745922,
      "grad_norm": 1.761985421180725,
      "learning_rate": 0.00016587103251960937,
      "loss": 0.7568842172622681,
      "step": 610
    },
    {
      "epoch": 0.2942274122675014,
      "grad_norm": 1.612274408340454,
      "learning_rate": 0.00016575358779220294,
      "loss": 0.6250135898590088,
      "step": 611
    },
    {
      "epoch": 0.29470896286041054,
      "grad_norm": 1.8893933296203613,
      "learning_rate": 0.00016563598307209204,
      "loss": 0.6475261449813843,
      "step": 612
    },
    {
      "epoch": 0.2951905134533197,
      "grad_norm": 2.8482584953308105,
      "learning_rate": 0.0001655182186454344,
      "loss": 0.8087155222892761,
      "step": 613
    },
    {
      "epoch": 0.29567206404622887,
      "grad_norm": 2.366551399230957,
      "learning_rate": 0.00016540029479877638,
      "loss": 0.7850953340530396,
      "step": 614
    },
    {
      "epoch": 0.29615361463913803,
      "grad_norm": 8.164939880371094,
      "learning_rate": 0.00016528221181905217,
      "loss": 1.076558232307434,
      "step": 615
    },
    {
      "epoch": 0.2966351652320472,
      "grad_norm": 1.8576300144195557,
      "learning_rate": 0.00016516396999358322,
      "loss": 0.9149329662322998,
      "step": 616
    },
    {
      "epoch": 0.29711671582495636,
      "grad_norm": 2.657912492752075,
      "learning_rate": 0.00016504556961007748,
      "loss": 0.9277459383010864,
      "step": 617
    },
    {
      "epoch": 0.2975982664178655,
      "grad_norm": 3.7927987575531006,
      "learning_rate": 0.00016492701095662866,
      "loss": 1.134142279624939,
      "step": 618
    },
    {
      "epoch": 0.2980798170107747,
      "grad_norm": 4.51692533493042,
      "learning_rate": 0.00016480829432171564,
      "loss": 0.8797689080238342,
      "step": 619
    },
    {
      "epoch": 0.29856136760368385,
      "grad_norm": 2.478346347808838,
      "learning_rate": 0.0001646894199942017,
      "loss": 0.6071556210517883,
      "step": 620
    },
    {
      "epoch": 0.299042918196593,
      "grad_norm": 2.1819894313812256,
      "learning_rate": 0.0001645703882633338,
      "loss": 1.0739173889160156,
      "step": 621
    },
    {
      "epoch": 0.2995244687895022,
      "grad_norm": 2.650787830352783,
      "learning_rate": 0.00016445119941874183,
      "loss": 0.7531375885009766,
      "step": 622
    },
    {
      "epoch": 0.30000601938241134,
      "grad_norm": 2.552433729171753,
      "learning_rate": 0.00016433185375043809,
      "loss": 0.9724037647247314,
      "step": 623
    },
    {
      "epoch": 0.30048756997532056,
      "grad_norm": 2.4958653450012207,
      "learning_rate": 0.00016421235154881638,
      "loss": 1.0594429969787598,
      "step": 624
    },
    {
      "epoch": 0.3009691205682297,
      "grad_norm": 1.87485671043396,
      "learning_rate": 0.00016409269310465146,
      "loss": 0.37132173776626587,
      "step": 625
    },
    {
      "epoch": 0.3014506711611389,
      "grad_norm": 2.2897658348083496,
      "learning_rate": 0.00016397287870909813,
      "loss": 0.8453037142753601,
      "step": 626
    },
    {
      "epoch": 0.30193222175404805,
      "grad_norm": 1.6302165985107422,
      "learning_rate": 0.00016385290865369079,
      "loss": 0.9922266602516174,
      "step": 627
    },
    {
      "epoch": 0.3024137723469572,
      "grad_norm": 5.031513214111328,
      "learning_rate": 0.00016373278323034255,
      "loss": 1.457411289215088,
      "step": 628
    },
    {
      "epoch": 0.3028953229398664,
      "grad_norm": 3.348341941833496,
      "learning_rate": 0.0001636125027313445,
      "loss": 0.801249086856842,
      "step": 629
    },
    {
      "epoch": 0.30337687353277554,
      "grad_norm": 2.86871337890625,
      "learning_rate": 0.00016349206744936518,
      "loss": 1.0821988582611084,
      "step": 630
    },
    {
      "epoch": 0.3038584241256847,
      "grad_norm": 2.2004575729370117,
      "learning_rate": 0.00016337147767744967,
      "loss": 1.0259231328964233,
      "step": 631
    },
    {
      "epoch": 0.30433997471859386,
      "grad_norm": 3.6990625858306885,
      "learning_rate": 0.0001632507337090189,
      "loss": 0.7700485587120056,
      "step": 632
    },
    {
      "epoch": 0.304821525311503,
      "grad_norm": 1.8415513038635254,
      "learning_rate": 0.0001631298358378692,
      "loss": 0.9068533182144165,
      "step": 633
    },
    {
      "epoch": 0.3053030759044122,
      "grad_norm": 2.650070905685425,
      "learning_rate": 0.00016300878435817113,
      "loss": 0.9891302585601807,
      "step": 634
    },
    {
      "epoch": 0.30578462649732135,
      "grad_norm": 2.9380407333374023,
      "learning_rate": 0.00016288757956446918,
      "loss": 1.0874038934707642,
      "step": 635
    },
    {
      "epoch": 0.3062661770902305,
      "grad_norm": 2.3508448600769043,
      "learning_rate": 0.00016276622175168083,
      "loss": 0.7192472815513611,
      "step": 636
    },
    {
      "epoch": 0.30674772768313974,
      "grad_norm": 1.7921020984649658,
      "learning_rate": 0.0001626447112150959,
      "loss": 1.236085295677185,
      "step": 637
    },
    {
      "epoch": 0.3072292782760489,
      "grad_norm": 1.971126675605774,
      "learning_rate": 0.00016252304825037576,
      "loss": 1.1444514989852905,
      "step": 638
    },
    {
      "epoch": 0.30771082886895806,
      "grad_norm": 3.100534677505493,
      "learning_rate": 0.0001624012331535528,
      "loss": 1.5551820993423462,
      "step": 639
    },
    {
      "epoch": 0.3081923794618672,
      "grad_norm": 1.2525639533996582,
      "learning_rate": 0.00016227926622102947,
      "loss": 0.8078435063362122,
      "step": 640
    },
    {
      "epoch": 0.3086739300547764,
      "grad_norm": 1.682363748550415,
      "learning_rate": 0.00016215714774957772,
      "loss": 0.9083467721939087,
      "step": 641
    },
    {
      "epoch": 0.30915548064768555,
      "grad_norm": 2.5103561878204346,
      "learning_rate": 0.00016203487803633822,
      "loss": 1.243340015411377,
      "step": 642
    },
    {
      "epoch": 0.3096370312405947,
      "grad_norm": 1.6590416431427002,
      "learning_rate": 0.00016191245737881956,
      "loss": 0.8234121799468994,
      "step": 643
    },
    {
      "epoch": 0.3101185818335039,
      "grad_norm": 4.016539573669434,
      "learning_rate": 0.00016178988607489777,
      "loss": 0.9046227931976318,
      "step": 644
    },
    {
      "epoch": 0.31060013242641304,
      "grad_norm": 2.99137282371521,
      "learning_rate": 0.00016166716442281528,
      "loss": 1.1642231941223145,
      "step": 645
    },
    {
      "epoch": 0.3110816830193222,
      "grad_norm": 3.5799002647399902,
      "learning_rate": 0.0001615442927211805,
      "loss": 0.9058552980422974,
      "step": 646
    },
    {
      "epoch": 0.31156323361223137,
      "grad_norm": 2.403265953063965,
      "learning_rate": 0.0001614212712689668,
      "loss": 1.0792720317840576,
      "step": 647
    },
    {
      "epoch": 0.31204478420514054,
      "grad_norm": 3.1600358486175537,
      "learning_rate": 0.00016129810036551198,
      "loss": 0.782172679901123,
      "step": 648
    },
    {
      "epoch": 0.3125263347980497,
      "grad_norm": 2.1347131729125977,
      "learning_rate": 0.00016117478031051755,
      "loss": 0.8863410353660583,
      "step": 649
    },
    {
      "epoch": 0.31300788539095886,
      "grad_norm": 5.152075290679932,
      "learning_rate": 0.00016105131140404787,
      "loss": 1.144633173942566,
      "step": 650
    },
    {
      "epoch": 0.3134894359838681,
      "grad_norm": 3.3020546436309814,
      "learning_rate": 0.00016092769394652947,
      "loss": 0.9027894139289856,
      "step": 651
    },
    {
      "epoch": 0.31397098657677724,
      "grad_norm": 3.2763631343841553,
      "learning_rate": 0.0001608039282387504,
      "loss": 0.7677838802337646,
      "step": 652
    },
    {
      "epoch": 0.3144525371696864,
      "grad_norm": 3.521336317062378,
      "learning_rate": 0.00016068001458185936,
      "loss": 1.2670947313308716,
      "step": 653
    },
    {
      "epoch": 0.31493408776259557,
      "grad_norm": 2.533520460128784,
      "learning_rate": 0.0001605559532773651,
      "loss": 0.8032466173171997,
      "step": 654
    },
    {
      "epoch": 0.31541563835550473,
      "grad_norm": 2.3603177070617676,
      "learning_rate": 0.00016043174462713566,
      "loss": 0.7805054187774658,
      "step": 655
    },
    {
      "epoch": 0.3158971889484139,
      "grad_norm": 2.912759780883789,
      "learning_rate": 0.00016030738893339753,
      "loss": 0.8116629123687744,
      "step": 656
    },
    {
      "epoch": 0.31637873954132306,
      "grad_norm": 2.510319709777832,
      "learning_rate": 0.00016018288649873497,
      "loss": 1.2192778587341309,
      "step": 657
    },
    {
      "epoch": 0.3168602901342322,
      "grad_norm": 2.3298234939575195,
      "learning_rate": 0.0001600582376260894,
      "loss": 0.4386475384235382,
      "step": 658
    },
    {
      "epoch": 0.3173418407271414,
      "grad_norm": 1.8634427785873413,
      "learning_rate": 0.00015993344261875847,
      "loss": 0.8052042126655579,
      "step": 659
    },
    {
      "epoch": 0.31782339132005055,
      "grad_norm": 3.112504720687866,
      "learning_rate": 0.00015980850178039547,
      "loss": 1.2988989353179932,
      "step": 660
    },
    {
      "epoch": 0.3183049419129597,
      "grad_norm": 2.100111484527588,
      "learning_rate": 0.00015968341541500842,
      "loss": 1.2103468179702759,
      "step": 661
    },
    {
      "epoch": 0.3187864925058689,
      "grad_norm": 1.5729902982711792,
      "learning_rate": 0.00015955818382695953,
      "loss": 0.6760388016700745,
      "step": 662
    },
    {
      "epoch": 0.31926804309877804,
      "grad_norm": 5.739004611968994,
      "learning_rate": 0.00015943280732096438,
      "loss": 0.7272723317146301,
      "step": 663
    },
    {
      "epoch": 0.3197495936916872,
      "grad_norm": 4.136562347412109,
      "learning_rate": 0.00015930728620209113,
      "loss": 1.1547679901123047,
      "step": 664
    },
    {
      "epoch": 0.3202311442845964,
      "grad_norm": 3.1759958267211914,
      "learning_rate": 0.00015918162077575976,
      "loss": 0.969549834728241,
      "step": 665
    },
    {
      "epoch": 0.3207126948775056,
      "grad_norm": 1.8113259077072144,
      "learning_rate": 0.00015905581134774153,
      "loss": 0.5392419695854187,
      "step": 666
    },
    {
      "epoch": 0.32119424547041475,
      "grad_norm": 2.3568577766418457,
      "learning_rate": 0.0001589298582241579,
      "loss": 1.3548619747161865,
      "step": 667
    },
    {
      "epoch": 0.3216757960633239,
      "grad_norm": 2.805354118347168,
      "learning_rate": 0.00015880376171148014,
      "loss": 1.150697946548462,
      "step": 668
    },
    {
      "epoch": 0.3221573466562331,
      "grad_norm": 3.0976903438568115,
      "learning_rate": 0.00015867752211652831,
      "loss": 0.4871942102909088,
      "step": 669
    },
    {
      "epoch": 0.32263889724914224,
      "grad_norm": 2.2747292518615723,
      "learning_rate": 0.00015855113974647068,
      "loss": 1.1653554439544678,
      "step": 670
    },
    {
      "epoch": 0.3231204478420514,
      "grad_norm": 1.2680680751800537,
      "learning_rate": 0.0001584246149088229,
      "loss": 0.6154179573059082,
      "step": 671
    },
    {
      "epoch": 0.32360199843496057,
      "grad_norm": 4.875119686126709,
      "learning_rate": 0.0001582979479114472,
      "loss": 0.7686120271682739,
      "step": 672
    },
    {
      "epoch": 0.32408354902786973,
      "grad_norm": 2.1111087799072266,
      "learning_rate": 0.0001581711390625519,
      "loss": 0.7352918982505798,
      "step": 673
    },
    {
      "epoch": 0.3245650996207789,
      "grad_norm": 2.1917271614074707,
      "learning_rate": 0.0001580441886706903,
      "loss": 0.523643970489502,
      "step": 674
    },
    {
      "epoch": 0.32504665021368806,
      "grad_norm": 3.474445343017578,
      "learning_rate": 0.00015791709704476015,
      "loss": 0.6348187327384949,
      "step": 675
    },
    {
      "epoch": 0.3255282008065972,
      "grad_norm": 1.6397799253463745,
      "learning_rate": 0.00015778986449400292,
      "loss": 0.6338925361633301,
      "step": 676
    },
    {
      "epoch": 0.3260097513995064,
      "grad_norm": 2.691070318222046,
      "learning_rate": 0.00015766249132800292,
      "loss": 0.6127196550369263,
      "step": 677
    },
    {
      "epoch": 0.3264913019924156,
      "grad_norm": 3.417691707611084,
      "learning_rate": 0.00015753497785668663,
      "loss": 0.6740036010742188,
      "step": 678
    },
    {
      "epoch": 0.32697285258532477,
      "grad_norm": 2.264055013656616,
      "learning_rate": 0.00015740732439032187,
      "loss": 1.0195417404174805,
      "step": 679
    },
    {
      "epoch": 0.32745440317823393,
      "grad_norm": 1.5161832571029663,
      "learning_rate": 0.00015727953123951716,
      "loss": 1.2092632055282593,
      "step": 680
    },
    {
      "epoch": 0.3279359537711431,
      "grad_norm": 1.5340229272842407,
      "learning_rate": 0.00015715159871522086,
      "loss": 0.437242329120636,
      "step": 681
    },
    {
      "epoch": 0.32841750436405226,
      "grad_norm": 5.840272426605225,
      "learning_rate": 0.00015702352712872056,
      "loss": 1.0941039323806763,
      "step": 682
    },
    {
      "epoch": 0.3288990549569614,
      "grad_norm": 2.0571744441986084,
      "learning_rate": 0.00015689531679164204,
      "loss": 1.1421433687210083,
      "step": 683
    },
    {
      "epoch": 0.3293806055498706,
      "grad_norm": 2.8197414875030518,
      "learning_rate": 0.00015676696801594886,
      "loss": 0.9986647963523865,
      "step": 684
    },
    {
      "epoch": 0.32986215614277975,
      "grad_norm": 2.4561047554016113,
      "learning_rate": 0.00015663848111394132,
      "loss": 1.3996104001998901,
      "step": 685
    },
    {
      "epoch": 0.3303437067356889,
      "grad_norm": 3.0542616844177246,
      "learning_rate": 0.00015650985639825585,
      "loss": 1.1885985136032104,
      "step": 686
    },
    {
      "epoch": 0.3308252573285981,
      "grad_norm": 3.789811134338379,
      "learning_rate": 0.00015638109418186424,
      "loss": 0.8562176823616028,
      "step": 687
    },
    {
      "epoch": 0.33130680792150724,
      "grad_norm": 1.5541062355041504,
      "learning_rate": 0.00015625219477807277,
      "loss": 0.8571232557296753,
      "step": 688
    },
    {
      "epoch": 0.3317883585144164,
      "grad_norm": 1.705386757850647,
      "learning_rate": 0.00015612315850052166,
      "loss": 0.9700482487678528,
      "step": 689
    },
    {
      "epoch": 0.33226990910732557,
      "grad_norm": 1.890159010887146,
      "learning_rate": 0.00015599398566318396,
      "loss": 0.6123092174530029,
      "step": 690
    },
    {
      "epoch": 0.33275145970023473,
      "grad_norm": 7.227348327636719,
      "learning_rate": 0.00015586467658036524,
      "loss": 0.9032357931137085,
      "step": 691
    },
    {
      "epoch": 0.33323301029314395,
      "grad_norm": 4.403627872467041,
      "learning_rate": 0.00015573523156670244,
      "loss": 0.7850395441055298,
      "step": 692
    },
    {
      "epoch": 0.3337145608860531,
      "grad_norm": 1.3737635612487793,
      "learning_rate": 0.0001556056509371633,
      "loss": 0.5132907032966614,
      "step": 693
    },
    {
      "epoch": 0.3341961114789623,
      "grad_norm": 1.0324561595916748,
      "learning_rate": 0.00015547593500704547,
      "loss": 0.93122798204422,
      "step": 694
    },
    {
      "epoch": 0.33467766207187144,
      "grad_norm": 1.1416678428649902,
      "learning_rate": 0.00015534608409197592,
      "loss": 0.6370548605918884,
      "step": 695
    },
    {
      "epoch": 0.3351592126647806,
      "grad_norm": 1.9818695783615112,
      "learning_rate": 0.00015521609850791004,
      "loss": 0.7577373385429382,
      "step": 696
    },
    {
      "epoch": 0.33564076325768977,
      "grad_norm": 2.411644458770752,
      "learning_rate": 0.0001550859785711308,
      "loss": 0.8775813579559326,
      "step": 697
    },
    {
      "epoch": 0.33612231385059893,
      "grad_norm": 1.5806745290756226,
      "learning_rate": 0.0001549557245982482,
      "loss": 0.7956851124763489,
      "step": 698
    },
    {
      "epoch": 0.3366038644435081,
      "grad_norm": 2.270986795425415,
      "learning_rate": 0.00015482533690619837,
      "loss": 0.691323459148407,
      "step": 699
    },
    {
      "epoch": 0.33708541503641726,
      "grad_norm": 1.6080154180526733,
      "learning_rate": 0.00015469481581224272,
      "loss": 0.7039358615875244,
      "step": 700
    },
    {
      "epoch": 0.3375669656293264,
      "grad_norm": 1.3919873237609863,
      "learning_rate": 0.0001545641616339673,
      "loss": 0.8251152038574219,
      "step": 701
    },
    {
      "epoch": 0.3380485162222356,
      "grad_norm": 1.937422752380371,
      "learning_rate": 0.00015443337468928206,
      "loss": 0.8872036933898926,
      "step": 702
    },
    {
      "epoch": 0.33853006681514475,
      "grad_norm": 2.56326961517334,
      "learning_rate": 0.00015430245529641986,
      "loss": 1.1989351511001587,
      "step": 703
    },
    {
      "epoch": 0.3390116174080539,
      "grad_norm": 2.1591696739196777,
      "learning_rate": 0.00015417140377393596,
      "loss": 0.9199816584587097,
      "step": 704
    },
    {
      "epoch": 0.3394931680009631,
      "grad_norm": 1.6416661739349365,
      "learning_rate": 0.00015404022044070704,
      "loss": 0.6598629951477051,
      "step": 705
    },
    {
      "epoch": 0.3399747185938723,
      "grad_norm": 3.789112091064453,
      "learning_rate": 0.00015390890561593052,
      "loss": 1.118451714515686,
      "step": 706
    },
    {
      "epoch": 0.34045626918678146,
      "grad_norm": 2.1995675563812256,
      "learning_rate": 0.0001537774596191238,
      "loss": 0.6980578303337097,
      "step": 707
    },
    {
      "epoch": 0.3409378197796906,
      "grad_norm": 3.908986806869507,
      "learning_rate": 0.00015364588277012344,
      "loss": 1.0182504653930664,
      "step": 708
    },
    {
      "epoch": 0.3414193703725998,
      "grad_norm": 1.7687630653381348,
      "learning_rate": 0.00015351417538908435,
      "loss": 1.094681978225708,
      "step": 709
    },
    {
      "epoch": 0.34190092096550895,
      "grad_norm": 1.680133581161499,
      "learning_rate": 0.0001533823377964791,
      "loss": 0.8249967694282532,
      "step": 710
    },
    {
      "epoch": 0.3423824715584181,
      "grad_norm": 2.9655508995056152,
      "learning_rate": 0.00015325037031309704,
      "loss": 1.4249920845031738,
      "step": 711
    },
    {
      "epoch": 0.3428640221513273,
      "grad_norm": 2.510322093963623,
      "learning_rate": 0.00015311827326004363,
      "loss": 0.7630861401557922,
      "step": 712
    },
    {
      "epoch": 0.34334557274423644,
      "grad_norm": 2.184629440307617,
      "learning_rate": 0.0001529860469587396,
      "loss": 1.183720350265503,
      "step": 713
    },
    {
      "epoch": 0.3438271233371456,
      "grad_norm": 1.8966466188430786,
      "learning_rate": 0.00015285369173092015,
      "loss": 0.39592230319976807,
      "step": 714
    },
    {
      "epoch": 0.34430867393005476,
      "grad_norm": 1.1290984153747559,
      "learning_rate": 0.00015272120789863413,
      "loss": 0.6176584362983704,
      "step": 715
    },
    {
      "epoch": 0.34479022452296393,
      "grad_norm": 1.3990130424499512,
      "learning_rate": 0.00015258859578424342,
      "loss": 0.9987316131591797,
      "step": 716
    },
    {
      "epoch": 0.3452717751158731,
      "grad_norm": 2.7190723419189453,
      "learning_rate": 0.00015245585571042194,
      "loss": 0.9911836385726929,
      "step": 717
    },
    {
      "epoch": 0.34575332570878226,
      "grad_norm": 2.9956939220428467,
      "learning_rate": 0.00015232298800015506,
      "loss": 0.871444821357727,
      "step": 718
    },
    {
      "epoch": 0.3462348763016914,
      "grad_norm": 2.845107316970825,
      "learning_rate": 0.00015218999297673862,
      "loss": 1.2158775329589844,
      "step": 719
    },
    {
      "epoch": 0.34671642689460064,
      "grad_norm": 3.0271718502044678,
      "learning_rate": 0.0001520568709637783,
      "loss": 1.0125917196273804,
      "step": 720
    },
    {
      "epoch": 0.3471979774875098,
      "grad_norm": 3.51896595954895,
      "learning_rate": 0.00015192362228518875,
      "loss": 0.8885514140129089,
      "step": 721
    },
    {
      "epoch": 0.34767952808041896,
      "grad_norm": 6.444211483001709,
      "learning_rate": 0.00015179024726519284,
      "loss": 1.0924811363220215,
      "step": 722
    },
    {
      "epoch": 0.34816107867332813,
      "grad_norm": 1.879628300666809,
      "learning_rate": 0.00015165674622832085,
      "loss": 0.8389526605606079,
      "step": 723
    },
    {
      "epoch": 0.3486426292662373,
      "grad_norm": 2.240924835205078,
      "learning_rate": 0.0001515231194994097,
      "loss": 1.4966366291046143,
      "step": 724
    },
    {
      "epoch": 0.34912417985914646,
      "grad_norm": 1.5642396211624146,
      "learning_rate": 0.00015138936740360207,
      "loss": 1.036865234375,
      "step": 725
    },
    {
      "epoch": 0.3496057304520556,
      "grad_norm": 2.7182729244232178,
      "learning_rate": 0.00015125549026634585,
      "loss": 0.589577317237854,
      "step": 726
    },
    {
      "epoch": 0.3500872810449648,
      "grad_norm": 1.568490743637085,
      "learning_rate": 0.00015112148841339295,
      "loss": 0.9795326590538025,
      "step": 727
    },
    {
      "epoch": 0.35056883163787395,
      "grad_norm": 2.714752674102783,
      "learning_rate": 0.000150987362170799,
      "loss": 1.006779670715332,
      "step": 728
    },
    {
      "epoch": 0.3510503822307831,
      "grad_norm": 3.4299421310424805,
      "learning_rate": 0.00015085311186492206,
      "loss": 0.7638811469078064,
      "step": 729
    },
    {
      "epoch": 0.3515319328236923,
      "grad_norm": 2.7987523078918457,
      "learning_rate": 0.00015071873782242223,
      "loss": 1.0115736722946167,
      "step": 730
    },
    {
      "epoch": 0.35201348341660144,
      "grad_norm": 1.8808013200759888,
      "learning_rate": 0.0001505842403702606,
      "loss": 0.7189298868179321,
      "step": 731
    },
    {
      "epoch": 0.3524950340095106,
      "grad_norm": 2.015634775161743,
      "learning_rate": 0.00015044961983569856,
      "loss": 0.8953535556793213,
      "step": 732
    },
    {
      "epoch": 0.3529765846024198,
      "grad_norm": 1.5096917152404785,
      "learning_rate": 0.00015031487654629702,
      "loss": 0.7522500157356262,
      "step": 733
    },
    {
      "epoch": 0.353458135195329,
      "grad_norm": 3.803304433822632,
      "learning_rate": 0.00015018001082991553,
      "loss": 0.922589898109436,
      "step": 734
    },
    {
      "epoch": 0.35393968578823815,
      "grad_norm": 0.7615075707435608,
      "learning_rate": 0.0001500450230147116,
      "loss": 0.6860317587852478,
      "step": 735
    },
    {
      "epoch": 0.3544212363811473,
      "grad_norm": 1.7672241926193237,
      "learning_rate": 0.00014990991342913974,
      "loss": 1.0541589260101318,
      "step": 736
    },
    {
      "epoch": 0.35490278697405647,
      "grad_norm": 2.9616379737854004,
      "learning_rate": 0.00014977468240195084,
      "loss": 0.7769518494606018,
      "step": 737
    },
    {
      "epoch": 0.35538433756696564,
      "grad_norm": 3.0688796043395996,
      "learning_rate": 0.0001496393302621912,
      "loss": 0.8056391477584839,
      "step": 738
    },
    {
      "epoch": 0.3558658881598748,
      "grad_norm": 1.8512976169586182,
      "learning_rate": 0.00014950385733920188,
      "loss": 0.7788792848587036,
      "step": 739
    },
    {
      "epoch": 0.35634743875278396,
      "grad_norm": 2.7824597358703613,
      "learning_rate": 0.00014936826396261783,
      "loss": 0.8260157704353333,
      "step": 740
    },
    {
      "epoch": 0.3568289893456931,
      "grad_norm": 2.353167772293091,
      "learning_rate": 0.00014923255046236705,
      "loss": 0.9451891183853149,
      "step": 741
    },
    {
      "epoch": 0.3573105399386023,
      "grad_norm": 2.8516533374786377,
      "learning_rate": 0.00014909671716866984,
      "loss": 0.6984317302703857,
      "step": 742
    },
    {
      "epoch": 0.35779209053151145,
      "grad_norm": 2.587214946746826,
      "learning_rate": 0.00014896076441203802,
      "loss": 0.9936342239379883,
      "step": 743
    },
    {
      "epoch": 0.3582736411244206,
      "grad_norm": 2.178942918777466,
      "learning_rate": 0.000148824692523274,
      "loss": 0.93442702293396,
      "step": 744
    },
    {
      "epoch": 0.3587551917173298,
      "grad_norm": 3.4893758296966553,
      "learning_rate": 0.0001486885018334702,
      "loss": 0.9630119204521179,
      "step": 745
    },
    {
      "epoch": 0.35923674231023894,
      "grad_norm": 2.7591657638549805,
      "learning_rate": 0.00014855219267400797,
      "loss": 0.9268670082092285,
      "step": 746
    },
    {
      "epoch": 0.35971829290314816,
      "grad_norm": 2.116692066192627,
      "learning_rate": 0.00014841576537655705,
      "loss": 0.9149730205535889,
      "step": 747
    },
    {
      "epoch": 0.3601998434960573,
      "grad_norm": 2.387363910675049,
      "learning_rate": 0.00014827922027307451,
      "loss": 0.9265523552894592,
      "step": 748
    },
    {
      "epoch": 0.3606813940889665,
      "grad_norm": 2.553795337677002,
      "learning_rate": 0.00014814255769580415,
      "loss": 0.8555904626846313,
      "step": 749
    },
    {
      "epoch": 0.36116294468187565,
      "grad_norm": 0.9291559457778931,
      "learning_rate": 0.00014800577797727558,
      "loss": 0.8189652562141418,
      "step": 750
    },
    {
      "epoch": 0.3616444952747848,
      "grad_norm": 1.9631831645965576,
      "learning_rate": 0.00014786888145030343,
      "loss": 1.18495774269104,
      "step": 751
    },
    {
      "epoch": 0.362126045867694,
      "grad_norm": 4.110176086425781,
      "learning_rate": 0.0001477318684479866,
      "loss": 0.8783632516860962,
      "step": 752
    },
    {
      "epoch": 0.36260759646060314,
      "grad_norm": 13.664667129516602,
      "learning_rate": 0.00014759473930370736,
      "loss": 0.5866649150848389,
      "step": 753
    },
    {
      "epoch": 0.3630891470535123,
      "grad_norm": 3.042158603668213,
      "learning_rate": 0.0001474574943511306,
      "loss": 0.7243727445602417,
      "step": 754
    },
    {
      "epoch": 0.36357069764642147,
      "grad_norm": 4.326662540435791,
      "learning_rate": 0.0001473201339242029,
      "loss": 1.0714976787567139,
      "step": 755
    },
    {
      "epoch": 0.36405224823933063,
      "grad_norm": 1.5671316385269165,
      "learning_rate": 0.000147182658357152,
      "loss": 0.3876141905784607,
      "step": 756
    },
    {
      "epoch": 0.3645337988322398,
      "grad_norm": 2.065976142883301,
      "learning_rate": 0.00014704506798448566,
      "loss": 0.5911609530448914,
      "step": 757
    },
    {
      "epoch": 0.36501534942514896,
      "grad_norm": 2.8105785846710205,
      "learning_rate": 0.00014690736314099101,
      "loss": 0.6328890919685364,
      "step": 758
    },
    {
      "epoch": 0.3654969000180581,
      "grad_norm": 2.536707639694214,
      "learning_rate": 0.00014676954416173373,
      "loss": 1.633359670639038,
      "step": 759
    },
    {
      "epoch": 0.3659784506109673,
      "grad_norm": 3.183945655822754,
      "learning_rate": 0.00014663161138205724,
      "loss": 0.8919988870620728,
      "step": 760
    },
    {
      "epoch": 0.3664600012038765,
      "grad_norm": 1.9773579835891724,
      "learning_rate": 0.00014649356513758176,
      "loss": 0.8086701035499573,
      "step": 761
    },
    {
      "epoch": 0.36694155179678567,
      "grad_norm": 3.7868988513946533,
      "learning_rate": 0.00014635540576420374,
      "loss": 0.9487038850784302,
      "step": 762
    },
    {
      "epoch": 0.36742310238969483,
      "grad_norm": 3.004542827606201,
      "learning_rate": 0.0001462171335980948,
      "loss": 0.6151854991912842,
      "step": 763
    },
    {
      "epoch": 0.367904652982604,
      "grad_norm": 2.0149967670440674,
      "learning_rate": 0.00014607874897570105,
      "loss": 0.40916958451271057,
      "step": 764
    },
    {
      "epoch": 0.36838620357551316,
      "grad_norm": 2.9412448406219482,
      "learning_rate": 0.0001459402522337422,
      "loss": 0.902844250202179,
      "step": 765
    },
    {
      "epoch": 0.3688677541684223,
      "grad_norm": 2.487145185470581,
      "learning_rate": 0.00014580164370921078,
      "loss": 0.529595136642456,
      "step": 766
    },
    {
      "epoch": 0.3693493047613315,
      "grad_norm": 3.0917794704437256,
      "learning_rate": 0.0001456629237393713,
      "loss": 0.8628991842269897,
      "step": 767
    },
    {
      "epoch": 0.36983085535424065,
      "grad_norm": 3.183295726776123,
      "learning_rate": 0.00014552409266175952,
      "loss": 0.5798790454864502,
      "step": 768
    },
    {
      "epoch": 0.3703124059471498,
      "grad_norm": 3.3284950256347656,
      "learning_rate": 0.00014538515081418142,
      "loss": 1.0323659181594849,
      "step": 769
    },
    {
      "epoch": 0.370793956540059,
      "grad_norm": 2.198798894882202,
      "learning_rate": 0.00014524609853471264,
      "loss": 0.8785564303398132,
      "step": 770
    },
    {
      "epoch": 0.37127550713296814,
      "grad_norm": 2.7441883087158203,
      "learning_rate": 0.00014510693616169741,
      "loss": 0.9428408145904541,
      "step": 771
    },
    {
      "epoch": 0.3717570577258773,
      "grad_norm": 1.532981038093567,
      "learning_rate": 0.0001449676640337479,
      "loss": 0.857248067855835,
      "step": 772
    },
    {
      "epoch": 0.37223860831878647,
      "grad_norm": 3.4213173389434814,
      "learning_rate": 0.00014482828248974335,
      "loss": 0.7497163414955139,
      "step": 773
    },
    {
      "epoch": 0.3727201589116957,
      "grad_norm": 1.356554627418518,
      "learning_rate": 0.00014468879186882916,
      "loss": 1.0976029634475708,
      "step": 774
    },
    {
      "epoch": 0.37320170950460485,
      "grad_norm": 4.762817859649658,
      "learning_rate": 0.00014454919251041622,
      "loss": 0.6355637907981873,
      "step": 775
    },
    {
      "epoch": 0.373683260097514,
      "grad_norm": 2.7758395671844482,
      "learning_rate": 0.00014440948475418,
      "loss": 0.6969090104103088,
      "step": 776
    },
    {
      "epoch": 0.3741648106904232,
      "grad_norm": 3.0153472423553467,
      "learning_rate": 0.00014426966894005966,
      "loss": 0.9582271575927734,
      "step": 777
    },
    {
      "epoch": 0.37464636128333234,
      "grad_norm": 2.0446996688842773,
      "learning_rate": 0.0001441297454082573,
      "loss": 0.9723330736160278,
      "step": 778
    },
    {
      "epoch": 0.3751279118762415,
      "grad_norm": 4.276880741119385,
      "learning_rate": 0.00014398971449923722,
      "loss": 1.2020187377929688,
      "step": 779
    },
    {
      "epoch": 0.37560946246915067,
      "grad_norm": 1.7344483137130737,
      "learning_rate": 0.00014384957655372483,
      "loss": 0.5605676174163818,
      "step": 780
    },
    {
      "epoch": 0.37609101306205983,
      "grad_norm": 5.550004959106445,
      "learning_rate": 0.00014370933191270617,
      "loss": 1.1839048862457275,
      "step": 781
    },
    {
      "epoch": 0.376572563654969,
      "grad_norm": 2.1490771770477295,
      "learning_rate": 0.0001435689809174267,
      "loss": 0.8717823028564453,
      "step": 782
    },
    {
      "epoch": 0.37705411424787816,
      "grad_norm": 4.095498561859131,
      "learning_rate": 0.0001434285239093908,
      "loss": 0.7953165173530579,
      "step": 783
    },
    {
      "epoch": 0.3775356648407873,
      "grad_norm": 4.037353038787842,
      "learning_rate": 0.00014328796123036071,
      "loss": 1.1269760131835938,
      "step": 784
    },
    {
      "epoch": 0.3780172154336965,
      "grad_norm": 3.893498420715332,
      "learning_rate": 0.0001431472932223559,
      "loss": 1.1473517417907715,
      "step": 785
    },
    {
      "epoch": 0.37849876602660565,
      "grad_norm": 2.345381021499634,
      "learning_rate": 0.00014300652022765207,
      "loss": 0.7361969947814941,
      "step": 786
    },
    {
      "epoch": 0.3789803166195148,
      "grad_norm": 5.63955545425415,
      "learning_rate": 0.00014286564258878033,
      "loss": 1.414058804512024,
      "step": 787
    },
    {
      "epoch": 0.37946186721242403,
      "grad_norm": 2.7331528663635254,
      "learning_rate": 0.00014272466064852644,
      "loss": 0.5617365837097168,
      "step": 788
    },
    {
      "epoch": 0.3799434178053332,
      "grad_norm": 3.676002025604248,
      "learning_rate": 0.00014258357474993,
      "loss": 0.5242584347724915,
      "step": 789
    },
    {
      "epoch": 0.38042496839824236,
      "grad_norm": 3.942762613296509,
      "learning_rate": 0.0001424423852362835,
      "loss": 0.9681786894798279,
      "step": 790
    },
    {
      "epoch": 0.3809065189911515,
      "grad_norm": 1.1877814531326294,
      "learning_rate": 0.00014230109245113158,
      "loss": 0.6973773241043091,
      "step": 791
    },
    {
      "epoch": 0.3813880695840607,
      "grad_norm": 3.188105821609497,
      "learning_rate": 0.00014215969673827018,
      "loss": 0.6791388392448425,
      "step": 792
    },
    {
      "epoch": 0.38186962017696985,
      "grad_norm": 2.562154531478882,
      "learning_rate": 0.00014201819844174564,
      "loss": 0.8226215243339539,
      "step": 793
    },
    {
      "epoch": 0.382351170769879,
      "grad_norm": 7.387059688568115,
      "learning_rate": 0.0001418765979058539,
      "loss": 1.0329314470291138,
      "step": 794
    },
    {
      "epoch": 0.3828327213627882,
      "grad_norm": 2.0933785438537598,
      "learning_rate": 0.00014173489547513973,
      "loss": 0.8333945870399475,
      "step": 795
    },
    {
      "epoch": 0.38331427195569734,
      "grad_norm": 3.3404147624969482,
      "learning_rate": 0.00014159309149439582,
      "loss": 0.7309401035308838,
      "step": 796
    },
    {
      "epoch": 0.3837958225486065,
      "grad_norm": 2.6134278774261475,
      "learning_rate": 0.00014145118630866187,
      "loss": 0.739777684211731,
      "step": 797
    },
    {
      "epoch": 0.38427737314151567,
      "grad_norm": 2.8803746700286865,
      "learning_rate": 0.000141309180263224,
      "loss": 0.8909446597099304,
      "step": 798
    },
    {
      "epoch": 0.38475892373442483,
      "grad_norm": 2.1177947521209717,
      "learning_rate": 0.0001411670737036135,
      "loss": 0.5141834020614624,
      "step": 799
    },
    {
      "epoch": 0.385240474327334,
      "grad_norm": 6.034994602203369,
      "learning_rate": 0.0001410248669756065,
      "loss": 0.7950483560562134,
      "step": 800
    },
    {
      "epoch": 0.38572202492024316,
      "grad_norm": 1.906010389328003,
      "learning_rate": 0.00014088256042522264,
      "loss": 1.084035873413086,
      "step": 801
    },
    {
      "epoch": 0.3862035755131524,
      "grad_norm": 1.750962257385254,
      "learning_rate": 0.00014074015439872458,
      "loss": 1.0769619941711426,
      "step": 802
    },
    {
      "epoch": 0.38668512610606154,
      "grad_norm": 2.9726738929748535,
      "learning_rate": 0.00014059764924261703,
      "loss": 0.6981932520866394,
      "step": 803
    },
    {
      "epoch": 0.3871666766989707,
      "grad_norm": 3.2621171474456787,
      "learning_rate": 0.00014045504530364584,
      "loss": 0.971074640750885,
      "step": 804
    },
    {
      "epoch": 0.38764822729187987,
      "grad_norm": 1.6707664728164673,
      "learning_rate": 0.00014031234292879725,
      "loss": 1.0051618814468384,
      "step": 805
    },
    {
      "epoch": 0.38812977788478903,
      "grad_norm": 2.2506284713745117,
      "learning_rate": 0.00014016954246529696,
      "loss": 0.90675288438797,
      "step": 806
    },
    {
      "epoch": 0.3886113284776982,
      "grad_norm": 4.92419958114624,
      "learning_rate": 0.00014002664426060942,
      "loss": 1.0551118850708008,
      "step": 807
    },
    {
      "epoch": 0.38909287907060736,
      "grad_norm": 3.4966378211975098,
      "learning_rate": 0.00013988364866243693,
      "loss": 1.4710934162139893,
      "step": 808
    },
    {
      "epoch": 0.3895744296635165,
      "grad_norm": 3.4958462715148926,
      "learning_rate": 0.00013974055601871868,
      "loss": 1.0214515924453735,
      "step": 809
    },
    {
      "epoch": 0.3900559802564257,
      "grad_norm": 2.0109124183654785,
      "learning_rate": 0.00013959736667762998,
      "loss": 0.9646590948104858,
      "step": 810
    },
    {
      "epoch": 0.39053753084933485,
      "grad_norm": 2.275374174118042,
      "learning_rate": 0.00013945408098758156,
      "loss": 0.942484974861145,
      "step": 811
    },
    {
      "epoch": 0.391019081442244,
      "grad_norm": 3.084449052810669,
      "learning_rate": 0.0001393106992972184,
      "loss": 0.9004904627799988,
      "step": 812
    },
    {
      "epoch": 0.3915006320351532,
      "grad_norm": 4.096373558044434,
      "learning_rate": 0.00013916722195541926,
      "loss": 0.7134367227554321,
      "step": 813
    },
    {
      "epoch": 0.39198218262806234,
      "grad_norm": 2.6820106506347656,
      "learning_rate": 0.00013902364931129557,
      "loss": 1.200644612312317,
      "step": 814
    },
    {
      "epoch": 0.39246373322097156,
      "grad_norm": 6.483953952789307,
      "learning_rate": 0.00013887998171419058,
      "loss": 0.8891584873199463,
      "step": 815
    },
    {
      "epoch": 0.3929452838138807,
      "grad_norm": 1.9730106592178345,
      "learning_rate": 0.00013873621951367862,
      "loss": 0.6953721642494202,
      "step": 816
    },
    {
      "epoch": 0.3934268344067899,
      "grad_norm": 7.508829593658447,
      "learning_rate": 0.00013859236305956425,
      "loss": 0.582947850227356,
      "step": 817
    },
    {
      "epoch": 0.39390838499969905,
      "grad_norm": 3.7755587100982666,
      "learning_rate": 0.00013844841270188132,
      "loss": 0.8725051879882812,
      "step": 818
    },
    {
      "epoch": 0.3943899355926082,
      "grad_norm": 2.484464645385742,
      "learning_rate": 0.00013830436879089228,
      "loss": 0.8891981840133667,
      "step": 819
    },
    {
      "epoch": 0.3948714861855174,
      "grad_norm": 4.062020778656006,
      "learning_rate": 0.00013816023167708704,
      "loss": 0.6328946948051453,
      "step": 820
    },
    {
      "epoch": 0.39535303677842654,
      "grad_norm": 3.9620730876922607,
      "learning_rate": 0.00013801600171118244,
      "loss": 0.776721179485321,
      "step": 821
    },
    {
      "epoch": 0.3958345873713357,
      "grad_norm": 2.2038917541503906,
      "learning_rate": 0.00013787167924412112,
      "loss": 0.6010950207710266,
      "step": 822
    },
    {
      "epoch": 0.39631613796424486,
      "grad_norm": 2.821363687515259,
      "learning_rate": 0.0001377272646270709,
      "loss": 0.8728996515274048,
      "step": 823
    },
    {
      "epoch": 0.396797688557154,
      "grad_norm": 1.586208701133728,
      "learning_rate": 0.00013758275821142382,
      "loss": 0.5970646142959595,
      "step": 824
    },
    {
      "epoch": 0.3972792391500632,
      "grad_norm": 4.052238464355469,
      "learning_rate": 0.00013743816034879523,
      "loss": 0.9887663125991821,
      "step": 825
    },
    {
      "epoch": 0.39776078974297235,
      "grad_norm": 7.808894157409668,
      "learning_rate": 0.000137293471391023,
      "loss": 0.6769440174102783,
      "step": 826
    },
    {
      "epoch": 0.3982423403358815,
      "grad_norm": 1.4124599695205688,
      "learning_rate": 0.00013714869169016667,
      "loss": 0.698244035243988,
      "step": 827
    },
    {
      "epoch": 0.3987238909287907,
      "grad_norm": 3.47186279296875,
      "learning_rate": 0.00013700382159850656,
      "loss": 0.5083252191543579,
      "step": 828
    },
    {
      "epoch": 0.3992054415216999,
      "grad_norm": 3.0755064487457275,
      "learning_rate": 0.00013685886146854297,
      "loss": 0.7256084680557251,
      "step": 829
    },
    {
      "epoch": 0.39968699211460906,
      "grad_norm": 2.346916913986206,
      "learning_rate": 0.00013671381165299525,
      "loss": 0.7663772106170654,
      "step": 830
    },
    {
      "epoch": 0.4001685427075182,
      "grad_norm": 4.709892749786377,
      "learning_rate": 0.00013656867250480098,
      "loss": 1.231966257095337,
      "step": 831
    },
    {
      "epoch": 0.4006500933004274,
      "grad_norm": 2.162738561630249,
      "learning_rate": 0.00013642344437711512,
      "loss": 0.6987869739532471,
      "step": 832
    },
    {
      "epoch": 0.40113164389333655,
      "grad_norm": 1.9764432907104492,
      "learning_rate": 0.00013627812762330912,
      "loss": 0.4391883909702301,
      "step": 833
    },
    {
      "epoch": 0.4016131944862457,
      "grad_norm": 1.9531248807907104,
      "learning_rate": 0.00013613272259697007,
      "loss": 0.8051517009735107,
      "step": 834
    },
    {
      "epoch": 0.4020947450791549,
      "grad_norm": 1.628053069114685,
      "learning_rate": 0.00013598722965189986,
      "loss": 1.2156120538711548,
      "step": 835
    },
    {
      "epoch": 0.40257629567206404,
      "grad_norm": 1.67246675491333,
      "learning_rate": 0.0001358416491421143,
      "loss": 0.7458651065826416,
      "step": 836
    },
    {
      "epoch": 0.4030578462649732,
      "grad_norm": 5.9809465408325195,
      "learning_rate": 0.00013569598142184225,
      "loss": 0.7124079465866089,
      "step": 837
    },
    {
      "epoch": 0.40353939685788237,
      "grad_norm": 1.8881200551986694,
      "learning_rate": 0.00013555022684552483,
      "loss": 0.8664608001708984,
      "step": 838
    },
    {
      "epoch": 0.40402094745079153,
      "grad_norm": 2.3725993633270264,
      "learning_rate": 0.00013540438576781441,
      "loss": 0.7445593476295471,
      "step": 839
    },
    {
      "epoch": 0.4045024980437007,
      "grad_norm": 2.2124617099761963,
      "learning_rate": 0.0001352584585435739,
      "loss": 1.4516873359680176,
      "step": 840
    },
    {
      "epoch": 0.40498404863660986,
      "grad_norm": 2.961352586746216,
      "learning_rate": 0.00013511244552787583,
      "loss": 0.752439558506012,
      "step": 841
    },
    {
      "epoch": 0.405465599229519,
      "grad_norm": 2.48881196975708,
      "learning_rate": 0.00013496634707600147,
      "loss": 1.0942041873931885,
      "step": 842
    },
    {
      "epoch": 0.40594714982242824,
      "grad_norm": 3.2813918590545654,
      "learning_rate": 0.0001348201635434399,
      "loss": 1.0821374654769897,
      "step": 843
    },
    {
      "epoch": 0.4064287004153374,
      "grad_norm": 3.466980218887329,
      "learning_rate": 0.0001346738952858873,
      "loss": 1.102767825126648,
      "step": 844
    },
    {
      "epoch": 0.40691025100824657,
      "grad_norm": 4.052308559417725,
      "learning_rate": 0.000134527542659246,
      "loss": 1.5272918939590454,
      "step": 845
    },
    {
      "epoch": 0.40739180160115573,
      "grad_norm": 3.340364694595337,
      "learning_rate": 0.00013438110601962362,
      "loss": 0.6943974494934082,
      "step": 846
    },
    {
      "epoch": 0.4078733521940649,
      "grad_norm": 1.8921301364898682,
      "learning_rate": 0.00013423458572333214,
      "loss": 0.9528016448020935,
      "step": 847
    },
    {
      "epoch": 0.40835490278697406,
      "grad_norm": 1.971108078956604,
      "learning_rate": 0.0001340879821268872,
      "loss": 1.0711705684661865,
      "step": 848
    },
    {
      "epoch": 0.4088364533798832,
      "grad_norm": 2.413546323776245,
      "learning_rate": 0.000133941295587007,
      "loss": 0.7476043701171875,
      "step": 849
    },
    {
      "epoch": 0.4093180039727924,
      "grad_norm": 1.7470178604125977,
      "learning_rate": 0.00013379452646061164,
      "loss": 0.5836044549942017,
      "step": 850
    },
    {
      "epoch": 0.40979955456570155,
      "grad_norm": 2.3459055423736572,
      "learning_rate": 0.0001336476751048222,
      "loss": 0.9842826128005981,
      "step": 851
    },
    {
      "epoch": 0.4102811051586107,
      "grad_norm": 3.0704832077026367,
      "learning_rate": 0.00013350074187695979,
      "loss": 1.0236726999282837,
      "step": 852
    },
    {
      "epoch": 0.4107626557515199,
      "grad_norm": 1.8277277946472168,
      "learning_rate": 0.00013335372713454467,
      "loss": 0.6626753807067871,
      "step": 853
    },
    {
      "epoch": 0.41124420634442904,
      "grad_norm": 2.284486770629883,
      "learning_rate": 0.0001332066312352956,
      "loss": 0.540708601474762,
      "step": 854
    },
    {
      "epoch": 0.4117257569373382,
      "grad_norm": 3.2437660694122314,
      "learning_rate": 0.00013305945453712868,
      "loss": 0.8092200756072998,
      "step": 855
    },
    {
      "epoch": 0.4122073075302474,
      "grad_norm": 16.97819709777832,
      "learning_rate": 0.0001329121973981567,
      "loss": 0.6793229579925537,
      "step": 856
    },
    {
      "epoch": 0.4126888581231566,
      "grad_norm": 3.0376884937286377,
      "learning_rate": 0.00013276486017668807,
      "loss": 0.7474631667137146,
      "step": 857
    },
    {
      "epoch": 0.41317040871606575,
      "grad_norm": 1.2816187143325806,
      "learning_rate": 0.0001326174432312262,
      "loss": 1.1490566730499268,
      "step": 858
    },
    {
      "epoch": 0.4136519593089749,
      "grad_norm": 1.441726803779602,
      "learning_rate": 0.00013246994692046836,
      "loss": 0.9337580800056458,
      "step": 859
    },
    {
      "epoch": 0.4141335099018841,
      "grad_norm": 3.029785633087158,
      "learning_rate": 0.000132322371603305,
      "loss": 0.7632018327713013,
      "step": 860
    },
    {
      "epoch": 0.41461506049479324,
      "grad_norm": 5.360157489776611,
      "learning_rate": 0.0001321747176388188,
      "loss": 0.856697678565979,
      "step": 861
    },
    {
      "epoch": 0.4150966110877024,
      "grad_norm": 2.823807716369629,
      "learning_rate": 0.00013202698538628376,
      "loss": 1.0925341844558716,
      "step": 862
    },
    {
      "epoch": 0.41557816168061157,
      "grad_norm": 3.7324116230010986,
      "learning_rate": 0.00013187917520516448,
      "loss": 1.181652545928955,
      "step": 863
    },
    {
      "epoch": 0.41605971227352073,
      "grad_norm": 3.2788307666778564,
      "learning_rate": 0.00013173128745511508,
      "loss": 1.2298600673675537,
      "step": 864
    },
    {
      "epoch": 0.4165412628664299,
      "grad_norm": 1.4664795398712158,
      "learning_rate": 0.0001315833224959784,
      "loss": 0.6844913959503174,
      "step": 865
    },
    {
      "epoch": 0.41702281345933906,
      "grad_norm": 2.564586877822876,
      "learning_rate": 0.00013143528068778525,
      "loss": 1.0977352857589722,
      "step": 866
    },
    {
      "epoch": 0.4175043640522482,
      "grad_norm": 3.34252667427063,
      "learning_rate": 0.00013128716239075338,
      "loss": 0.9098133444786072,
      "step": 867
    },
    {
      "epoch": 0.4179859146451574,
      "grad_norm": 2.997368335723877,
      "learning_rate": 0.00013113896796528664,
      "loss": 0.6700395941734314,
      "step": 868
    },
    {
      "epoch": 0.41846746523806655,
      "grad_norm": 1.5699840784072876,
      "learning_rate": 0.00013099069777197412,
      "loss": 0.7059319019317627,
      "step": 869
    },
    {
      "epoch": 0.41894901583097577,
      "grad_norm": 3.758190631866455,
      "learning_rate": 0.0001308423521715893,
      "loss": 1.0462746620178223,
      "step": 870
    },
    {
      "epoch": 0.41943056642388493,
      "grad_norm": 4.388060569763184,
      "learning_rate": 0.00013069393152508906,
      "loss": 1.2692842483520508,
      "step": 871
    },
    {
      "epoch": 0.4199121170167941,
      "grad_norm": 6.121324062347412,
      "learning_rate": 0.00013054543619361303,
      "loss": 1.72324538230896,
      "step": 872
    },
    {
      "epoch": 0.42039366760970326,
      "grad_norm": 2.058579683303833,
      "learning_rate": 0.0001303968665384824,
      "loss": 0.5607519745826721,
      "step": 873
    },
    {
      "epoch": 0.4208752182026124,
      "grad_norm": 2.755772352218628,
      "learning_rate": 0.00013024822292119934,
      "loss": 1.167569875717163,
      "step": 874
    },
    {
      "epoch": 0.4213567687955216,
      "grad_norm": 1.8952524662017822,
      "learning_rate": 0.0001300995057034459,
      "loss": 0.5059840679168701,
      "step": 875
    },
    {
      "epoch": 0.42183831938843075,
      "grad_norm": 2.785203695297241,
      "learning_rate": 0.00012995071524708325,
      "loss": 0.9211249351501465,
      "step": 876
    },
    {
      "epoch": 0.4223198699813399,
      "grad_norm": 4.891737461090088,
      "learning_rate": 0.00012980185191415074,
      "loss": 0.7493455410003662,
      "step": 877
    },
    {
      "epoch": 0.4228014205742491,
      "grad_norm": 2.06239914894104,
      "learning_rate": 0.0001296529160668651,
      "loss": 1.014078140258789,
      "step": 878
    },
    {
      "epoch": 0.42328297116715824,
      "grad_norm": 5.26932430267334,
      "learning_rate": 0.00012950390806761944,
      "loss": 0.6335924863815308,
      "step": 879
    },
    {
      "epoch": 0.4237645217600674,
      "grad_norm": 2.8031914234161377,
      "learning_rate": 0.0001293548282789825,
      "loss": 0.8351728916168213,
      "step": 880
    },
    {
      "epoch": 0.42424607235297657,
      "grad_norm": 1.6703650951385498,
      "learning_rate": 0.00012920567706369758,
      "loss": 0.7832145094871521,
      "step": 881
    },
    {
      "epoch": 0.42472762294588573,
      "grad_norm": 1.514344573020935,
      "learning_rate": 0.00012905645478468192,
      "loss": 1.0918382406234741,
      "step": 882
    },
    {
      "epoch": 0.4252091735387949,
      "grad_norm": 2.142754077911377,
      "learning_rate": 0.00012890716180502564,
      "loss": 0.7013830542564392,
      "step": 883
    },
    {
      "epoch": 0.4256907241317041,
      "grad_norm": 1.1120011806488037,
      "learning_rate": 0.00012875779848799078,
      "loss": 1.0959222316741943,
      "step": 884
    },
    {
      "epoch": 0.4261722747246133,
      "grad_norm": 2.1146645545959473,
      "learning_rate": 0.00012860836519701063,
      "loss": 1.017622947692871,
      "step": 885
    },
    {
      "epoch": 0.42665382531752244,
      "grad_norm": 2.158938407897949,
      "learning_rate": 0.00012845886229568873,
      "loss": 0.9938173890113831,
      "step": 886
    },
    {
      "epoch": 0.4271353759104316,
      "grad_norm": 2.879587411880493,
      "learning_rate": 0.00012830929014779797,
      "loss": 0.8292675018310547,
      "step": 887
    },
    {
      "epoch": 0.42761692650334077,
      "grad_norm": 2.129265785217285,
      "learning_rate": 0.0001281596491172797,
      "loss": 0.7805866599082947,
      "step": 888
    },
    {
      "epoch": 0.42809847709624993,
      "grad_norm": 2.7378902435302734,
      "learning_rate": 0.00012800993956824303,
      "loss": 1.0834482908248901,
      "step": 889
    },
    {
      "epoch": 0.4285800276891591,
      "grad_norm": 4.303222179412842,
      "learning_rate": 0.00012786016186496358,
      "loss": 0.879001259803772,
      "step": 890
    },
    {
      "epoch": 0.42906157828206826,
      "grad_norm": 1.383887529373169,
      "learning_rate": 0.000127710316371883,
      "loss": 0.6555094122886658,
      "step": 891
    },
    {
      "epoch": 0.4295431288749774,
      "grad_norm": 3.3802478313446045,
      "learning_rate": 0.0001275604034536077,
      "loss": 0.6983332633972168,
      "step": 892
    },
    {
      "epoch": 0.4300246794678866,
      "grad_norm": 4.484005451202393,
      "learning_rate": 0.0001274104234749083,
      "loss": 1.1950817108154297,
      "step": 893
    },
    {
      "epoch": 0.43050623006079575,
      "grad_norm": 2.093848466873169,
      "learning_rate": 0.00012726037680071853,
      "loss": 0.8492384552955627,
      "step": 894
    },
    {
      "epoch": 0.4309877806537049,
      "grad_norm": 6.30825138092041,
      "learning_rate": 0.00012711026379613434,
      "loss": 0.769837498664856,
      "step": 895
    },
    {
      "epoch": 0.4314693312466141,
      "grad_norm": 1.8871920108795166,
      "learning_rate": 0.00012696008482641325,
      "loss": 0.6886523962020874,
      "step": 896
    },
    {
      "epoch": 0.43195088183952324,
      "grad_norm": 2.9344093799591064,
      "learning_rate": 0.00012680984025697313,
      "loss": 0.6091852784156799,
      "step": 897
    },
    {
      "epoch": 0.43243243243243246,
      "grad_norm": 2.977684259414673,
      "learning_rate": 0.00012665953045339152,
      "loss": 1.206083059310913,
      "step": 898
    },
    {
      "epoch": 0.4329139830253416,
      "grad_norm": 3.1293866634368896,
      "learning_rate": 0.0001265091557814047,
      "loss": 1.1070964336395264,
      "step": 899
    },
    {
      "epoch": 0.4333955336182508,
      "grad_norm": 4.85739278793335,
      "learning_rate": 0.00012635871660690676,
      "loss": 0.6476233601570129,
      "step": 900
    },
    {
      "epoch": 0.43387708421115995,
      "grad_norm": 2.5916333198547363,
      "learning_rate": 0.0001262082132959488,
      "loss": 0.9050962328910828,
      "step": 901
    },
    {
      "epoch": 0.4343586348040691,
      "grad_norm": 4.994722366333008,
      "learning_rate": 0.00012605764621473792,
      "loss": 0.9139664769172668,
      "step": 902
    },
    {
      "epoch": 0.4348401853969783,
      "grad_norm": 1.8033146858215332,
      "learning_rate": 0.00012590701572963642,
      "loss": 0.5417568683624268,
      "step": 903
    },
    {
      "epoch": 0.43532173598988744,
      "grad_norm": 3.6167919635772705,
      "learning_rate": 0.00012575632220716078,
      "loss": 0.7129384279251099,
      "step": 904
    },
    {
      "epoch": 0.4358032865827966,
      "grad_norm": 1.6472055912017822,
      "learning_rate": 0.000125605566013981,
      "loss": 0.9040926694869995,
      "step": 905
    },
    {
      "epoch": 0.43628483717570576,
      "grad_norm": 1.6665151119232178,
      "learning_rate": 0.00012545474751691953,
      "loss": 0.8669220209121704,
      "step": 906
    },
    {
      "epoch": 0.4367663877686149,
      "grad_norm": 2.999077796936035,
      "learning_rate": 0.00012530386708295036,
      "loss": 1.1189666986465454,
      "step": 907
    },
    {
      "epoch": 0.4372479383615241,
      "grad_norm": 2.541163682937622,
      "learning_rate": 0.00012515292507919829,
      "loss": 0.8834406733512878,
      "step": 908
    },
    {
      "epoch": 0.43772948895443325,
      "grad_norm": 1.8075464963912964,
      "learning_rate": 0.0001250019218729378,
      "loss": 0.5413083434104919,
      "step": 909
    },
    {
      "epoch": 0.4382110395473424,
      "grad_norm": 6.940054893493652,
      "learning_rate": 0.00012485085783159238,
      "loss": 0.5122719407081604,
      "step": 910
    },
    {
      "epoch": 0.43869259014025164,
      "grad_norm": 2.529057025909424,
      "learning_rate": 0.00012469973332273354,
      "loss": 0.9689932465553284,
      "step": 911
    },
    {
      "epoch": 0.4391741407331608,
      "grad_norm": 3.0322325229644775,
      "learning_rate": 0.00012454854871407994,
      "loss": 1.0562812089920044,
      "step": 912
    },
    {
      "epoch": 0.43965569132606996,
      "grad_norm": 2.4834840297698975,
      "learning_rate": 0.00012439730437349635,
      "loss": 0.9881972670555115,
      "step": 913
    },
    {
      "epoch": 0.4401372419189791,
      "grad_norm": 1.512488603591919,
      "learning_rate": 0.00012424600066899302,
      "loss": 0.9199472069740295,
      "step": 914
    },
    {
      "epoch": 0.4406187925118883,
      "grad_norm": 3.734846591949463,
      "learning_rate": 0.00012409463796872464,
      "loss": 1.1432019472122192,
      "step": 915
    },
    {
      "epoch": 0.44110034310479745,
      "grad_norm": 1.0262116193771362,
      "learning_rate": 0.0001239432166409893,
      "loss": 1.0165278911590576,
      "step": 916
    },
    {
      "epoch": 0.4415818936977066,
      "grad_norm": 2.11399245262146,
      "learning_rate": 0.00012379173705422795,
      "loss": 0.9167724251747131,
      "step": 917
    },
    {
      "epoch": 0.4420634442906158,
      "grad_norm": 1.3735132217407227,
      "learning_rate": 0.00012364019957702315,
      "loss": 0.4771014451980591,
      "step": 918
    },
    {
      "epoch": 0.44254499488352494,
      "grad_norm": 2.018268585205078,
      "learning_rate": 0.00012348860457809838,
      "loss": 0.7764768004417419,
      "step": 919
    },
    {
      "epoch": 0.4430265454764341,
      "grad_norm": 2.184174060821533,
      "learning_rate": 0.00012333695242631705,
      "loss": 0.7174931168556213,
      "step": 920
    },
    {
      "epoch": 0.44350809606934327,
      "grad_norm": 3.354451894760132,
      "learning_rate": 0.0001231852434906817,
      "loss": 0.931488037109375,
      "step": 921
    },
    {
      "epoch": 0.44398964666225244,
      "grad_norm": 3.204756498336792,
      "learning_rate": 0.00012303347814033292,
      "loss": 0.7502593398094177,
      "step": 922
    },
    {
      "epoch": 0.4444711972551616,
      "grad_norm": 1.5013762712478638,
      "learning_rate": 0.0001228816567445487,
      "loss": 1.0279090404510498,
      "step": 923
    },
    {
      "epoch": 0.44495274784807076,
      "grad_norm": 2.3139233589172363,
      "learning_rate": 0.0001227297796727433,
      "loss": 0.6862879395484924,
      "step": 924
    },
    {
      "epoch": 0.44543429844098,
      "grad_norm": 4.01719856262207,
      "learning_rate": 0.00012257784729446656,
      "loss": 1.393019437789917,
      "step": 925
    },
    {
      "epoch": 0.44591584903388914,
      "grad_norm": 2.0590579509735107,
      "learning_rate": 0.00012242585997940275,
      "loss": 0.7485421299934387,
      "step": 926
    },
    {
      "epoch": 0.4463973996267983,
      "grad_norm": 2.715683937072754,
      "learning_rate": 0.0001222738180973699,
      "loss": 0.7426998615264893,
      "step": 927
    },
    {
      "epoch": 0.44687895021970747,
      "grad_norm": 3.121952533721924,
      "learning_rate": 0.00012212172201831885,
      "loss": 0.47169557213783264,
      "step": 928
    },
    {
      "epoch": 0.44736050081261663,
      "grad_norm": 2.231208324432373,
      "learning_rate": 0.00012196957211233222,
      "loss": 1.1421630382537842,
      "step": 929
    },
    {
      "epoch": 0.4478420514055258,
      "grad_norm": 1.7500450611114502,
      "learning_rate": 0.00012181736874962371,
      "loss": 1.121241807937622,
      "step": 930
    },
    {
      "epoch": 0.44832360199843496,
      "grad_norm": 3.2750349044799805,
      "learning_rate": 0.00012166511230053696,
      "loss": 0.7124077081680298,
      "step": 931
    },
    {
      "epoch": 0.4488051525913441,
      "grad_norm": 6.525684356689453,
      "learning_rate": 0.00012151280313554486,
      "loss": 1.450913429260254,
      "step": 932
    },
    {
      "epoch": 0.4492867031842533,
      "grad_norm": 2.9183151721954346,
      "learning_rate": 0.00012136044162524858,
      "loss": 0.5961760878562927,
      "step": 933
    },
    {
      "epoch": 0.44976825377716245,
      "grad_norm": 3.11704683303833,
      "learning_rate": 0.00012120802814037663,
      "loss": 0.8525784611701965,
      "step": 934
    },
    {
      "epoch": 0.4502498043700716,
      "grad_norm": 3.9269542694091797,
      "learning_rate": 0.00012105556305178399,
      "loss": 1.108290195465088,
      "step": 935
    },
    {
      "epoch": 0.4507313549629808,
      "grad_norm": 2.7036521434783936,
      "learning_rate": 0.00012090304673045123,
      "loss": 1.036728858947754,
      "step": 936
    },
    {
      "epoch": 0.45121290555588994,
      "grad_norm": 0.9536427855491638,
      "learning_rate": 0.00012075047954748353,
      "loss": 0.415727436542511,
      "step": 937
    },
    {
      "epoch": 0.4516944561487991,
      "grad_norm": 2.0500926971435547,
      "learning_rate": 0.00012059786187410984,
      "loss": 1.376224398612976,
      "step": 938
    },
    {
      "epoch": 0.4521760067417083,
      "grad_norm": 2.452744483947754,
      "learning_rate": 0.000120445194081682,
      "loss": 0.5674085021018982,
      "step": 939
    },
    {
      "epoch": 0.4526575573346175,
      "grad_norm": 2.1720454692840576,
      "learning_rate": 0.00012029247654167379,
      "loss": 0.9531777501106262,
      "step": 940
    },
    {
      "epoch": 0.45313910792752665,
      "grad_norm": 3.1240689754486084,
      "learning_rate": 0.00012013970962568002,
      "loss": 0.8404605388641357,
      "step": 941
    },
    {
      "epoch": 0.4536206585204358,
      "grad_norm": 2.0688564777374268,
      "learning_rate": 0.00011998689370541562,
      "loss": 0.5712980628013611,
      "step": 942
    },
    {
      "epoch": 0.454102209113345,
      "grad_norm": 2.4777586460113525,
      "learning_rate": 0.00011983402915271478,
      "loss": 0.43273207545280457,
      "step": 943
    },
    {
      "epoch": 0.45458375970625414,
      "grad_norm": 3.779033660888672,
      "learning_rate": 0.00011968111633953007,
      "loss": 0.8489125370979309,
      "step": 944
    },
    {
      "epoch": 0.4550653102991633,
      "grad_norm": 2.9321274757385254,
      "learning_rate": 0.0001195281556379314,
      "loss": 0.8262616395950317,
      "step": 945
    },
    {
      "epoch": 0.45554686089207247,
      "grad_norm": 5.756001949310303,
      "learning_rate": 0.0001193751474201053,
      "loss": 0.606381356716156,
      "step": 946
    },
    {
      "epoch": 0.45602841148498163,
      "grad_norm": 2.845503807067871,
      "learning_rate": 0.00011922209205835382,
      "loss": 0.8182961344718933,
      "step": 947
    },
    {
      "epoch": 0.4565099620778908,
      "grad_norm": 3.4336414337158203,
      "learning_rate": 0.0001190689899250938,
      "loss": 0.8047642707824707,
      "step": 948
    },
    {
      "epoch": 0.45699151267079996,
      "grad_norm": 2.5517613887786865,
      "learning_rate": 0.00011891584139285582,
      "loss": 1.008400559425354,
      "step": 949
    },
    {
      "epoch": 0.4574730632637091,
      "grad_norm": 2.110889196395874,
      "learning_rate": 0.00011876264683428344,
      "loss": 0.9205650091171265,
      "step": 950
    },
    {
      "epoch": 0.4579546138566183,
      "grad_norm": 3.2672603130340576,
      "learning_rate": 0.00011860940662213211,
      "loss": 0.635238766670227,
      "step": 951
    },
    {
      "epoch": 0.4584361644495275,
      "grad_norm": 1.8438022136688232,
      "learning_rate": 0.00011845612112926843,
      "loss": 0.7883297801017761,
      "step": 952
    },
    {
      "epoch": 0.45891771504243667,
      "grad_norm": 1.3595741987228394,
      "learning_rate": 0.00011830279072866921,
      "loss": 0.609693169593811,
      "step": 953
    },
    {
      "epoch": 0.45939926563534583,
      "grad_norm": 2.6202378273010254,
      "learning_rate": 0.00011814941579342044,
      "loss": 0.8235170841217041,
      "step": 954
    },
    {
      "epoch": 0.459880816228255,
      "grad_norm": 2.131608724594116,
      "learning_rate": 0.00011799599669671654,
      "loss": 0.8079139590263367,
      "step": 955
    },
    {
      "epoch": 0.46036236682116416,
      "grad_norm": 2.5156984329223633,
      "learning_rate": 0.00011784253381185937,
      "loss": 0.6958327293395996,
      "step": 956
    },
    {
      "epoch": 0.4608439174140733,
      "grad_norm": 3.036233901977539,
      "learning_rate": 0.0001176890275122573,
      "loss": 1.0472469329833984,
      "step": 957
    },
    {
      "epoch": 0.4613254680069825,
      "grad_norm": 1.7393897771835327,
      "learning_rate": 0.0001175354781714244,
      "loss": 0.5142320394515991,
      "step": 958
    },
    {
      "epoch": 0.46180701859989165,
      "grad_norm": 2.7154998779296875,
      "learning_rate": 0.0001173818861629794,
      "loss": 0.9809057116508484,
      "step": 959
    },
    {
      "epoch": 0.4622885691928008,
      "grad_norm": 2.047426462173462,
      "learning_rate": 0.00011722825186064494,
      "loss": 0.9878953695297241,
      "step": 960
    },
    {
      "epoch": 0.46277011978571,
      "grad_norm": 2.798541784286499,
      "learning_rate": 0.00011707457563824646,
      "loss": 1.0951030254364014,
      "step": 961
    },
    {
      "epoch": 0.46325167037861914,
      "grad_norm": 2.197075843811035,
      "learning_rate": 0.00011692085786971149,
      "loss": 0.6801528930664062,
      "step": 962
    },
    {
      "epoch": 0.4637332209715283,
      "grad_norm": 2.722748279571533,
      "learning_rate": 0.00011676709892906858,
      "loss": 0.8837396502494812,
      "step": 963
    },
    {
      "epoch": 0.46421477156443747,
      "grad_norm": 0.9209997057914734,
      "learning_rate": 0.00011661329919044656,
      "loss": 0.8588595986366272,
      "step": 964
    },
    {
      "epoch": 0.46469632215734663,
      "grad_norm": 2.9362680912017822,
      "learning_rate": 0.00011645945902807341,
      "loss": 0.9675197601318359,
      "step": 965
    },
    {
      "epoch": 0.46517787275025585,
      "grad_norm": 1.631209373474121,
      "learning_rate": 0.00011630557881627553,
      "loss": 0.6794362664222717,
      "step": 966
    },
    {
      "epoch": 0.465659423343165,
      "grad_norm": 2.272028684616089,
      "learning_rate": 0.0001161516589294768,
      "loss": 0.9334878921508789,
      "step": 967
    },
    {
      "epoch": 0.4661409739360742,
      "grad_norm": 2.298605442047119,
      "learning_rate": 0.00011599769974219757,
      "loss": 0.3982749283313751,
      "step": 968
    },
    {
      "epoch": 0.46662252452898334,
      "grad_norm": 5.029486179351807,
      "learning_rate": 0.0001158437016290539,
      "loss": 1.0167627334594727,
      "step": 969
    },
    {
      "epoch": 0.4671040751218925,
      "grad_norm": 2.409154176712036,
      "learning_rate": 0.00011568966496475649,
      "loss": 0.5409466028213501,
      "step": 970
    },
    {
      "epoch": 0.46758562571480167,
      "grad_norm": 2.911445140838623,
      "learning_rate": 0.00011553559012410984,
      "loss": 0.9783698916435242,
      "step": 971
    },
    {
      "epoch": 0.46806717630771083,
      "grad_norm": 3.0666348934173584,
      "learning_rate": 0.00011538147748201138,
      "loss": 1.022337555885315,
      "step": 972
    },
    {
      "epoch": 0.46854872690062,
      "grad_norm": 1.7170687913894653,
      "learning_rate": 0.00011522732741345053,
      "loss": 0.7173493504524231,
      "step": 973
    },
    {
      "epoch": 0.46903027749352916,
      "grad_norm": 1.648453712463379,
      "learning_rate": 0.00011507314029350776,
      "loss": 0.5984148979187012,
      "step": 974
    },
    {
      "epoch": 0.4695118280864383,
      "grad_norm": 9.832086563110352,
      "learning_rate": 0.00011491891649735366,
      "loss": 0.8754693865776062,
      "step": 975
    },
    {
      "epoch": 0.4699933786793475,
      "grad_norm": 2.168177604675293,
      "learning_rate": 0.00011476465640024814,
      "loss": 0.8928840160369873,
      "step": 976
    },
    {
      "epoch": 0.47047492927225665,
      "grad_norm": 2.2388288974761963,
      "learning_rate": 0.00011461036037753934,
      "loss": 0.6539017558097839,
      "step": 977
    },
    {
      "epoch": 0.4709564798651658,
      "grad_norm": 2.8901429176330566,
      "learning_rate": 0.00011445602880466288,
      "loss": 0.8645316362380981,
      "step": 978
    },
    {
      "epoch": 0.471438030458075,
      "grad_norm": 2.232909917831421,
      "learning_rate": 0.00011430166205714088,
      "loss": 0.8136779069900513,
      "step": 979
    },
    {
      "epoch": 0.4719195810509842,
      "grad_norm": 3.610781669616699,
      "learning_rate": 0.00011414726051058102,
      "loss": 0.5626087188720703,
      "step": 980
    },
    {
      "epoch": 0.47240113164389336,
      "grad_norm": 2.35752534866333,
      "learning_rate": 0.0001139928245406757,
      "loss": 0.6648074388504028,
      "step": 981
    },
    {
      "epoch": 0.4728826822368025,
      "grad_norm": 2.00398325920105,
      "learning_rate": 0.00011383835452320097,
      "loss": 1.2014191150665283,
      "step": 982
    },
    {
      "epoch": 0.4733642328297117,
      "grad_norm": 2.0243752002716064,
      "learning_rate": 0.00011368385083401585,
      "loss": 0.9910991191864014,
      "step": 983
    },
    {
      "epoch": 0.47384578342262085,
      "grad_norm": 1.3499548435211182,
      "learning_rate": 0.00011352931384906125,
      "loss": 0.6833926439285278,
      "step": 984
    },
    {
      "epoch": 0.47432733401553,
      "grad_norm": 2.449747323989868,
      "learning_rate": 0.00011337474394435908,
      "loss": 0.9368841648101807,
      "step": 985
    },
    {
      "epoch": 0.4748088846084392,
      "grad_norm": 2.3449792861938477,
      "learning_rate": 0.00011322014149601136,
      "loss": 0.9484957456588745,
      "step": 986
    },
    {
      "epoch": 0.47529043520134834,
      "grad_norm": 3.125023603439331,
      "learning_rate": 0.00011306550688019926,
      "loss": 0.91358482837677,
      "step": 987
    },
    {
      "epoch": 0.4757719857942575,
      "grad_norm": 2.818984270095825,
      "learning_rate": 0.0001129108404731823,
      "loss": 0.8100882768630981,
      "step": 988
    },
    {
      "epoch": 0.47625353638716666,
      "grad_norm": 2.428812026977539,
      "learning_rate": 0.0001127561426512973,
      "loss": 1.0841569900512695,
      "step": 989
    },
    {
      "epoch": 0.47673508698007583,
      "grad_norm": 1.7471508979797363,
      "learning_rate": 0.0001126014137909575,
      "loss": 0.8236819505691528,
      "step": 990
    },
    {
      "epoch": 0.477216637572985,
      "grad_norm": 2.4005544185638428,
      "learning_rate": 0.00011244665426865174,
      "loss": 1.4280376434326172,
      "step": 991
    },
    {
      "epoch": 0.47769818816589416,
      "grad_norm": 1.483923077583313,
      "learning_rate": 0.00011229186446094338,
      "loss": 0.6031376123428345,
      "step": 992
    },
    {
      "epoch": 0.4781797387588034,
      "grad_norm": 2.8969626426696777,
      "learning_rate": 0.00011213704474446951,
      "loss": 1.460785984992981,
      "step": 993
    },
    {
      "epoch": 0.47866128935171254,
      "grad_norm": 2.5944559574127197,
      "learning_rate": 0.00011198219549594,
      "loss": 0.8315947651863098,
      "step": 994
    },
    {
      "epoch": 0.4791428399446217,
      "grad_norm": 2.070571184158325,
      "learning_rate": 0.00011182731709213659,
      "loss": 0.737067699432373,
      "step": 995
    },
    {
      "epoch": 0.47962439053753086,
      "grad_norm": 2.449599266052246,
      "learning_rate": 0.00011167240990991192,
      "loss": 0.47890785336494446,
      "step": 996
    },
    {
      "epoch": 0.48010594113044003,
      "grad_norm": 3.1011149883270264,
      "learning_rate": 0.00011151747432618871,
      "loss": 0.8006799817085266,
      "step": 997
    },
    {
      "epoch": 0.4805874917233492,
      "grad_norm": 1.2630785703659058,
      "learning_rate": 0.00011136251071795871,
      "loss": 0.7390092015266418,
      "step": 998
    },
    {
      "epoch": 0.48106904231625836,
      "grad_norm": 2.8609020709991455,
      "learning_rate": 0.00011120751946228197,
      "loss": 0.9070338010787964,
      "step": 999
    },
    {
      "epoch": 0.4815505929091675,
      "grad_norm": 3.1906204223632812,
      "learning_rate": 0.00011105250093628565,
      "loss": 1.106128215789795,
      "step": 1000
    },
    {
      "epoch": 0.4820321435020767,
      "grad_norm": 3.9935097694396973,
      "learning_rate": 0.00011089745551716344,
      "loss": 1.0615726709365845,
      "step": 1001
    },
    {
      "epoch": 0.48251369409498585,
      "grad_norm": 2.0722532272338867,
      "learning_rate": 0.00011074238358217437,
      "loss": 1.059751272201538,
      "step": 1002
    },
    {
      "epoch": 0.482995244687895,
      "grad_norm": 2.5334818363189697,
      "learning_rate": 0.00011058728550864197,
      "loss": 0.6454775929450989,
      "step": 1003
    },
    {
      "epoch": 0.4834767952808042,
      "grad_norm": 1.7356927394866943,
      "learning_rate": 0.00011043216167395344,
      "loss": 0.7123833894729614,
      "step": 1004
    },
    {
      "epoch": 0.48395834587371334,
      "grad_norm": 2.0227949619293213,
      "learning_rate": 0.00011027701245555865,
      "loss": 0.6236705780029297,
      "step": 1005
    },
    {
      "epoch": 0.4844398964666225,
      "grad_norm": 2.2741832733154297,
      "learning_rate": 0.00011012183823096917,
      "loss": 0.7136357426643372,
      "step": 1006
    },
    {
      "epoch": 0.4849214470595317,
      "grad_norm": 6.329826354980469,
      "learning_rate": 0.00010996663937775751,
      "loss": 0.9109950661659241,
      "step": 1007
    },
    {
      "epoch": 0.4854029976524409,
      "grad_norm": 3.8215112686157227,
      "learning_rate": 0.000109811416273556,
      "loss": 1.312212347984314,
      "step": 1008
    },
    {
      "epoch": 0.48588454824535005,
      "grad_norm": 2.0453786849975586,
      "learning_rate": 0.00010965616929605609,
      "loss": 0.6476212739944458,
      "step": 1009
    },
    {
      "epoch": 0.4863660988382592,
      "grad_norm": 4.787343502044678,
      "learning_rate": 0.0001095008988230072,
      "loss": 1.1054847240447998,
      "step": 1010
    },
    {
      "epoch": 0.48684764943116837,
      "grad_norm": 2.2525241374969482,
      "learning_rate": 0.00010934560523221602,
      "loss": 0.9463966488838196,
      "step": 1011
    },
    {
      "epoch": 0.48732920002407754,
      "grad_norm": 2.4682202339172363,
      "learning_rate": 0.00010919028890154543,
      "loss": 0.7147688269615173,
      "step": 1012
    },
    {
      "epoch": 0.4878107506169867,
      "grad_norm": 2.827991485595703,
      "learning_rate": 0.00010903495020891375,
      "loss": 0.6355025172233582,
      "step": 1013
    },
    {
      "epoch": 0.48829230120989586,
      "grad_norm": 1.2764986753463745,
      "learning_rate": 0.00010887958953229349,
      "loss": 1.1802456378936768,
      "step": 1014
    },
    {
      "epoch": 0.488773851802805,
      "grad_norm": 1.4150707721710205,
      "learning_rate": 0.00010872420724971088,
      "loss": 0.6859856247901917,
      "step": 1015
    },
    {
      "epoch": 0.4892554023957142,
      "grad_norm": 4.007930278778076,
      "learning_rate": 0.0001085688037392446,
      "loss": 0.8577039241790771,
      "step": 1016
    },
    {
      "epoch": 0.48973695298862335,
      "grad_norm": 2.009462833404541,
      "learning_rate": 0.000108413379379025,
      "loss": 0.5408729314804077,
      "step": 1017
    },
    {
      "epoch": 0.4902185035815325,
      "grad_norm": 2.3193359375,
      "learning_rate": 0.00010825793454723325,
      "loss": 1.0953378677368164,
      "step": 1018
    },
    {
      "epoch": 0.4907000541744417,
      "grad_norm": 1.2775864601135254,
      "learning_rate": 0.00010810246962210018,
      "loss": 0.6733578443527222,
      "step": 1019
    },
    {
      "epoch": 0.49118160476735084,
      "grad_norm": 1.7805418968200684,
      "learning_rate": 0.00010794698498190557,
      "loss": 0.5374642014503479,
      "step": 1020
    },
    {
      "epoch": 0.49166315536026006,
      "grad_norm": 1.5644253492355347,
      "learning_rate": 0.00010779148100497722,
      "loss": 0.9437937140464783,
      "step": 1021
    },
    {
      "epoch": 0.4921447059531692,
      "grad_norm": 3.08471417427063,
      "learning_rate": 0.00010763595806968996,
      "loss": 1.0844614505767822,
      "step": 1022
    },
    {
      "epoch": 0.4926262565460784,
      "grad_norm": 2.5671825408935547,
      "learning_rate": 0.00010748041655446473,
      "loss": 1.0744214057922363,
      "step": 1023
    },
    {
      "epoch": 0.49310780713898755,
      "grad_norm": 4.144514560699463,
      "learning_rate": 0.00010732485683776768,
      "loss": 0.9581397175788879,
      "step": 1024
    },
    {
      "epoch": 0.4935893577318967,
      "grad_norm": 4.9848785400390625,
      "learning_rate": 0.00010716927929810925,
      "loss": 1.303045392036438,
      "step": 1025
    },
    {
      "epoch": 0.4940709083248059,
      "grad_norm": 2.6964612007141113,
      "learning_rate": 0.00010701368431404326,
      "loss": 1.090564489364624,
      "step": 1026
    },
    {
      "epoch": 0.49455245891771504,
      "grad_norm": 2.313910484313965,
      "learning_rate": 0.00010685807226416598,
      "loss": 0.843859076499939,
      "step": 1027
    },
    {
      "epoch": 0.4950340095106242,
      "grad_norm": 1.8410593271255493,
      "learning_rate": 0.00010670244352711518,
      "loss": 0.7944561839103699,
      "step": 1028
    },
    {
      "epoch": 0.49551556010353337,
      "grad_norm": 2.529728412628174,
      "learning_rate": 0.00010654679848156925,
      "loss": 0.5943077206611633,
      "step": 1029
    },
    {
      "epoch": 0.49599711069644253,
      "grad_norm": 2.1078760623931885,
      "learning_rate": 0.00010639113750624625,
      "loss": 0.6243797540664673,
      "step": 1030
    },
    {
      "epoch": 0.4964786612893517,
      "grad_norm": 3.5972940921783447,
      "learning_rate": 0.00010623546097990303,
      "loss": 1.0445196628570557,
      "step": 1031
    },
    {
      "epoch": 0.49696021188226086,
      "grad_norm": 1.6727979183197021,
      "learning_rate": 0.00010607976928133423,
      "loss": 0.9892148971557617,
      "step": 1032
    },
    {
      "epoch": 0.49744176247517,
      "grad_norm": 1.7264652252197266,
      "learning_rate": 0.00010592406278937144,
      "loss": 0.8948615789413452,
      "step": 1033
    },
    {
      "epoch": 0.4979233130680792,
      "grad_norm": 2.3174989223480225,
      "learning_rate": 0.00010576834188288226,
      "loss": 0.7108620405197144,
      "step": 1034
    },
    {
      "epoch": 0.4984048636609884,
      "grad_norm": 3.0876448154449463,
      "learning_rate": 0.00010561260694076935,
      "loss": 0.8531850576400757,
      "step": 1035
    },
    {
      "epoch": 0.49888641425389757,
      "grad_norm": 4.166559219360352,
      "learning_rate": 0.00010545685834196948,
      "loss": 1.3934071063995361,
      "step": 1036
    },
    {
      "epoch": 0.49936796484680673,
      "grad_norm": 1.7676146030426025,
      "learning_rate": 0.00010530109646545272,
      "loss": 0.542904794216156,
      "step": 1037
    },
    {
      "epoch": 0.4998495154397159,
      "grad_norm": 3.6352951526641846,
      "learning_rate": 0.0001051453216902214,
      "loss": 1.1154600381851196,
      "step": 1038
    },
    {
      "epoch": 0.500331066032625,
      "grad_norm": 1.6613993644714355,
      "learning_rate": 0.00010498953439530925,
      "loss": 1.0085630416870117,
      "step": 1039
    },
    {
      "epoch": 0.5008126166255342,
      "grad_norm": 2.341618776321411,
      "learning_rate": 0.00010483373495978046,
      "loss": 0.4881035089492798,
      "step": 1040
    },
    {
      "epoch": 0.5012941672184434,
      "grad_norm": 3.430499792098999,
      "learning_rate": 0.00010467792376272877,
      "loss": 0.6486775279045105,
      "step": 1041
    },
    {
      "epoch": 0.5017757178113526,
      "grad_norm": 2.436282157897949,
      "learning_rate": 0.00010452210118327652,
      "loss": 0.6542079448699951,
      "step": 1042
    },
    {
      "epoch": 0.5022572684042618,
      "grad_norm": 2.2035601139068604,
      "learning_rate": 0.00010436626760057378,
      "loss": 0.70213782787323,
      "step": 1043
    },
    {
      "epoch": 0.5027388189971709,
      "grad_norm": 3.07226824760437,
      "learning_rate": 0.00010421042339379732,
      "loss": 0.9099203944206238,
      "step": 1044
    },
    {
      "epoch": 0.5032203695900801,
      "grad_norm": 3.180781602859497,
      "learning_rate": 0.00010405456894214987,
      "loss": 0.6206411123275757,
      "step": 1045
    },
    {
      "epoch": 0.5037019201829892,
      "grad_norm": 4.420149803161621,
      "learning_rate": 0.00010389870462485902,
      "loss": 1.7011241912841797,
      "step": 1046
    },
    {
      "epoch": 0.5041834707758984,
      "grad_norm": 2.2351741790771484,
      "learning_rate": 0.00010374283082117635,
      "loss": 0.5734311938285828,
      "step": 1047
    },
    {
      "epoch": 0.5046650213688075,
      "grad_norm": 2.9196605682373047,
      "learning_rate": 0.00010358694791037653,
      "loss": 0.7947977781295776,
      "step": 1048
    },
    {
      "epoch": 0.5051465719617168,
      "grad_norm": 2.5870511531829834,
      "learning_rate": 0.00010343105627175644,
      "loss": 0.8223186731338501,
      "step": 1049
    },
    {
      "epoch": 0.5056281225546259,
      "grad_norm": 2.015101432800293,
      "learning_rate": 0.00010327515628463415,
      "loss": 0.7247661352157593,
      "step": 1050
    },
    {
      "epoch": 0.5061096731475351,
      "grad_norm": 2.108649253845215,
      "learning_rate": 0.00010311924832834808,
      "loss": 0.9195833802223206,
      "step": 1051
    },
    {
      "epoch": 0.5065912237404442,
      "grad_norm": 2.794048309326172,
      "learning_rate": 0.00010296333278225599,
      "loss": 0.6164246797561646,
      "step": 1052
    },
    {
      "epoch": 0.5070727743333534,
      "grad_norm": 2.6275827884674072,
      "learning_rate": 0.00010280741002573413,
      "loss": 0.4217431843280792,
      "step": 1053
    },
    {
      "epoch": 0.5075543249262625,
      "grad_norm": 2.049316167831421,
      "learning_rate": 0.00010265148043817632,
      "loss": 1.3008800745010376,
      "step": 1054
    },
    {
      "epoch": 0.5080358755191717,
      "grad_norm": 1.6879538297653198,
      "learning_rate": 0.00010249554439899298,
      "loss": 1.1763681173324585,
      "step": 1055
    },
    {
      "epoch": 0.508517426112081,
      "grad_norm": 2.6271748542785645,
      "learning_rate": 0.00010233960228761022,
      "loss": 0.9513725638389587,
      "step": 1056
    },
    {
      "epoch": 0.5089989767049901,
      "grad_norm": 1.9166399240493774,
      "learning_rate": 0.00010218365448346893,
      "loss": 0.8998247981071472,
      "step": 1057
    },
    {
      "epoch": 0.5094805272978993,
      "grad_norm": 3.4821462631225586,
      "learning_rate": 0.00010202770136602388,
      "loss": 1.230936884880066,
      "step": 1058
    },
    {
      "epoch": 0.5099620778908084,
      "grad_norm": 2.641911745071411,
      "learning_rate": 0.00010187174331474271,
      "loss": 0.6528239250183105,
      "step": 1059
    },
    {
      "epoch": 0.5104436284837176,
      "grad_norm": 1.5440243482589722,
      "learning_rate": 0.00010171578070910512,
      "loss": 0.5896283388137817,
      "step": 1060
    },
    {
      "epoch": 0.5109251790766267,
      "grad_norm": 2.0644617080688477,
      "learning_rate": 0.00010155981392860185,
      "loss": 0.7703726887702942,
      "step": 1061
    },
    {
      "epoch": 0.5114067296695359,
      "grad_norm": 3.1954174041748047,
      "learning_rate": 0.00010140384335273386,
      "loss": 0.9193657040596008,
      "step": 1062
    },
    {
      "epoch": 0.511888280262445,
      "grad_norm": 2.789104700088501,
      "learning_rate": 0.00010124786936101127,
      "loss": 0.9817596673965454,
      "step": 1063
    },
    {
      "epoch": 0.5123698308553543,
      "grad_norm": 2.242867946624756,
      "learning_rate": 0.00010109189233295255,
      "loss": 1.5660446882247925,
      "step": 1064
    },
    {
      "epoch": 0.5128513814482634,
      "grad_norm": 2.287654399871826,
      "learning_rate": 0.00010093591264808358,
      "loss": 0.6991084814071655,
      "step": 1065
    },
    {
      "epoch": 0.5133329320411726,
      "grad_norm": 1.873935580253601,
      "learning_rate": 0.00010077993068593663,
      "loss": 1.168310523033142,
      "step": 1066
    },
    {
      "epoch": 0.5138144826340817,
      "grad_norm": 1.3550395965576172,
      "learning_rate": 0.00010062394682604963,
      "loss": 0.45806559920310974,
      "step": 1067
    },
    {
      "epoch": 0.5142960332269909,
      "grad_norm": 2.6076066493988037,
      "learning_rate": 0.00010046796144796497,
      "loss": 0.8964013457298279,
      "step": 1068
    },
    {
      "epoch": 0.5147775838199001,
      "grad_norm": 2.46705961227417,
      "learning_rate": 0.0001003119749312289,
      "loss": 0.8759859800338745,
      "step": 1069
    },
    {
      "epoch": 0.5152591344128092,
      "grad_norm": 2.0368616580963135,
      "learning_rate": 0.00010015598765539031,
      "loss": 0.7912530303001404,
      "step": 1070
    },
    {
      "epoch": 0.5157406850057185,
      "grad_norm": 2.2679877281188965,
      "learning_rate": 0.0001,
      "loss": 0.7956958413124084,
      "step": 1071
    },
    {
      "epoch": 0.5162222355986276,
      "grad_norm": 2.586744785308838,
      "learning_rate": 9.984401234460971e-05,
      "loss": 1.0444368124008179,
      "step": 1072
    },
    {
      "epoch": 0.5167037861915368,
      "grad_norm": 4.488743305206299,
      "learning_rate": 9.968802506877111e-05,
      "loss": 1.7576448917388916,
      "step": 1073
    },
    {
      "epoch": 0.5171853367844459,
      "grad_norm": 2.6637063026428223,
      "learning_rate": 9.953203855203504e-05,
      "loss": 0.5657933354377747,
      "step": 1074
    },
    {
      "epoch": 0.5176668873773551,
      "grad_norm": 0.839038610458374,
      "learning_rate": 9.93760531739504e-05,
      "loss": 0.5093435645103455,
      "step": 1075
    },
    {
      "epoch": 0.5181484379702642,
      "grad_norm": 2.691375732421875,
      "learning_rate": 9.922006931406338e-05,
      "loss": 1.2426650524139404,
      "step": 1076
    },
    {
      "epoch": 0.5186299885631734,
      "grad_norm": 2.898726463317871,
      "learning_rate": 9.906408735191643e-05,
      "loss": 1.0973114967346191,
      "step": 1077
    },
    {
      "epoch": 0.5191115391560825,
      "grad_norm": 3.0733864307403564,
      "learning_rate": 9.890810766704745e-05,
      "loss": 0.9917636513710022,
      "step": 1078
    },
    {
      "epoch": 0.5195930897489918,
      "grad_norm": 3.097318410873413,
      "learning_rate": 9.875213063898875e-05,
      "loss": 0.54969322681427,
      "step": 1079
    },
    {
      "epoch": 0.5200746403419009,
      "grad_norm": 3.345762014389038,
      "learning_rate": 9.859615664726615e-05,
      "loss": 0.9232475161552429,
      "step": 1080
    },
    {
      "epoch": 0.5205561909348101,
      "grad_norm": 1.6197614669799805,
      "learning_rate": 9.844018607139818e-05,
      "loss": 0.6463713049888611,
      "step": 1081
    },
    {
      "epoch": 0.5210377415277193,
      "grad_norm": 2.2506837844848633,
      "learning_rate": 9.828421929089493e-05,
      "loss": 0.7813989520072937,
      "step": 1082
    },
    {
      "epoch": 0.5215192921206284,
      "grad_norm": 3.0212738513946533,
      "learning_rate": 9.812825668525733e-05,
      "loss": 0.6759187579154968,
      "step": 1083
    },
    {
      "epoch": 0.5220008427135376,
      "grad_norm": 1.6942336559295654,
      "learning_rate": 9.797229863397615e-05,
      "loss": 0.38814470171928406,
      "step": 1084
    },
    {
      "epoch": 0.5224823933064467,
      "grad_norm": 1.8985040187835693,
      "learning_rate": 9.781634551653108e-05,
      "loss": 0.866794228553772,
      "step": 1085
    },
    {
      "epoch": 0.522963943899356,
      "grad_norm": 1.7284749746322632,
      "learning_rate": 9.766039771238982e-05,
      "loss": 0.8945998549461365,
      "step": 1086
    },
    {
      "epoch": 0.5234454944922651,
      "grad_norm": 1.0572261810302734,
      "learning_rate": 9.750445560100706e-05,
      "loss": 0.8334730863571167,
      "step": 1087
    },
    {
      "epoch": 0.5239270450851743,
      "grad_norm": 1.5680509805679321,
      "learning_rate": 9.73485195618237e-05,
      "loss": 0.6145600080490112,
      "step": 1088
    },
    {
      "epoch": 0.5244085956780834,
      "grad_norm": 1.5271968841552734,
      "learning_rate": 9.719258997426588e-05,
      "loss": 0.6018796563148499,
      "step": 1089
    },
    {
      "epoch": 0.5248901462709926,
      "grad_norm": 3.9521241188049316,
      "learning_rate": 9.703666721774402e-05,
      "loss": 0.8241912722587585,
      "step": 1090
    },
    {
      "epoch": 0.5253716968639017,
      "grad_norm": 1.552260160446167,
      "learning_rate": 9.688075167165194e-05,
      "loss": 0.6726410388946533,
      "step": 1091
    },
    {
      "epoch": 0.525853247456811,
      "grad_norm": 3.280787944793701,
      "learning_rate": 9.672484371536586e-05,
      "loss": 0.8309072852134705,
      "step": 1092
    },
    {
      "epoch": 0.52633479804972,
      "grad_norm": 1.4258042573928833,
      "learning_rate": 9.656894372824358e-05,
      "loss": 0.7800561189651489,
      "step": 1093
    },
    {
      "epoch": 0.5268163486426293,
      "grad_norm": 2.567262649536133,
      "learning_rate": 9.64130520896235e-05,
      "loss": 0.6320695877075195,
      "step": 1094
    },
    {
      "epoch": 0.5272978992355384,
      "grad_norm": 1.290222406387329,
      "learning_rate": 9.625716917882367e-05,
      "loss": 0.6697812080383301,
      "step": 1095
    },
    {
      "epoch": 0.5277794498284476,
      "grad_norm": 2.8315935134887695,
      "learning_rate": 9.6101295375141e-05,
      "loss": 0.9362088441848755,
      "step": 1096
    },
    {
      "epoch": 0.5282610004213568,
      "grad_norm": 1.239498496055603,
      "learning_rate": 9.594543105785013e-05,
      "loss": 0.6236352920532227,
      "step": 1097
    },
    {
      "epoch": 0.5287425510142659,
      "grad_norm": 3.848062038421631,
      "learning_rate": 9.578957660620267e-05,
      "loss": 1.053107500076294,
      "step": 1098
    },
    {
      "epoch": 0.5292241016071751,
      "grad_norm": 3.380949020385742,
      "learning_rate": 9.563373239942623e-05,
      "loss": 0.7207359671592712,
      "step": 1099
    },
    {
      "epoch": 0.5297056522000843,
      "grad_norm": 1.3245621919631958,
      "learning_rate": 9.547789881672348e-05,
      "loss": 0.5151536464691162,
      "step": 1100
    },
    {
      "epoch": 0.5301872027929935,
      "grad_norm": 2.008918285369873,
      "learning_rate": 9.532207623727126e-05,
      "loss": 0.8836523294448853,
      "step": 1101
    },
    {
      "epoch": 0.5306687533859026,
      "grad_norm": 2.313354015350342,
      "learning_rate": 9.516626504021957e-05,
      "loss": 0.5717525482177734,
      "step": 1102
    },
    {
      "epoch": 0.5311503039788118,
      "grad_norm": 2.152496337890625,
      "learning_rate": 9.501046560469079e-05,
      "loss": 0.677817165851593,
      "step": 1103
    },
    {
      "epoch": 0.5316318545717209,
      "grad_norm": 2.158090829849243,
      "learning_rate": 9.485467830977864e-05,
      "loss": 0.6780133843421936,
      "step": 1104
    },
    {
      "epoch": 0.5321134051646301,
      "grad_norm": 5.504623889923096,
      "learning_rate": 9.469890353454732e-05,
      "loss": 1.0965919494628906,
      "step": 1105
    },
    {
      "epoch": 0.5325949557575392,
      "grad_norm": 2.040245532989502,
      "learning_rate": 9.454314165803054e-05,
      "loss": 0.7461580634117126,
      "step": 1106
    },
    {
      "epoch": 0.5330765063504485,
      "grad_norm": 3.3329811096191406,
      "learning_rate": 9.438739305923067e-05,
      "loss": 0.7621337175369263,
      "step": 1107
    },
    {
      "epoch": 0.5335580569433576,
      "grad_norm": 1.8127444982528687,
      "learning_rate": 9.423165811711777e-05,
      "loss": 0.8671072721481323,
      "step": 1108
    },
    {
      "epoch": 0.5340396075362668,
      "grad_norm": 2.84906005859375,
      "learning_rate": 9.407593721062859e-05,
      "loss": 0.6759154200553894,
      "step": 1109
    },
    {
      "epoch": 0.534521158129176,
      "grad_norm": 1.8998093605041504,
      "learning_rate": 9.39202307186658e-05,
      "loss": 0.5612306594848633,
      "step": 1110
    },
    {
      "epoch": 0.5350027087220851,
      "grad_norm": 3.594270706176758,
      "learning_rate": 9.3764539020097e-05,
      "loss": 1.3456085920333862,
      "step": 1111
    },
    {
      "epoch": 0.5354842593149943,
      "grad_norm": 2.0066394805908203,
      "learning_rate": 9.360886249375376e-05,
      "loss": 1.008877158164978,
      "step": 1112
    },
    {
      "epoch": 0.5359658099079034,
      "grad_norm": 3.534285306930542,
      "learning_rate": 9.345320151843078e-05,
      "loss": 0.681387722492218,
      "step": 1113
    },
    {
      "epoch": 0.5364473605008127,
      "grad_norm": 1.2602880001068115,
      "learning_rate": 9.329755647288485e-05,
      "loss": 0.5982921719551086,
      "step": 1114
    },
    {
      "epoch": 0.5369289110937218,
      "grad_norm": 2.8723275661468506,
      "learning_rate": 9.314192773583403e-05,
      "loss": 1.0891187191009521,
      "step": 1115
    },
    {
      "epoch": 0.537410461686631,
      "grad_norm": 1.634235143661499,
      "learning_rate": 9.298631568595674e-05,
      "loss": 0.49135875701904297,
      "step": 1116
    },
    {
      "epoch": 0.5378920122795401,
      "grad_norm": 3.787536382675171,
      "learning_rate": 9.283072070189075e-05,
      "loss": 0.6476027369499207,
      "step": 1117
    },
    {
      "epoch": 0.5383735628724493,
      "grad_norm": 1.2070212364196777,
      "learning_rate": 9.267514316223234e-05,
      "loss": 0.4674101173877716,
      "step": 1118
    },
    {
      "epoch": 0.5388551134653584,
      "grad_norm": 2.6074249744415283,
      "learning_rate": 9.251958344553528e-05,
      "loss": 0.7353014945983887,
      "step": 1119
    },
    {
      "epoch": 0.5393366640582676,
      "grad_norm": 1.2813712358474731,
      "learning_rate": 9.23640419303101e-05,
      "loss": 0.4414594769477844,
      "step": 1120
    },
    {
      "epoch": 0.5398182146511767,
      "grad_norm": 3.1918694972991943,
      "learning_rate": 9.220851899502283e-05,
      "loss": 0.9425276517868042,
      "step": 1121
    },
    {
      "epoch": 0.540299765244086,
      "grad_norm": 2.1694791316986084,
      "learning_rate": 9.205301501809448e-05,
      "loss": 0.5796248316764832,
      "step": 1122
    },
    {
      "epoch": 0.5407813158369951,
      "grad_norm": 6.8798627853393555,
      "learning_rate": 9.189753037789987e-05,
      "loss": 1.253668189048767,
      "step": 1123
    },
    {
      "epoch": 0.5412628664299043,
      "grad_norm": 1.3222750425338745,
      "learning_rate": 9.174206545276677e-05,
      "loss": 0.7048219442367554,
      "step": 1124
    },
    {
      "epoch": 0.5417444170228135,
      "grad_norm": 1.9702863693237305,
      "learning_rate": 9.158662062097501e-05,
      "loss": 0.8644969463348389,
      "step": 1125
    },
    {
      "epoch": 0.5422259676157226,
      "grad_norm": 2.6311800479888916,
      "learning_rate": 9.143119626075542e-05,
      "loss": 1.1505922079086304,
      "step": 1126
    },
    {
      "epoch": 0.5427075182086318,
      "grad_norm": 2.7044169902801514,
      "learning_rate": 9.127579275028914e-05,
      "loss": 1.1839011907577515,
      "step": 1127
    },
    {
      "epoch": 0.5431890688015409,
      "grad_norm": 1.588924765586853,
      "learning_rate": 9.112041046770653e-05,
      "loss": 1.02664053440094,
      "step": 1128
    },
    {
      "epoch": 0.5436706193944502,
      "grad_norm": 1.242644190788269,
      "learning_rate": 9.096504979108629e-05,
      "loss": 0.9475134015083313,
      "step": 1129
    },
    {
      "epoch": 0.5441521699873593,
      "grad_norm": 2.0665194988250732,
      "learning_rate": 9.080971109845458e-05,
      "loss": 1.0283668041229248,
      "step": 1130
    },
    {
      "epoch": 0.5446337205802685,
      "grad_norm": 1.7460771799087524,
      "learning_rate": 9.0654394767784e-05,
      "loss": 1.1584253311157227,
      "step": 1131
    },
    {
      "epoch": 0.5451152711731776,
      "grad_norm": 1.6345937252044678,
      "learning_rate": 9.049910117699281e-05,
      "loss": 0.748465359210968,
      "step": 1132
    },
    {
      "epoch": 0.5455968217660868,
      "grad_norm": 2.6525137424468994,
      "learning_rate": 9.034383070394393e-05,
      "loss": 1.1963119506835938,
      "step": 1133
    },
    {
      "epoch": 0.5460783723589959,
      "grad_norm": 1.5838695764541626,
      "learning_rate": 9.0188583726444e-05,
      "loss": 0.8124378323554993,
      "step": 1134
    },
    {
      "epoch": 0.5465599229519051,
      "grad_norm": 2.3336410522460938,
      "learning_rate": 9.00333606222425e-05,
      "loss": 0.6154911518096924,
      "step": 1135
    },
    {
      "epoch": 0.5470414735448142,
      "grad_norm": 3.852142095565796,
      "learning_rate": 8.987816176903082e-05,
      "loss": 1.235611081123352,
      "step": 1136
    },
    {
      "epoch": 0.5475230241377235,
      "grad_norm": 1.8240224123001099,
      "learning_rate": 8.972298754444136e-05,
      "loss": 0.6516082286834717,
      "step": 1137
    },
    {
      "epoch": 0.5480045747306327,
      "grad_norm": 2.7947466373443604,
      "learning_rate": 8.956783832604654e-05,
      "loss": 0.7206995487213135,
      "step": 1138
    },
    {
      "epoch": 0.5484861253235418,
      "grad_norm": 1.6543246507644653,
      "learning_rate": 8.941271449135806e-05,
      "loss": 0.7875213623046875,
      "step": 1139
    },
    {
      "epoch": 0.548967675916451,
      "grad_norm": 3.1106162071228027,
      "learning_rate": 8.925761641782567e-05,
      "loss": 0.9625113010406494,
      "step": 1140
    },
    {
      "epoch": 0.5494492265093601,
      "grad_norm": 2.879199504852295,
      "learning_rate": 8.910254448283659e-05,
      "loss": 0.6439436078071594,
      "step": 1141
    },
    {
      "epoch": 0.5499307771022693,
      "grad_norm": 1.7155425548553467,
      "learning_rate": 8.894749906371439e-05,
      "loss": 0.5890610218048096,
      "step": 1142
    },
    {
      "epoch": 0.5504123276951784,
      "grad_norm": 1.9412592649459839,
      "learning_rate": 8.87924805377181e-05,
      "loss": 0.42438191175460815,
      "step": 1143
    },
    {
      "epoch": 0.5508938782880877,
      "grad_norm": 3.774172782897949,
      "learning_rate": 8.863748928204131e-05,
      "loss": 0.6759569048881531,
      "step": 1144
    },
    {
      "epoch": 0.5513754288809968,
      "grad_norm": 2.8683576583862305,
      "learning_rate": 8.848252567381131e-05,
      "loss": 0.7741227149963379,
      "step": 1145
    },
    {
      "epoch": 0.551856979473906,
      "grad_norm": 1.7375504970550537,
      "learning_rate": 8.83275900900881e-05,
      "loss": 0.518896222114563,
      "step": 1146
    },
    {
      "epoch": 0.5523385300668151,
      "grad_norm": 4.198785781860352,
      "learning_rate": 8.817268290786343e-05,
      "loss": 0.8628325462341309,
      "step": 1147
    },
    {
      "epoch": 0.5528200806597243,
      "grad_norm": 2.672045946121216,
      "learning_rate": 8.801780450406002e-05,
      "loss": 0.7736381888389587,
      "step": 1148
    },
    {
      "epoch": 0.5533016312526334,
      "grad_norm": 2.145855188369751,
      "learning_rate": 8.786295525553053e-05,
      "loss": 0.8363268375396729,
      "step": 1149
    },
    {
      "epoch": 0.5537831818455426,
      "grad_norm": 1.9035331010818481,
      "learning_rate": 8.770813553905664e-05,
      "loss": 1.0555862188339233,
      "step": 1150
    },
    {
      "epoch": 0.5542647324384519,
      "grad_norm": 2.813894510269165,
      "learning_rate": 8.755334573134829e-05,
      "loss": 0.8247569799423218,
      "step": 1151
    },
    {
      "epoch": 0.554746283031361,
      "grad_norm": 2.5057973861694336,
      "learning_rate": 8.739858620904251e-05,
      "loss": 1.0859413146972656,
      "step": 1152
    },
    {
      "epoch": 0.5552278336242702,
      "grad_norm": 1.9540780782699585,
      "learning_rate": 8.724385734870271e-05,
      "loss": 1.2197812795639038,
      "step": 1153
    },
    {
      "epoch": 0.5557093842171793,
      "grad_norm": 1.9933992624282837,
      "learning_rate": 8.708915952681769e-05,
      "loss": 0.9951842427253723,
      "step": 1154
    },
    {
      "epoch": 0.5561909348100885,
      "grad_norm": 1.414975881576538,
      "learning_rate": 8.693449311980074e-05,
      "loss": 0.8459328413009644,
      "step": 1155
    },
    {
      "epoch": 0.5566724854029976,
      "grad_norm": 1.296830415725708,
      "learning_rate": 8.677985850398866e-05,
      "loss": 0.6186532974243164,
      "step": 1156
    },
    {
      "epoch": 0.5571540359959068,
      "grad_norm": 2.1881940364837646,
      "learning_rate": 8.662525605564093e-05,
      "loss": 0.6511512994766235,
      "step": 1157
    },
    {
      "epoch": 0.557635586588816,
      "grad_norm": 1.8507907390594482,
      "learning_rate": 8.647068615093875e-05,
      "loss": 1.050092101097107,
      "step": 1158
    },
    {
      "epoch": 0.5581171371817252,
      "grad_norm": 2.0452117919921875,
      "learning_rate": 8.631614916598419e-05,
      "loss": 0.6867355704307556,
      "step": 1159
    },
    {
      "epoch": 0.5585986877746343,
      "grad_norm": 3.1383056640625,
      "learning_rate": 8.616164547679906e-05,
      "loss": 0.9598536491394043,
      "step": 1160
    },
    {
      "epoch": 0.5590802383675435,
      "grad_norm": 3.5082128047943115,
      "learning_rate": 8.600717545932435e-05,
      "loss": 0.7770359516143799,
      "step": 1161
    },
    {
      "epoch": 0.5595617889604526,
      "grad_norm": 3.237555980682373,
      "learning_rate": 8.5852739489419e-05,
      "loss": 0.665165901184082,
      "step": 1162
    },
    {
      "epoch": 0.5600433395533618,
      "grad_norm": 15.439366340637207,
      "learning_rate": 8.569833794285915e-05,
      "loss": 0.7772696614265442,
      "step": 1163
    },
    {
      "epoch": 0.5605248901462709,
      "grad_norm": 1.0626342296600342,
      "learning_rate": 8.554397119533714e-05,
      "loss": 0.828998863697052,
      "step": 1164
    },
    {
      "epoch": 0.5610064407391802,
      "grad_norm": 2.143800973892212,
      "learning_rate": 8.538963962246069e-05,
      "loss": 0.6868883371353149,
      "step": 1165
    },
    {
      "epoch": 0.5614879913320894,
      "grad_norm": 4.05539083480835,
      "learning_rate": 8.523534359975189e-05,
      "loss": 0.7314295172691345,
      "step": 1166
    },
    {
      "epoch": 0.5619695419249985,
      "grad_norm": 2.270596742630005,
      "learning_rate": 8.508108350264635e-05,
      "loss": 0.7261523008346558,
      "step": 1167
    },
    {
      "epoch": 0.5624510925179077,
      "grad_norm": 3.601720094680786,
      "learning_rate": 8.492685970649228e-05,
      "loss": 1.0951374769210815,
      "step": 1168
    },
    {
      "epoch": 0.5629326431108168,
      "grad_norm": 1.5328428745269775,
      "learning_rate": 8.477267258654949e-05,
      "loss": 1.0970330238342285,
      "step": 1169
    },
    {
      "epoch": 0.563414193703726,
      "grad_norm": 6.813167095184326,
      "learning_rate": 8.461852251798866e-05,
      "loss": 0.9881186485290527,
      "step": 1170
    },
    {
      "epoch": 0.5638957442966351,
      "grad_norm": 1.7621617317199707,
      "learning_rate": 8.44644098758902e-05,
      "loss": 0.9316292405128479,
      "step": 1171
    },
    {
      "epoch": 0.5643772948895444,
      "grad_norm": 6.586948871612549,
      "learning_rate": 8.431033503524354e-05,
      "loss": 0.7971272468566895,
      "step": 1172
    },
    {
      "epoch": 0.5648588454824535,
      "grad_norm": 2.0917298793792725,
      "learning_rate": 8.415629837094611e-05,
      "loss": 1.0194541215896606,
      "step": 1173
    },
    {
      "epoch": 0.5653403960753627,
      "grad_norm": 2.8051044940948486,
      "learning_rate": 8.400230025780243e-05,
      "loss": 0.8005639910697937,
      "step": 1174
    },
    {
      "epoch": 0.5658219466682718,
      "grad_norm": 2.8116307258605957,
      "learning_rate": 8.384834107052321e-05,
      "loss": 0.7625358700752258,
      "step": 1175
    },
    {
      "epoch": 0.566303497261181,
      "grad_norm": 3.2399468421936035,
      "learning_rate": 8.369442118372447e-05,
      "loss": 0.9739376306533813,
      "step": 1176
    },
    {
      "epoch": 0.5667850478540901,
      "grad_norm": 1.1742022037506104,
      "learning_rate": 8.35405409719266e-05,
      "loss": 0.6930390000343323,
      "step": 1177
    },
    {
      "epoch": 0.5672665984469993,
      "grad_norm": 2.1908392906188965,
      "learning_rate": 8.338670080955349e-05,
      "loss": 0.5992050766944885,
      "step": 1178
    },
    {
      "epoch": 0.5677481490399086,
      "grad_norm": 2.0192325115203857,
      "learning_rate": 8.323290107093143e-05,
      "loss": 1.0240747928619385,
      "step": 1179
    },
    {
      "epoch": 0.5682296996328177,
      "grad_norm": 1.9871251583099365,
      "learning_rate": 8.307914213028856e-05,
      "loss": 0.7399676442146301,
      "step": 1180
    },
    {
      "epoch": 0.5687112502257269,
      "grad_norm": 2.819214105606079,
      "learning_rate": 8.292542436175356e-05,
      "loss": 1.2702053785324097,
      "step": 1181
    },
    {
      "epoch": 0.569192800818636,
      "grad_norm": 1.923147201538086,
      "learning_rate": 8.277174813935508e-05,
      "loss": 0.7850733995437622,
      "step": 1182
    },
    {
      "epoch": 0.5696743514115452,
      "grad_norm": 2.719367027282715,
      "learning_rate": 8.261811383702061e-05,
      "loss": 0.6898080706596375,
      "step": 1183
    },
    {
      "epoch": 0.5701559020044543,
      "grad_norm": 8.6840238571167,
      "learning_rate": 8.246452182857562e-05,
      "loss": 0.7938961982727051,
      "step": 1184
    },
    {
      "epoch": 0.5706374525973635,
      "grad_norm": 2.5878841876983643,
      "learning_rate": 8.231097248774274e-05,
      "loss": 0.6973369717597961,
      "step": 1185
    },
    {
      "epoch": 0.5711190031902726,
      "grad_norm": 2.47227144241333,
      "learning_rate": 8.215746618814067e-05,
      "loss": 0.6305696964263916,
      "step": 1186
    },
    {
      "epoch": 0.5716005537831819,
      "grad_norm": 3.4036216735839844,
      "learning_rate": 8.200400330328348e-05,
      "loss": 1.2133344411849976,
      "step": 1187
    },
    {
      "epoch": 0.572082104376091,
      "grad_norm": 1.6595686674118042,
      "learning_rate": 8.185058420657957e-05,
      "loss": 0.9603378772735596,
      "step": 1188
    },
    {
      "epoch": 0.5725636549690002,
      "grad_norm": 2.6230967044830322,
      "learning_rate": 8.16972092713308e-05,
      "loss": 0.696787416934967,
      "step": 1189
    },
    {
      "epoch": 0.5730452055619093,
      "grad_norm": 1.9638707637786865,
      "learning_rate": 8.154387887073158e-05,
      "loss": 0.6159970760345459,
      "step": 1190
    },
    {
      "epoch": 0.5735267561548185,
      "grad_norm": 1.7063583135604858,
      "learning_rate": 8.139059337786792e-05,
      "loss": 0.6273808479309082,
      "step": 1191
    },
    {
      "epoch": 0.5740083067477277,
      "grad_norm": 2.2902870178222656,
      "learning_rate": 8.12373531657166e-05,
      "loss": 0.6002784371376038,
      "step": 1192
    },
    {
      "epoch": 0.5744898573406368,
      "grad_norm": 1.6169003248214722,
      "learning_rate": 8.108415860714418e-05,
      "loss": 0.8624545335769653,
      "step": 1193
    },
    {
      "epoch": 0.5749714079335461,
      "grad_norm": 2.1170918941497803,
      "learning_rate": 8.093101007490622e-05,
      "loss": 0.905326783657074,
      "step": 1194
    },
    {
      "epoch": 0.5754529585264552,
      "grad_norm": 2.1471760272979736,
      "learning_rate": 8.077790794164619e-05,
      "loss": 0.6312342882156372,
      "step": 1195
    },
    {
      "epoch": 0.5759345091193644,
      "grad_norm": 3.737238645553589,
      "learning_rate": 8.062485257989471e-05,
      "loss": 1.0070664882659912,
      "step": 1196
    },
    {
      "epoch": 0.5764160597122735,
      "grad_norm": 2.7662718296051025,
      "learning_rate": 8.047184436206864e-05,
      "loss": 0.9549069404602051,
      "step": 1197
    },
    {
      "epoch": 0.5768976103051827,
      "grad_norm": 1.6752015352249146,
      "learning_rate": 8.031888366046998e-05,
      "loss": 0.4790445864200592,
      "step": 1198
    },
    {
      "epoch": 0.5773791608980918,
      "grad_norm": 3.104614019393921,
      "learning_rate": 8.016597084728526e-05,
      "loss": 0.9265120625495911,
      "step": 1199
    },
    {
      "epoch": 0.577860711491001,
      "grad_norm": 3.4039664268493652,
      "learning_rate": 8.001310629458443e-05,
      "loss": 1.2424333095550537,
      "step": 1200
    },
    {
      "epoch": 0.5783422620839102,
      "grad_norm": 2.116501569747925,
      "learning_rate": 7.986029037432002e-05,
      "loss": 1.3340606689453125,
      "step": 1201
    },
    {
      "epoch": 0.5788238126768194,
      "grad_norm": 3.4259469509124756,
      "learning_rate": 7.970752345832623e-05,
      "loss": 0.6767507195472717,
      "step": 1202
    },
    {
      "epoch": 0.5793053632697285,
      "grad_norm": 2.1776535511016846,
      "learning_rate": 7.9554805918318e-05,
      "loss": 0.6985357999801636,
      "step": 1203
    },
    {
      "epoch": 0.5797869138626377,
      "grad_norm": 2.03326678276062,
      "learning_rate": 7.940213812589018e-05,
      "loss": 0.8311415314674377,
      "step": 1204
    },
    {
      "epoch": 0.5802684644555468,
      "grad_norm": 0.9500651359558105,
      "learning_rate": 7.92495204525165e-05,
      "loss": 0.40140607953071594,
      "step": 1205
    },
    {
      "epoch": 0.580750015048456,
      "grad_norm": 2.4238381385803223,
      "learning_rate": 7.909695326954878e-05,
      "loss": 0.711327850818634,
      "step": 1206
    },
    {
      "epoch": 0.5812315656413652,
      "grad_norm": 2.3756251335144043,
      "learning_rate": 7.894443694821602e-05,
      "loss": 0.9460272789001465,
      "step": 1207
    },
    {
      "epoch": 0.5817131162342744,
      "grad_norm": 1.896917462348938,
      "learning_rate": 7.879197185962339e-05,
      "loss": 0.7879883646965027,
      "step": 1208
    },
    {
      "epoch": 0.5821946668271836,
      "grad_norm": 2.270230293273926,
      "learning_rate": 7.863955837475144e-05,
      "loss": 1.0135605335235596,
      "step": 1209
    },
    {
      "epoch": 0.5826762174200927,
      "grad_norm": 4.318577289581299,
      "learning_rate": 7.848719686445515e-05,
      "loss": 1.2733519077301025,
      "step": 1210
    },
    {
      "epoch": 0.5831577680130019,
      "grad_norm": 1.5087001323699951,
      "learning_rate": 7.833488769946306e-05,
      "loss": 1.406731367111206,
      "step": 1211
    },
    {
      "epoch": 0.583639318605911,
      "grad_norm": 2.8327383995056152,
      "learning_rate": 7.818263125037633e-05,
      "loss": 0.9411525726318359,
      "step": 1212
    },
    {
      "epoch": 0.5841208691988202,
      "grad_norm": 3.125861167907715,
      "learning_rate": 7.803042788766777e-05,
      "loss": 0.8555670380592346,
      "step": 1213
    },
    {
      "epoch": 0.5846024197917293,
      "grad_norm": 1.9403356313705444,
      "learning_rate": 7.787827798168115e-05,
      "loss": 0.6103419661521912,
      "step": 1214
    },
    {
      "epoch": 0.5850839703846386,
      "grad_norm": 1.446379542350769,
      "learning_rate": 7.772618190263009e-05,
      "loss": 0.7615257501602173,
      "step": 1215
    },
    {
      "epoch": 0.5855655209775477,
      "grad_norm": 2.8812882900238037,
      "learning_rate": 7.757414002059726e-05,
      "loss": 0.8465603590011597,
      "step": 1216
    },
    {
      "epoch": 0.5860470715704569,
      "grad_norm": 1.3343104124069214,
      "learning_rate": 7.742215270553349e-05,
      "loss": 0.6286378502845764,
      "step": 1217
    },
    {
      "epoch": 0.586528622163366,
      "grad_norm": 1.9730757474899292,
      "learning_rate": 7.727022032725672e-05,
      "loss": 0.8061712384223938,
      "step": 1218
    },
    {
      "epoch": 0.5870101727562752,
      "grad_norm": 2.839226245880127,
      "learning_rate": 7.711834325545135e-05,
      "loss": 0.6400884389877319,
      "step": 1219
    },
    {
      "epoch": 0.5874917233491844,
      "grad_norm": 1.4755616188049316,
      "learning_rate": 7.696652185966711e-05,
      "loss": 0.6182642579078674,
      "step": 1220
    },
    {
      "epoch": 0.5879732739420935,
      "grad_norm": 2.689772844314575,
      "learning_rate": 7.681475650931834e-05,
      "loss": 1.127432942390442,
      "step": 1221
    },
    {
      "epoch": 0.5884548245350028,
      "grad_norm": 6.828489303588867,
      "learning_rate": 7.666304757368297e-05,
      "loss": 0.9215345978736877,
      "step": 1222
    },
    {
      "epoch": 0.5889363751279119,
      "grad_norm": 3.5342772006988525,
      "learning_rate": 7.651139542190164e-05,
      "loss": 1.1560536623001099,
      "step": 1223
    },
    {
      "epoch": 0.5894179257208211,
      "grad_norm": 1.9722071886062622,
      "learning_rate": 7.635980042297687e-05,
      "loss": 0.726015031337738,
      "step": 1224
    },
    {
      "epoch": 0.5898994763137302,
      "grad_norm": 1.4691028594970703,
      "learning_rate": 7.620826294577208e-05,
      "loss": 0.4432721734046936,
      "step": 1225
    },
    {
      "epoch": 0.5903810269066394,
      "grad_norm": 2.7923078536987305,
      "learning_rate": 7.605678335901071e-05,
      "loss": 0.7103666067123413,
      "step": 1226
    },
    {
      "epoch": 0.5908625774995485,
      "grad_norm": 2.1559343338012695,
      "learning_rate": 7.59053620312754e-05,
      "loss": 0.8040168285369873,
      "step": 1227
    },
    {
      "epoch": 0.5913441280924577,
      "grad_norm": 19.709592819213867,
      "learning_rate": 7.575399933100697e-05,
      "loss": 0.8758564591407776,
      "step": 1228
    },
    {
      "epoch": 0.5918256786853668,
      "grad_norm": 2.7344141006469727,
      "learning_rate": 7.560269562650368e-05,
      "loss": 0.7944309711456299,
      "step": 1229
    },
    {
      "epoch": 0.5923072292782761,
      "grad_norm": 2.51283597946167,
      "learning_rate": 7.54514512859201e-05,
      "loss": 1.2313385009765625,
      "step": 1230
    },
    {
      "epoch": 0.5927887798711852,
      "grad_norm": 2.552638292312622,
      "learning_rate": 7.530026667726645e-05,
      "loss": 0.3206713795661926,
      "step": 1231
    },
    {
      "epoch": 0.5932703304640944,
      "grad_norm": 7.434596538543701,
      "learning_rate": 7.51491421684076e-05,
      "loss": 0.8569538593292236,
      "step": 1232
    },
    {
      "epoch": 0.5937518810570036,
      "grad_norm": 3.027329683303833,
      "learning_rate": 7.49980781270622e-05,
      "loss": 0.7863670587539673,
      "step": 1233
    },
    {
      "epoch": 0.5942334316499127,
      "grad_norm": 2.122748613357544,
      "learning_rate": 7.484707492080172e-05,
      "loss": 0.7018932104110718,
      "step": 1234
    },
    {
      "epoch": 0.5947149822428219,
      "grad_norm": 2.7135818004608154,
      "learning_rate": 7.469613291704962e-05,
      "loss": 0.6621221303939819,
      "step": 1235
    },
    {
      "epoch": 0.595196532835731,
      "grad_norm": 1.889063835144043,
      "learning_rate": 7.45452524830805e-05,
      "loss": 1.0902376174926758,
      "step": 1236
    },
    {
      "epoch": 0.5956780834286403,
      "grad_norm": 2.056445837020874,
      "learning_rate": 7.439443398601903e-05,
      "loss": 1.0527076721191406,
      "step": 1237
    },
    {
      "epoch": 0.5961596340215494,
      "grad_norm": 3.3864147663116455,
      "learning_rate": 7.424367779283926e-05,
      "loss": 1.0964192152023315,
      "step": 1238
    },
    {
      "epoch": 0.5966411846144586,
      "grad_norm": 2.0588386058807373,
      "learning_rate": 7.409298427036364e-05,
      "loss": 0.6538578867912292,
      "step": 1239
    },
    {
      "epoch": 0.5971227352073677,
      "grad_norm": 1.9627223014831543,
      "learning_rate": 7.39423537852621e-05,
      "loss": 0.5833548307418823,
      "step": 1240
    },
    {
      "epoch": 0.5976042858002769,
      "grad_norm": 3.9132301807403564,
      "learning_rate": 7.379178670405123e-05,
      "loss": 0.5358046293258667,
      "step": 1241
    },
    {
      "epoch": 0.598085836393186,
      "grad_norm": 2.2976489067077637,
      "learning_rate": 7.364128339309326e-05,
      "loss": 0.6436305046081543,
      "step": 1242
    },
    {
      "epoch": 0.5985673869860952,
      "grad_norm": 2.9549405574798584,
      "learning_rate": 7.349084421859533e-05,
      "loss": 0.7930302619934082,
      "step": 1243
    },
    {
      "epoch": 0.5990489375790043,
      "grad_norm": 1.628475308418274,
      "learning_rate": 7.334046954660852e-05,
      "loss": 0.4897911846637726,
      "step": 1244
    },
    {
      "epoch": 0.5995304881719136,
      "grad_norm": 2.060633897781372,
      "learning_rate": 7.31901597430269e-05,
      "loss": 0.7500059008598328,
      "step": 1245
    },
    {
      "epoch": 0.6000120387648227,
      "grad_norm": 2.478471040725708,
      "learning_rate": 7.303991517358678e-05,
      "loss": 1.014409065246582,
      "step": 1246
    },
    {
      "epoch": 0.6004935893577319,
      "grad_norm": 1.4871796369552612,
      "learning_rate": 7.288973620386568e-05,
      "loss": 1.267449975013733,
      "step": 1247
    },
    {
      "epoch": 0.6009751399506411,
      "grad_norm": 1.7497665882110596,
      "learning_rate": 7.273962319928151e-05,
      "loss": 0.574657678604126,
      "step": 1248
    },
    {
      "epoch": 0.6014566905435502,
      "grad_norm": 2.9989984035491943,
      "learning_rate": 7.258957652509171e-05,
      "loss": 1.1509230136871338,
      "step": 1249
    },
    {
      "epoch": 0.6019382411364594,
      "grad_norm": 3.3296890258789062,
      "learning_rate": 7.24395965463923e-05,
      "loss": 0.5525704026222229,
      "step": 1250
    },
    {
      "epoch": 0.6024197917293685,
      "grad_norm": 5.2618632316589355,
      "learning_rate": 7.228968362811702e-05,
      "loss": 0.7114471197128296,
      "step": 1251
    },
    {
      "epoch": 0.6029013423222778,
      "grad_norm": 3.188866138458252,
      "learning_rate": 7.21398381350364e-05,
      "loss": 0.9359039664268494,
      "step": 1252
    },
    {
      "epoch": 0.6033828929151869,
      "grad_norm": 1.8079476356506348,
      "learning_rate": 7.199006043175698e-05,
      "loss": 0.8708849549293518,
      "step": 1253
    },
    {
      "epoch": 0.6038644435080961,
      "grad_norm": 4.308001518249512,
      "learning_rate": 7.184035088272028e-05,
      "loss": 0.7232476472854614,
      "step": 1254
    },
    {
      "epoch": 0.6043459941010052,
      "grad_norm": 1.541079044342041,
      "learning_rate": 7.169070985220208e-05,
      "loss": 0.5265953540802002,
      "step": 1255
    },
    {
      "epoch": 0.6048275446939144,
      "grad_norm": 2.096688747406006,
      "learning_rate": 7.154113770431132e-05,
      "loss": 0.7937042117118835,
      "step": 1256
    },
    {
      "epoch": 0.6053090952868235,
      "grad_norm": 3.0756218433380127,
      "learning_rate": 7.13916348029894e-05,
      "loss": 0.5530334115028381,
      "step": 1257
    },
    {
      "epoch": 0.6057906458797327,
      "grad_norm": 0.9245099425315857,
      "learning_rate": 7.124220151200926e-05,
      "loss": 0.5336117148399353,
      "step": 1258
    },
    {
      "epoch": 0.6062721964726419,
      "grad_norm": 2.570974826812744,
      "learning_rate": 7.10928381949744e-05,
      "loss": 0.772148609161377,
      "step": 1259
    },
    {
      "epoch": 0.6067537470655511,
      "grad_norm": 1.8352408409118652,
      "learning_rate": 7.094354521531807e-05,
      "loss": 0.6643396615982056,
      "step": 1260
    },
    {
      "epoch": 0.6072352976584603,
      "grad_norm": 6.521425724029541,
      "learning_rate": 7.079432293630244e-05,
      "loss": 1.0640923976898193,
      "step": 1261
    },
    {
      "epoch": 0.6077168482513694,
      "grad_norm": 3.8641836643218994,
      "learning_rate": 7.064517172101753e-05,
      "loss": 1.0460093021392822,
      "step": 1262
    },
    {
      "epoch": 0.6081983988442786,
      "grad_norm": 1.6016931533813477,
      "learning_rate": 7.04960919323806e-05,
      "loss": 0.24079592525959015,
      "step": 1263
    },
    {
      "epoch": 0.6086799494371877,
      "grad_norm": 1.9926220178604126,
      "learning_rate": 7.034708393313493e-05,
      "loss": 1.0020270347595215,
      "step": 1264
    },
    {
      "epoch": 0.609161500030097,
      "grad_norm": 1.8591305017471313,
      "learning_rate": 7.019814808584928e-05,
      "loss": 0.6165023446083069,
      "step": 1265
    },
    {
      "epoch": 0.609643050623006,
      "grad_norm": 1.7767674922943115,
      "learning_rate": 7.004928475291678e-05,
      "loss": 0.49050724506378174,
      "step": 1266
    },
    {
      "epoch": 0.6101246012159153,
      "grad_norm": 1.8133375644683838,
      "learning_rate": 6.990049429655412e-05,
      "loss": 0.9639424085617065,
      "step": 1267
    },
    {
      "epoch": 0.6106061518088244,
      "grad_norm": 1.6175096035003662,
      "learning_rate": 6.97517770788007e-05,
      "loss": 0.8122717142105103,
      "step": 1268
    },
    {
      "epoch": 0.6110877024017336,
      "grad_norm": 2.0364584922790527,
      "learning_rate": 6.960313346151761e-05,
      "loss": 1.1467472314834595,
      "step": 1269
    },
    {
      "epoch": 0.6115692529946427,
      "grad_norm": 2.000424861907959,
      "learning_rate": 6.9454563806387e-05,
      "loss": 1.1381384134292603,
      "step": 1270
    },
    {
      "epoch": 0.6120508035875519,
      "grad_norm": 1.4513589143753052,
      "learning_rate": 6.930606847491094e-05,
      "loss": 0.535015881061554,
      "step": 1271
    },
    {
      "epoch": 0.612532354180461,
      "grad_norm": 2.351076602935791,
      "learning_rate": 6.915764782841072e-05,
      "loss": 0.8914245367050171,
      "step": 1272
    },
    {
      "epoch": 0.6130139047733703,
      "grad_norm": 2.7353837490081787,
      "learning_rate": 6.900930222802588e-05,
      "loss": 0.9201171398162842,
      "step": 1273
    },
    {
      "epoch": 0.6134954553662795,
      "grad_norm": 3.2564220428466797,
      "learning_rate": 6.886103203471337e-05,
      "loss": 0.9169067740440369,
      "step": 1274
    },
    {
      "epoch": 0.6139770059591886,
      "grad_norm": 1.768539309501648,
      "learning_rate": 6.871283760924665e-05,
      "loss": 0.5857947468757629,
      "step": 1275
    },
    {
      "epoch": 0.6144585565520978,
      "grad_norm": 1.419161319732666,
      "learning_rate": 6.856471931221478e-05,
      "loss": 0.8163211941719055,
      "step": 1276
    },
    {
      "epoch": 0.6149401071450069,
      "grad_norm": 1.2443370819091797,
      "learning_rate": 6.841667750402162e-05,
      "loss": 0.5108415484428406,
      "step": 1277
    },
    {
      "epoch": 0.6154216577379161,
      "grad_norm": 1.4883040189743042,
      "learning_rate": 6.826871254488496e-05,
      "loss": 1.2236932516098022,
      "step": 1278
    },
    {
      "epoch": 0.6159032083308252,
      "grad_norm": 2.4167792797088623,
      "learning_rate": 6.812082479483553e-05,
      "loss": 0.8932603001594543,
      "step": 1279
    },
    {
      "epoch": 0.6163847589237345,
      "grad_norm": 2.2708561420440674,
      "learning_rate": 6.797301461371625e-05,
      "loss": 0.7447341084480286,
      "step": 1280
    },
    {
      "epoch": 0.6168663095166436,
      "grad_norm": 2.546278715133667,
      "learning_rate": 6.782528236118124e-05,
      "loss": 0.7705029249191284,
      "step": 1281
    },
    {
      "epoch": 0.6173478601095528,
      "grad_norm": 2.1538925170898438,
      "learning_rate": 6.767762839669503e-05,
      "loss": 0.6113386154174805,
      "step": 1282
    },
    {
      "epoch": 0.6178294107024619,
      "grad_norm": 2.145097494125366,
      "learning_rate": 6.753005307953167e-05,
      "loss": 0.5513904690742493,
      "step": 1283
    },
    {
      "epoch": 0.6183109612953711,
      "grad_norm": 2.934802293777466,
      "learning_rate": 6.738255676877381e-05,
      "loss": 0.6201481819152832,
      "step": 1284
    },
    {
      "epoch": 0.6187925118882802,
      "grad_norm": 1.853514552116394,
      "learning_rate": 6.723513982331195e-05,
      "loss": 0.6032390594482422,
      "step": 1285
    },
    {
      "epoch": 0.6192740624811894,
      "grad_norm": 1.4172900915145874,
      "learning_rate": 6.708780260184333e-05,
      "loss": 0.4911664128303528,
      "step": 1286
    },
    {
      "epoch": 0.6197556130740985,
      "grad_norm": 2.883531332015991,
      "learning_rate": 6.694054546287132e-05,
      "loss": 0.5742973685264587,
      "step": 1287
    },
    {
      "epoch": 0.6202371636670078,
      "grad_norm": 1.188624620437622,
      "learning_rate": 6.679336876470441e-05,
      "loss": 0.6552760601043701,
      "step": 1288
    },
    {
      "epoch": 0.620718714259917,
      "grad_norm": 3.3380281925201416,
      "learning_rate": 6.664627286545535e-05,
      "loss": 1.04770028591156,
      "step": 1289
    },
    {
      "epoch": 0.6212002648528261,
      "grad_norm": 1.6602662801742554,
      "learning_rate": 6.649925812304025e-05,
      "loss": 0.587570309638977,
      "step": 1290
    },
    {
      "epoch": 0.6216818154457353,
      "grad_norm": 1.2718095779418945,
      "learning_rate": 6.635232489517782e-05,
      "loss": 1.1010924577713013,
      "step": 1291
    },
    {
      "epoch": 0.6221633660386444,
      "grad_norm": 3.140336513519287,
      "learning_rate": 6.620547353938836e-05,
      "loss": 0.7243585586547852,
      "step": 1292
    },
    {
      "epoch": 0.6226449166315536,
      "grad_norm": 3.731562852859497,
      "learning_rate": 6.605870441299302e-05,
      "loss": 1.3233602046966553,
      "step": 1293
    },
    {
      "epoch": 0.6231264672244627,
      "grad_norm": 4.593099117279053,
      "learning_rate": 6.591201787311285e-05,
      "loss": 0.9525260925292969,
      "step": 1294
    },
    {
      "epoch": 0.623608017817372,
      "grad_norm": 1.4077935218811035,
      "learning_rate": 6.57654142766679e-05,
      "loss": 0.8255711793899536,
      "step": 1295
    },
    {
      "epoch": 0.6240895684102811,
      "grad_norm": 2.333665132522583,
      "learning_rate": 6.561889398037643e-05,
      "loss": 0.7672783732414246,
      "step": 1296
    },
    {
      "epoch": 0.6245711190031903,
      "grad_norm": 3.2461068630218506,
      "learning_rate": 6.547245734075403e-05,
      "loss": 0.7859494686126709,
      "step": 1297
    },
    {
      "epoch": 0.6250526695960994,
      "grad_norm": 2.298067569732666,
      "learning_rate": 6.532610471411274e-05,
      "loss": 0.6992470026016235,
      "step": 1298
    },
    {
      "epoch": 0.6255342201890086,
      "grad_norm": 2.7416765689849854,
      "learning_rate": 6.517983645656014e-05,
      "loss": 0.9306595921516418,
      "step": 1299
    },
    {
      "epoch": 0.6260157707819177,
      "grad_norm": 0.9679139852523804,
      "learning_rate": 6.503365292399857e-05,
      "loss": 0.8975527882575989,
      "step": 1300
    },
    {
      "epoch": 0.6264973213748269,
      "grad_norm": 1.561290979385376,
      "learning_rate": 6.488755447212418e-05,
      "loss": 0.48123905062675476,
      "step": 1301
    },
    {
      "epoch": 0.6269788719677362,
      "grad_norm": 3.4579672813415527,
      "learning_rate": 6.474154145642612e-05,
      "loss": 0.8294971585273743,
      "step": 1302
    },
    {
      "epoch": 0.6274604225606453,
      "grad_norm": 1.3913123607635498,
      "learning_rate": 6.459561423218561e-05,
      "loss": 0.6386594176292419,
      "step": 1303
    },
    {
      "epoch": 0.6279419731535545,
      "grad_norm": 3.5648066997528076,
      "learning_rate": 6.444977315447521e-05,
      "loss": 1.0743132829666138,
      "step": 1304
    },
    {
      "epoch": 0.6284235237464636,
      "grad_norm": 1.9397964477539062,
      "learning_rate": 6.430401857815776e-05,
      "loss": 0.7587765455245972,
      "step": 1305
    },
    {
      "epoch": 0.6289050743393728,
      "grad_norm": 1.7409212589263916,
      "learning_rate": 6.415835085788575e-05,
      "loss": 0.8555688261985779,
      "step": 1306
    },
    {
      "epoch": 0.6293866249322819,
      "grad_norm": 1.6474158763885498,
      "learning_rate": 6.401277034810017e-05,
      "loss": 0.2949734628200531,
      "step": 1307
    },
    {
      "epoch": 0.6298681755251911,
      "grad_norm": 1.5710214376449585,
      "learning_rate": 6.386727740302994e-05,
      "loss": 0.5082312822341919,
      "step": 1308
    },
    {
      "epoch": 0.6303497261181003,
      "grad_norm": 3.9894864559173584,
      "learning_rate": 6.37218723766909e-05,
      "loss": 0.506136953830719,
      "step": 1309
    },
    {
      "epoch": 0.6308312767110095,
      "grad_norm": 1.8334898948669434,
      "learning_rate": 6.357655562288488e-05,
      "loss": 0.8781048655509949,
      "step": 1310
    },
    {
      "epoch": 0.6313128273039186,
      "grad_norm": 1.5686639547348022,
      "learning_rate": 6.343132749519902e-05,
      "loss": 0.8222221732139587,
      "step": 1311
    },
    {
      "epoch": 0.6317943778968278,
      "grad_norm": 1.482987642288208,
      "learning_rate": 6.328618834700474e-05,
      "loss": 0.7976973652839661,
      "step": 1312
    },
    {
      "epoch": 0.6322759284897369,
      "grad_norm": 2.0139546394348145,
      "learning_rate": 6.314113853145703e-05,
      "loss": 0.6463433504104614,
      "step": 1313
    },
    {
      "epoch": 0.6327574790826461,
      "grad_norm": 2.2550463676452637,
      "learning_rate": 6.299617840149349e-05,
      "loss": 0.7010633945465088,
      "step": 1314
    },
    {
      "epoch": 0.6332390296755553,
      "grad_norm": 2.5890138149261475,
      "learning_rate": 6.285130830983339e-05,
      "loss": 1.2857202291488647,
      "step": 1315
    },
    {
      "epoch": 0.6337205802684645,
      "grad_norm": 1.242794394493103,
      "learning_rate": 6.270652860897704e-05,
      "loss": 0.703166663646698,
      "step": 1316
    },
    {
      "epoch": 0.6342021308613737,
      "grad_norm": 1.908077597618103,
      "learning_rate": 6.25618396512048e-05,
      "loss": 0.5847074389457703,
      "step": 1317
    },
    {
      "epoch": 0.6346836814542828,
      "grad_norm": 2.4328458309173584,
      "learning_rate": 6.24172417885762e-05,
      "loss": 0.5482038855552673,
      "step": 1318
    },
    {
      "epoch": 0.635165232047192,
      "grad_norm": 2.597142219543457,
      "learning_rate": 6.227273537292911e-05,
      "loss": 0.9418822526931763,
      "step": 1319
    },
    {
      "epoch": 0.6356467826401011,
      "grad_norm": 2.848977565765381,
      "learning_rate": 6.212832075587891e-05,
      "loss": 0.7264025211334229,
      "step": 1320
    },
    {
      "epoch": 0.6361283332330103,
      "grad_norm": 3.3098719120025635,
      "learning_rate": 6.19839982888176e-05,
      "loss": 0.7752615213394165,
      "step": 1321
    },
    {
      "epoch": 0.6366098838259194,
      "grad_norm": 3.0315816402435303,
      "learning_rate": 6.183976832291296e-05,
      "loss": 0.9584868550300598,
      "step": 1322
    },
    {
      "epoch": 0.6370914344188287,
      "grad_norm": 1.9445630311965942,
      "learning_rate": 6.169563120910775e-05,
      "loss": 1.2950618267059326,
      "step": 1323
    },
    {
      "epoch": 0.6375729850117378,
      "grad_norm": 2.6105964183807373,
      "learning_rate": 6.155158729811867e-05,
      "loss": 0.9339355230331421,
      "step": 1324
    },
    {
      "epoch": 0.638054535604647,
      "grad_norm": 7.7421698570251465,
      "learning_rate": 6.140763694043578e-05,
      "loss": 1.1200129985809326,
      "step": 1325
    },
    {
      "epoch": 0.6385360861975561,
      "grad_norm": 2.582887887954712,
      "learning_rate": 6.126378048632139e-05,
      "loss": 0.865228533744812,
      "step": 1326
    },
    {
      "epoch": 0.6390176367904653,
      "grad_norm": 2.1188411712646484,
      "learning_rate": 6.112001828580944e-05,
      "loss": 0.7778995633125305,
      "step": 1327
    },
    {
      "epoch": 0.6394991873833744,
      "grad_norm": 1.919268012046814,
      "learning_rate": 6.0976350688704455e-05,
      "loss": 0.46954837441444397,
      "step": 1328
    },
    {
      "epoch": 0.6399807379762836,
      "grad_norm": 1.1841535568237305,
      "learning_rate": 6.083277804458072e-05,
      "loss": 0.6684697270393372,
      "step": 1329
    },
    {
      "epoch": 0.6404622885691929,
      "grad_norm": 1.8455570936203003,
      "learning_rate": 6.068930070278159e-05,
      "loss": 0.38171032071113586,
      "step": 1330
    },
    {
      "epoch": 0.640943839162102,
      "grad_norm": 1.8052548170089722,
      "learning_rate": 6.054591901241846e-05,
      "loss": 0.5547836422920227,
      "step": 1331
    },
    {
      "epoch": 0.6414253897550112,
      "grad_norm": 2.119178056716919,
      "learning_rate": 6.040263332237002e-05,
      "loss": 0.7391781210899353,
      "step": 1332
    },
    {
      "epoch": 0.6419069403479203,
      "grad_norm": 3.0023255348205566,
      "learning_rate": 6.025944398128137e-05,
      "loss": 1.1047343015670776,
      "step": 1333
    },
    {
      "epoch": 0.6423884909408295,
      "grad_norm": 1.737441062927246,
      "learning_rate": 6.011635133756309e-05,
      "loss": 0.5697471499443054,
      "step": 1334
    },
    {
      "epoch": 0.6428700415337386,
      "grad_norm": 3.9862446784973145,
      "learning_rate": 5.99733557393906e-05,
      "loss": 1.3332350254058838,
      "step": 1335
    },
    {
      "epoch": 0.6433515921266478,
      "grad_norm": 1.3148013353347778,
      "learning_rate": 5.983045753470308e-05,
      "loss": 1.1608415842056274,
      "step": 1336
    },
    {
      "epoch": 0.6438331427195569,
      "grad_norm": 2.815699577331543,
      "learning_rate": 5.96876570712028e-05,
      "loss": 1.2337565422058105,
      "step": 1337
    },
    {
      "epoch": 0.6443146933124662,
      "grad_norm": 2.5290703773498535,
      "learning_rate": 5.954495469635417e-05,
      "loss": 0.9341751933097839,
      "step": 1338
    },
    {
      "epoch": 0.6447962439053753,
      "grad_norm": 2.237555742263794,
      "learning_rate": 5.940235075738296e-05,
      "loss": 0.8382311463356018,
      "step": 1339
    },
    {
      "epoch": 0.6452777944982845,
      "grad_norm": 1.6974987983703613,
      "learning_rate": 5.925984560127542e-05,
      "loss": 0.7167918086051941,
      "step": 1340
    },
    {
      "epoch": 0.6457593450911936,
      "grad_norm": 4.838293075561523,
      "learning_rate": 5.911743957477739e-05,
      "loss": 0.534832239151001,
      "step": 1341
    },
    {
      "epoch": 0.6462408956841028,
      "grad_norm": 2.0869569778442383,
      "learning_rate": 5.897513302439355e-05,
      "loss": 0.7934316396713257,
      "step": 1342
    },
    {
      "epoch": 0.646722446277012,
      "grad_norm": 1.3564095497131348,
      "learning_rate": 5.883292629638651e-05,
      "loss": 1.0168604850769043,
      "step": 1343
    },
    {
      "epoch": 0.6472039968699211,
      "grad_norm": 1.5895377397537231,
      "learning_rate": 5.869081973677604e-05,
      "loss": 0.9991562366485596,
      "step": 1344
    },
    {
      "epoch": 0.6476855474628304,
      "grad_norm": 2.8290328979492188,
      "learning_rate": 5.8548813691338134e-05,
      "loss": 0.7760152220726013,
      "step": 1345
    },
    {
      "epoch": 0.6481670980557395,
      "grad_norm": 1.9063435792922974,
      "learning_rate": 5.84069085056042e-05,
      "loss": 0.8138294816017151,
      "step": 1346
    },
    {
      "epoch": 0.6486486486486487,
      "grad_norm": 2.9390857219696045,
      "learning_rate": 5.826510452486027e-05,
      "loss": 0.8707611560821533,
      "step": 1347
    },
    {
      "epoch": 0.6491301992415578,
      "grad_norm": 3.932231903076172,
      "learning_rate": 5.81234020941461e-05,
      "loss": 0.7644728422164917,
      "step": 1348
    },
    {
      "epoch": 0.649611749834467,
      "grad_norm": 1.248726487159729,
      "learning_rate": 5.798180155825437e-05,
      "loss": 0.6264711022377014,
      "step": 1349
    },
    {
      "epoch": 0.6500933004273761,
      "grad_norm": 1.2341344356536865,
      "learning_rate": 5.784030326172981e-05,
      "loss": 0.8610283136367798,
      "step": 1350
    },
    {
      "epoch": 0.6505748510202853,
      "grad_norm": 2.3828272819519043,
      "learning_rate": 5.7698907548868395e-05,
      "loss": 0.6295274496078491,
      "step": 1351
    },
    {
      "epoch": 0.6510564016131944,
      "grad_norm": 2.036634922027588,
      "learning_rate": 5.755761476371653e-05,
      "loss": 0.5899828672409058,
      "step": 1352
    },
    {
      "epoch": 0.6515379522061037,
      "grad_norm": 1.150680422782898,
      "learning_rate": 5.741642525007003e-05,
      "loss": 0.8901524543762207,
      "step": 1353
    },
    {
      "epoch": 0.6520195027990128,
      "grad_norm": 3.1640279293060303,
      "learning_rate": 5.727533935147359e-05,
      "loss": 1.1335471868515015,
      "step": 1354
    },
    {
      "epoch": 0.652501053391922,
      "grad_norm": 5.0667572021484375,
      "learning_rate": 5.713435741121975e-05,
      "loss": 0.49279093742370605,
      "step": 1355
    },
    {
      "epoch": 0.6529826039848312,
      "grad_norm": 1.2700073719024658,
      "learning_rate": 5.699347977234799e-05,
      "loss": 0.5746053457260132,
      "step": 1356
    },
    {
      "epoch": 0.6534641545777403,
      "grad_norm": 1.4186991453170776,
      "learning_rate": 5.685270677764412e-05,
      "loss": 0.5925418138504028,
      "step": 1357
    },
    {
      "epoch": 0.6539457051706495,
      "grad_norm": 1.1606782674789429,
      "learning_rate": 5.671203876963931e-05,
      "loss": 0.6767562031745911,
      "step": 1358
    },
    {
      "epoch": 0.6544272557635586,
      "grad_norm": 2.6344268321990967,
      "learning_rate": 5.657147609060924e-05,
      "loss": 0.539024829864502,
      "step": 1359
    },
    {
      "epoch": 0.6549088063564679,
      "grad_norm": 2.0170819759368896,
      "learning_rate": 5.643101908257333e-05,
      "loss": 0.612555980682373,
      "step": 1360
    },
    {
      "epoch": 0.655390356949377,
      "grad_norm": 1.9014548063278198,
      "learning_rate": 5.629066808729385e-05,
      "loss": 0.5272603631019592,
      "step": 1361
    },
    {
      "epoch": 0.6558719075422862,
      "grad_norm": 2.820272445678711,
      "learning_rate": 5.6150423446275144e-05,
      "loss": 0.9678093194961548,
      "step": 1362
    },
    {
      "epoch": 0.6563534581351953,
      "grad_norm": 3.59275484085083,
      "learning_rate": 5.601028550076277e-05,
      "loss": 0.5986140370368958,
      "step": 1363
    },
    {
      "epoch": 0.6568350087281045,
      "grad_norm": 2.3459818363189697,
      "learning_rate": 5.587025459174271e-05,
      "loss": 0.7716139554977417,
      "step": 1364
    },
    {
      "epoch": 0.6573165593210136,
      "grad_norm": 3.167153835296631,
      "learning_rate": 5.573033105994038e-05,
      "loss": 0.5715650320053101,
      "step": 1365
    },
    {
      "epoch": 0.6577981099139228,
      "grad_norm": 1.8883392810821533,
      "learning_rate": 5.559051524582002e-05,
      "loss": 0.973146915435791,
      "step": 1366
    },
    {
      "epoch": 0.658279660506832,
      "grad_norm": 2.793569564819336,
      "learning_rate": 5.5450807489583777e-05,
      "loss": 0.6978902220726013,
      "step": 1367
    },
    {
      "epoch": 0.6587612110997412,
      "grad_norm": 4.652721881866455,
      "learning_rate": 5.531120813117085e-05,
      "loss": 0.9357098937034607,
      "step": 1368
    },
    {
      "epoch": 0.6592427616926503,
      "grad_norm": 2.6184303760528564,
      "learning_rate": 5.517171751025667e-05,
      "loss": 0.4999481439590454,
      "step": 1369
    },
    {
      "epoch": 0.6597243122855595,
      "grad_norm": 1.9336432218551636,
      "learning_rate": 5.5032335966252103e-05,
      "loss": 0.5195287466049194,
      "step": 1370
    },
    {
      "epoch": 0.6602058628784687,
      "grad_norm": 3.689028263092041,
      "learning_rate": 5.489306383830258e-05,
      "loss": 0.6519877910614014,
      "step": 1371
    },
    {
      "epoch": 0.6606874134713778,
      "grad_norm": 1.4407304525375366,
      "learning_rate": 5.475390146528738e-05,
      "loss": 0.8569602370262146,
      "step": 1372
    },
    {
      "epoch": 0.661168964064287,
      "grad_norm": 1.3111518621444702,
      "learning_rate": 5.461484918581858e-05,
      "loss": 0.6398709416389465,
      "step": 1373
    },
    {
      "epoch": 0.6616505146571962,
      "grad_norm": 2.1044137477874756,
      "learning_rate": 5.4475907338240494e-05,
      "loss": 0.775865912437439,
      "step": 1374
    },
    {
      "epoch": 0.6621320652501054,
      "grad_norm": 2.2225348949432373,
      "learning_rate": 5.43370762606287e-05,
      "loss": 0.7651700973510742,
      "step": 1375
    },
    {
      "epoch": 0.6626136158430145,
      "grad_norm": 1.5968869924545288,
      "learning_rate": 5.4198356290789276e-05,
      "loss": 1.028122067451477,
      "step": 1376
    },
    {
      "epoch": 0.6630951664359237,
      "grad_norm": 1.3744912147521973,
      "learning_rate": 5.405974776625785e-05,
      "loss": 0.34380024671554565,
      "step": 1377
    },
    {
      "epoch": 0.6635767170288328,
      "grad_norm": 2.1190967559814453,
      "learning_rate": 5.392125102429899e-05,
      "loss": 0.6510315537452698,
      "step": 1378
    },
    {
      "epoch": 0.664058267621742,
      "grad_norm": 6.662756443023682,
      "learning_rate": 5.378286640190522e-05,
      "loss": 0.9724109768867493,
      "step": 1379
    },
    {
      "epoch": 0.6645398182146511,
      "grad_norm": 1.7120323181152344,
      "learning_rate": 5.364459423579629e-05,
      "loss": 0.8208657503128052,
      "step": 1380
    },
    {
      "epoch": 0.6650213688075604,
      "grad_norm": 4.565060615539551,
      "learning_rate": 5.350643486241825e-05,
      "loss": 0.7308831214904785,
      "step": 1381
    },
    {
      "epoch": 0.6655029194004695,
      "grad_norm": 2.599998950958252,
      "learning_rate": 5.33683886179428e-05,
      "loss": 0.7606978416442871,
      "step": 1382
    },
    {
      "epoch": 0.6659844699933787,
      "grad_norm": 2.6620562076568604,
      "learning_rate": 5.3230455838266266e-05,
      "loss": 0.9607159495353699,
      "step": 1383
    },
    {
      "epoch": 0.6664660205862879,
      "grad_norm": 3.301551342010498,
      "learning_rate": 5.309263685900898e-05,
      "loss": 0.7540823221206665,
      "step": 1384
    },
    {
      "epoch": 0.666947571179197,
      "grad_norm": 2.7178406715393066,
      "learning_rate": 5.295493201551433e-05,
      "loss": 0.7821422815322876,
      "step": 1385
    },
    {
      "epoch": 0.6674291217721062,
      "grad_norm": 2.4588863849639893,
      "learning_rate": 5.281734164284802e-05,
      "loss": 0.6028894186019897,
      "step": 1386
    },
    {
      "epoch": 0.6679106723650153,
      "grad_norm": 1.4585126638412476,
      "learning_rate": 5.26798660757971e-05,
      "loss": 1.2175313234329224,
      "step": 1387
    },
    {
      "epoch": 0.6683922229579246,
      "grad_norm": 1.173219084739685,
      "learning_rate": 5.2542505648869434e-05,
      "loss": 0.29551905393600464,
      "step": 1388
    },
    {
      "epoch": 0.6688737735508337,
      "grad_norm": 3.0974905490875244,
      "learning_rate": 5.240526069629265e-05,
      "loss": 0.9472237229347229,
      "step": 1389
    },
    {
      "epoch": 0.6693553241437429,
      "grad_norm": 1.552543044090271,
      "learning_rate": 5.22681315520134e-05,
      "loss": 0.8448207974433899,
      "step": 1390
    },
    {
      "epoch": 0.669836874736652,
      "grad_norm": 3.851508378982544,
      "learning_rate": 5.213111854969661e-05,
      "loss": 0.9650540947914124,
      "step": 1391
    },
    {
      "epoch": 0.6703184253295612,
      "grad_norm": 2.3694324493408203,
      "learning_rate": 5.199422202272448e-05,
      "loss": 0.3335321545600891,
      "step": 1392
    },
    {
      "epoch": 0.6707999759224703,
      "grad_norm": 2.6251461505889893,
      "learning_rate": 5.185744230419589e-05,
      "loss": 0.8398869037628174,
      "step": 1393
    },
    {
      "epoch": 0.6712815265153795,
      "grad_norm": 1.785504698753357,
      "learning_rate": 5.172077972692553e-05,
      "loss": 0.5403978824615479,
      "step": 1394
    },
    {
      "epoch": 0.6717630771082886,
      "grad_norm": 1.9077867269515991,
      "learning_rate": 5.1584234623442974e-05,
      "loss": 0.7148364186286926,
      "step": 1395
    },
    {
      "epoch": 0.6722446277011979,
      "grad_norm": 3.0032958984375,
      "learning_rate": 5.1447807325992025e-05,
      "loss": 0.7190037369728088,
      "step": 1396
    },
    {
      "epoch": 0.6727261782941071,
      "grad_norm": 1.241286277770996,
      "learning_rate": 5.13114981665298e-05,
      "loss": 0.9692134261131287,
      "step": 1397
    },
    {
      "epoch": 0.6732077288870162,
      "grad_norm": 3.4759392738342285,
      "learning_rate": 5.117530747672603e-05,
      "loss": 0.7307540774345398,
      "step": 1398
    },
    {
      "epoch": 0.6736892794799254,
      "grad_norm": 1.7262109518051147,
      "learning_rate": 5.103923558796203e-05,
      "loss": 0.7541236281394958,
      "step": 1399
    },
    {
      "epoch": 0.6741708300728345,
      "grad_norm": 2.2756550312042236,
      "learning_rate": 5.090328283133019e-05,
      "loss": 1.4233908653259277,
      "step": 1400
    },
    {
      "epoch": 0.6746523806657437,
      "grad_norm": 3.1226863861083984,
      "learning_rate": 5.0767449537632986e-05,
      "loss": 0.5946443676948547,
      "step": 1401
    },
    {
      "epoch": 0.6751339312586528,
      "grad_norm": 2.213266134262085,
      "learning_rate": 5.06317360373822e-05,
      "loss": 0.5166051387786865,
      "step": 1402
    },
    {
      "epoch": 0.6756154818515621,
      "grad_norm": 1.666943907737732,
      "learning_rate": 5.049614266079813e-05,
      "loss": 0.34427306056022644,
      "step": 1403
    },
    {
      "epoch": 0.6760970324444712,
      "grad_norm": 4.887892723083496,
      "learning_rate": 5.036066973780882e-05,
      "loss": 0.6936655640602112,
      "step": 1404
    },
    {
      "epoch": 0.6765785830373804,
      "grad_norm": 1.3032301664352417,
      "learning_rate": 5.022531759804918e-05,
      "loss": 1.0333144664764404,
      "step": 1405
    },
    {
      "epoch": 0.6770601336302895,
      "grad_norm": 0.9449478983879089,
      "learning_rate": 5.009008657086025e-05,
      "loss": 0.3657134473323822,
      "step": 1406
    },
    {
      "epoch": 0.6775416842231987,
      "grad_norm": 3.762364149093628,
      "learning_rate": 4.9954976985288395e-05,
      "loss": 0.8035635352134705,
      "step": 1407
    },
    {
      "epoch": 0.6780232348161078,
      "grad_norm": 2.9945201873779297,
      "learning_rate": 4.981998917008448e-05,
      "loss": 0.6599687933921814,
      "step": 1408
    },
    {
      "epoch": 0.678504785409017,
      "grad_norm": 1.929681658744812,
      "learning_rate": 4.9685123453703e-05,
      "loss": 0.9616090059280396,
      "step": 1409
    },
    {
      "epoch": 0.6789863360019261,
      "grad_norm": 2.38234281539917,
      "learning_rate": 4.955038016430149e-05,
      "loss": 0.7233282923698425,
      "step": 1410
    },
    {
      "epoch": 0.6794678865948354,
      "grad_norm": 2.5650625228881836,
      "learning_rate": 4.9415759629739455e-05,
      "loss": 0.5705734491348267,
      "step": 1411
    },
    {
      "epoch": 0.6799494371877446,
      "grad_norm": 1.6509586572647095,
      "learning_rate": 4.928126217757782e-05,
      "loss": 0.8393734693527222,
      "step": 1412
    },
    {
      "epoch": 0.6804309877806537,
      "grad_norm": 1.938231348991394,
      "learning_rate": 4.914688813507797e-05,
      "loss": 0.7624192237854004,
      "step": 1413
    },
    {
      "epoch": 0.6809125383735629,
      "grad_norm": 2.5296216011047363,
      "learning_rate": 4.901263782920105e-05,
      "loss": 0.8140773773193359,
      "step": 1414
    },
    {
      "epoch": 0.681394088966472,
      "grad_norm": 3.5682215690612793,
      "learning_rate": 4.887851158660706e-05,
      "loss": 0.9878526926040649,
      "step": 1415
    },
    {
      "epoch": 0.6818756395593812,
      "grad_norm": 2.109893798828125,
      "learning_rate": 4.8744509733654184e-05,
      "loss": 0.5913165211677551,
      "step": 1416
    },
    {
      "epoch": 0.6823571901522903,
      "grad_norm": 1.3570090532302856,
      "learning_rate": 4.861063259639793e-05,
      "loss": 0.7238122224807739,
      "step": 1417
    },
    {
      "epoch": 0.6828387407451996,
      "grad_norm": 2.3462166786193848,
      "learning_rate": 4.847688050059033e-05,
      "loss": 0.7800824642181396,
      "step": 1418
    },
    {
      "epoch": 0.6833202913381087,
      "grad_norm": 2.9588794708251953,
      "learning_rate": 4.8343253771679155e-05,
      "loss": 0.5499364137649536,
      "step": 1419
    },
    {
      "epoch": 0.6838018419310179,
      "grad_norm": 1.4072751998901367,
      "learning_rate": 4.82097527348072e-05,
      "loss": 1.0598034858703613,
      "step": 1420
    },
    {
      "epoch": 0.684283392523927,
      "grad_norm": 4.153375625610352,
      "learning_rate": 4.8076377714811284e-05,
      "loss": 0.8945078253746033,
      "step": 1421
    },
    {
      "epoch": 0.6847649431168362,
      "grad_norm": 1.7934489250183105,
      "learning_rate": 4.7943129036221735e-05,
      "loss": 0.6439244151115417,
      "step": 1422
    },
    {
      "epoch": 0.6852464937097453,
      "grad_norm": 1.9287291765213013,
      "learning_rate": 4.781000702326142e-05,
      "loss": 0.8950934410095215,
      "step": 1423
    },
    {
      "epoch": 0.6857280443026545,
      "grad_norm": 1.059610366821289,
      "learning_rate": 4.767701199984497e-05,
      "loss": 0.5218794941902161,
      "step": 1424
    },
    {
      "epoch": 0.6862095948955638,
      "grad_norm": 3.465240240097046,
      "learning_rate": 4.7544144289578066e-05,
      "loss": 0.9760928750038147,
      "step": 1425
    },
    {
      "epoch": 0.6866911454884729,
      "grad_norm": 2.3100907802581787,
      "learning_rate": 4.7411404215756594e-05,
      "loss": 0.521527111530304,
      "step": 1426
    },
    {
      "epoch": 0.6871726960813821,
      "grad_norm": 1.141473650932312,
      "learning_rate": 4.7278792101365866e-05,
      "loss": 0.6822175979614258,
      "step": 1427
    },
    {
      "epoch": 0.6876542466742912,
      "grad_norm": 1.551011562347412,
      "learning_rate": 4.714630826907985e-05,
      "loss": 0.6446740031242371,
      "step": 1428
    },
    {
      "epoch": 0.6881357972672004,
      "grad_norm": 1.2976975440979004,
      "learning_rate": 4.701395304126038e-05,
      "loss": 0.8829476833343506,
      "step": 1429
    },
    {
      "epoch": 0.6886173478601095,
      "grad_norm": 1.626734972000122,
      "learning_rate": 4.6881726739956375e-05,
      "loss": 0.6615743637084961,
      "step": 1430
    },
    {
      "epoch": 0.6890988984530187,
      "grad_norm": 2.3418686389923096,
      "learning_rate": 4.6749629686902984e-05,
      "loss": 0.86656254529953,
      "step": 1431
    },
    {
      "epoch": 0.6895804490459279,
      "grad_norm": 2.5859110355377197,
      "learning_rate": 4.661766220352097e-05,
      "loss": 1.26038658618927,
      "step": 1432
    },
    {
      "epoch": 0.6900619996388371,
      "grad_norm": 1.7766305208206177,
      "learning_rate": 4.64858246109157e-05,
      "loss": 0.856540322303772,
      "step": 1433
    },
    {
      "epoch": 0.6905435502317462,
      "grad_norm": 2.465031147003174,
      "learning_rate": 4.63541172298766e-05,
      "loss": 1.0491377115249634,
      "step": 1434
    },
    {
      "epoch": 0.6910251008246554,
      "grad_norm": 2.6391608715057373,
      "learning_rate": 4.622254038087622e-05,
      "loss": 1.118534803390503,
      "step": 1435
    },
    {
      "epoch": 0.6915066514175645,
      "grad_norm": 1.0951498746871948,
      "learning_rate": 4.60910943840695e-05,
      "loss": 0.29133671522140503,
      "step": 1436
    },
    {
      "epoch": 0.6919882020104737,
      "grad_norm": 2.76126766204834,
      "learning_rate": 4.5959779559292985e-05,
      "loss": 0.6496342420578003,
      "step": 1437
    },
    {
      "epoch": 0.6924697526033828,
      "grad_norm": 1.727417230606079,
      "learning_rate": 4.582859622606406e-05,
      "loss": 0.783515214920044,
      "step": 1438
    },
    {
      "epoch": 0.6929513031962921,
      "grad_norm": 1.9444777965545654,
      "learning_rate": 4.569754470358014e-05,
      "loss": 0.9542943835258484,
      "step": 1439
    },
    {
      "epoch": 0.6934328537892013,
      "grad_norm": 1.9966187477111816,
      "learning_rate": 4.556662531071796e-05,
      "loss": 0.736075222492218,
      "step": 1440
    },
    {
      "epoch": 0.6939144043821104,
      "grad_norm": 2.242565155029297,
      "learning_rate": 4.54358383660327e-05,
      "loss": 0.41106507182121277,
      "step": 1441
    },
    {
      "epoch": 0.6943959549750196,
      "grad_norm": 1.9280340671539307,
      "learning_rate": 4.530518418775733e-05,
      "loss": 0.9464333057403564,
      "step": 1442
    },
    {
      "epoch": 0.6948775055679287,
      "grad_norm": 1.8711456060409546,
      "learning_rate": 4.5174663093801674e-05,
      "loss": 0.3764048218727112,
      "step": 1443
    },
    {
      "epoch": 0.6953590561608379,
      "grad_norm": 1.8176897764205933,
      "learning_rate": 4.504427540175181e-05,
      "loss": 0.6044738292694092,
      "step": 1444
    },
    {
      "epoch": 0.695840606753747,
      "grad_norm": 2.3583931922912598,
      "learning_rate": 4.491402142886922e-05,
      "loss": 0.43546566367149353,
      "step": 1445
    },
    {
      "epoch": 0.6963221573466563,
      "grad_norm": 1.7809016704559326,
      "learning_rate": 4.4783901492089984e-05,
      "loss": 1.0412065982818604,
      "step": 1446
    },
    {
      "epoch": 0.6968037079395654,
      "grad_norm": 1.316541075706482,
      "learning_rate": 4.465391590802407e-05,
      "loss": 0.5663067698478699,
      "step": 1447
    },
    {
      "epoch": 0.6972852585324746,
      "grad_norm": 1.7204660177230835,
      "learning_rate": 4.4524064992954516e-05,
      "loss": 0.8451316952705383,
      "step": 1448
    },
    {
      "epoch": 0.6977668091253837,
      "grad_norm": 1.29192054271698,
      "learning_rate": 4.4394349062836736e-05,
      "loss": 0.7978814244270325,
      "step": 1449
    },
    {
      "epoch": 0.6982483597182929,
      "grad_norm": 2.1909849643707275,
      "learning_rate": 4.4264768433297565e-05,
      "loss": 0.7180624604225159,
      "step": 1450
    },
    {
      "epoch": 0.698729910311202,
      "grad_norm": 1.8029638528823853,
      "learning_rate": 4.4135323419634766e-05,
      "loss": 1.070571780204773,
      "step": 1451
    },
    {
      "epoch": 0.6992114609041112,
      "grad_norm": 1.7395559549331665,
      "learning_rate": 4.4006014336816035e-05,
      "loss": 0.5328083038330078,
      "step": 1452
    },
    {
      "epoch": 0.6996930114970205,
      "grad_norm": 1.4880378246307373,
      "learning_rate": 4.387684149947837e-05,
      "loss": 0.6413246393203735,
      "step": 1453
    },
    {
      "epoch": 0.7001745620899296,
      "grad_norm": 1.763173222541809,
      "learning_rate": 4.374780522192726e-05,
      "loss": 1.100099802017212,
      "step": 1454
    },
    {
      "epoch": 0.7006561126828388,
      "grad_norm": 3.848066568374634,
      "learning_rate": 4.3618905818135805e-05,
      "loss": 0.8637110590934753,
      "step": 1455
    },
    {
      "epoch": 0.7011376632757479,
      "grad_norm": 2.2468152046203613,
      "learning_rate": 4.349014360174417e-05,
      "loss": 0.9409898519515991,
      "step": 1456
    },
    {
      "epoch": 0.7016192138686571,
      "grad_norm": 2.685436964035034,
      "learning_rate": 4.336151888605871e-05,
      "loss": 0.8795504570007324,
      "step": 1457
    },
    {
      "epoch": 0.7021007644615662,
      "grad_norm": 2.1604654788970947,
      "learning_rate": 4.323303198405117e-05,
      "loss": 1.1707496643066406,
      "step": 1458
    },
    {
      "epoch": 0.7025823150544754,
      "grad_norm": 2.949472665786743,
      "learning_rate": 4.310468320835796e-05,
      "loss": 0.7554198503494263,
      "step": 1459
    },
    {
      "epoch": 0.7030638656473845,
      "grad_norm": 2.6601123809814453,
      "learning_rate": 4.297647287127946e-05,
      "loss": 1.2713550329208374,
      "step": 1460
    },
    {
      "epoch": 0.7035454162402938,
      "grad_norm": 3.171978235244751,
      "learning_rate": 4.284840128477913e-05,
      "loss": 0.8736757040023804,
      "step": 1461
    },
    {
      "epoch": 0.7040269668332029,
      "grad_norm": 3.365537643432617,
      "learning_rate": 4.2720468760482854e-05,
      "loss": 0.8785121440887451,
      "step": 1462
    },
    {
      "epoch": 0.7045085174261121,
      "grad_norm": 1.3108716011047363,
      "learning_rate": 4.2592675609678135e-05,
      "loss": 0.6028298139572144,
      "step": 1463
    },
    {
      "epoch": 0.7049900680190212,
      "grad_norm": 1.8261479139328003,
      "learning_rate": 4.24650221433134e-05,
      "loss": 0.697373628616333,
      "step": 1464
    },
    {
      "epoch": 0.7054716186119304,
      "grad_norm": 0.8967378735542297,
      "learning_rate": 4.2337508671997086e-05,
      "loss": 0.44815781712532043,
      "step": 1465
    },
    {
      "epoch": 0.7059531692048396,
      "grad_norm": 2.953753709793091,
      "learning_rate": 4.221013550599707e-05,
      "loss": 0.6674547791481018,
      "step": 1466
    },
    {
      "epoch": 0.7064347197977487,
      "grad_norm": 1.9956505298614502,
      "learning_rate": 4.208290295523984e-05,
      "loss": 1.0039540529251099,
      "step": 1467
    },
    {
      "epoch": 0.706916270390658,
      "grad_norm": 4.109732151031494,
      "learning_rate": 4.1955811329309746e-05,
      "loss": 1.230612874031067,
      "step": 1468
    },
    {
      "epoch": 0.7073978209835671,
      "grad_norm": 3.3807551860809326,
      "learning_rate": 4.182886093744813e-05,
      "loss": 1.2144087553024292,
      "step": 1469
    },
    {
      "epoch": 0.7078793715764763,
      "grad_norm": 2.252040147781372,
      "learning_rate": 4.170205208855281e-05,
      "loss": 0.3772548735141754,
      "step": 1470
    },
    {
      "epoch": 0.7083609221693854,
      "grad_norm": 1.2868578433990479,
      "learning_rate": 4.157538509117714e-05,
      "loss": 0.6164572238922119,
      "step": 1471
    },
    {
      "epoch": 0.7088424727622946,
      "grad_norm": 1.6125785112380981,
      "learning_rate": 4.144886025352934e-05,
      "loss": 1.0032025575637817,
      "step": 1472
    },
    {
      "epoch": 0.7093240233552037,
      "grad_norm": 1.678605318069458,
      "learning_rate": 4.13224778834717e-05,
      "loss": 0.5588351488113403,
      "step": 1473
    },
    {
      "epoch": 0.7098055739481129,
      "grad_norm": 2.6014554500579834,
      "learning_rate": 4.1196238288519874e-05,
      "loss": 0.8449669480323792,
      "step": 1474
    },
    {
      "epoch": 0.710287124541022,
      "grad_norm": 1.7891353368759155,
      "learning_rate": 4.107014177584211e-05,
      "loss": 0.7263489365577698,
      "step": 1475
    },
    {
      "epoch": 0.7107686751339313,
      "grad_norm": 1.4081364870071411,
      "learning_rate": 4.094418865225853e-05,
      "loss": 0.5538519620895386,
      "step": 1476
    },
    {
      "epoch": 0.7112502257268404,
      "grad_norm": 2.760237455368042,
      "learning_rate": 4.081837922424027e-05,
      "loss": 1.1000702381134033,
      "step": 1477
    },
    {
      "epoch": 0.7117317763197496,
      "grad_norm": 2.332379102706909,
      "learning_rate": 4.069271379790891e-05,
      "loss": 0.6511553525924683,
      "step": 1478
    },
    {
      "epoch": 0.7122133269126587,
      "grad_norm": 1.203919768333435,
      "learning_rate": 4.0567192679035636e-05,
      "loss": 0.7412837147712708,
      "step": 1479
    },
    {
      "epoch": 0.7126948775055679,
      "grad_norm": 1.9322272539138794,
      "learning_rate": 4.044181617304048e-05,
      "loss": 0.7486695647239685,
      "step": 1480
    },
    {
      "epoch": 0.7131764280984771,
      "grad_norm": 3.997180938720703,
      "learning_rate": 4.03165845849916e-05,
      "loss": 0.9312503337860107,
      "step": 1481
    },
    {
      "epoch": 0.7136579786913863,
      "grad_norm": 2.0899388790130615,
      "learning_rate": 4.019149821960455e-05,
      "loss": 0.4357759952545166,
      "step": 1482
    },
    {
      "epoch": 0.7141395292842955,
      "grad_norm": 1.9055157899856567,
      "learning_rate": 4.006655738124152e-05,
      "loss": 0.74583899974823,
      "step": 1483
    },
    {
      "epoch": 0.7146210798772046,
      "grad_norm": 1.6668399572372437,
      "learning_rate": 3.9941762373910586e-05,
      "loss": 0.8640475273132324,
      "step": 1484
    },
    {
      "epoch": 0.7151026304701138,
      "grad_norm": 1.9727829694747925,
      "learning_rate": 3.9817113501265016e-05,
      "loss": 0.4828895032405853,
      "step": 1485
    },
    {
      "epoch": 0.7155841810630229,
      "grad_norm": 3.055988073348999,
      "learning_rate": 3.9692611066602516e-05,
      "loss": 0.6246371269226074,
      "step": 1486
    },
    {
      "epoch": 0.7160657316559321,
      "grad_norm": 1.6440119743347168,
      "learning_rate": 3.956825537286436e-05,
      "loss": 0.632970929145813,
      "step": 1487
    },
    {
      "epoch": 0.7165472822488412,
      "grad_norm": 2.3147213459014893,
      "learning_rate": 3.944404672263494e-05,
      "loss": 1.031375765800476,
      "step": 1488
    },
    {
      "epoch": 0.7170288328417505,
      "grad_norm": 3.188918352127075,
      "learning_rate": 3.931998541814069e-05,
      "loss": 0.9395540356636047,
      "step": 1489
    },
    {
      "epoch": 0.7175103834346596,
      "grad_norm": 3.050300121307373,
      "learning_rate": 3.919607176124966e-05,
      "loss": 0.8490335941314697,
      "step": 1490
    },
    {
      "epoch": 0.7179919340275688,
      "grad_norm": 2.2424325942993164,
      "learning_rate": 3.9072306053470566e-05,
      "loss": 0.4993773102760315,
      "step": 1491
    },
    {
      "epoch": 0.7184734846204779,
      "grad_norm": 3.1395554542541504,
      "learning_rate": 3.8948688595952164e-05,
      "loss": 0.9753080010414124,
      "step": 1492
    },
    {
      "epoch": 0.7189550352133871,
      "grad_norm": 1.3993074893951416,
      "learning_rate": 3.882521968948246e-05,
      "loss": 0.7700310945510864,
      "step": 1493
    },
    {
      "epoch": 0.7194365858062963,
      "grad_norm": 1.5683131217956543,
      "learning_rate": 3.8701899634488014e-05,
      "loss": 0.5624740719795227,
      "step": 1494
    },
    {
      "epoch": 0.7199181363992054,
      "grad_norm": 1.920570969581604,
      "learning_rate": 3.857872873103322e-05,
      "loss": 0.5469199419021606,
      "step": 1495
    },
    {
      "epoch": 0.7203996869921147,
      "grad_norm": 1.838671088218689,
      "learning_rate": 3.8455707278819507e-05,
      "loss": 0.883967399597168,
      "step": 1496
    },
    {
      "epoch": 0.7208812375850238,
      "grad_norm": 3.9505457878112793,
      "learning_rate": 3.833283557718471e-05,
      "loss": 0.762613832950592,
      "step": 1497
    },
    {
      "epoch": 0.721362788177933,
      "grad_norm": 1.6243174076080322,
      "learning_rate": 3.821011392510228e-05,
      "loss": 0.9172532558441162,
      "step": 1498
    },
    {
      "epoch": 0.7218443387708421,
      "grad_norm": 2.40016770362854,
      "learning_rate": 3.808754262118046e-05,
      "loss": 0.4597705006599426,
      "step": 1499
    },
    {
      "epoch": 0.7223258893637513,
      "grad_norm": 1.8272314071655273,
      "learning_rate": 3.796512196366182e-05,
      "loss": 0.6962916851043701,
      "step": 1500
    },
    {
      "epoch": 0.7228074399566604,
      "grad_norm": 2.057837963104248,
      "learning_rate": 3.784285225042229e-05,
      "loss": 0.7898488640785217,
      "step": 1501
    },
    {
      "epoch": 0.7232889905495696,
      "grad_norm": 2.8366823196411133,
      "learning_rate": 3.772073377897052e-05,
      "loss": 0.9540159702301025,
      "step": 1502
    },
    {
      "epoch": 0.7237705411424787,
      "grad_norm": 2.988560199737549,
      "learning_rate": 3.7598766846447184e-05,
      "loss": 1.212519884109497,
      "step": 1503
    },
    {
      "epoch": 0.724252091735388,
      "grad_norm": 2.815720319747925,
      "learning_rate": 3.747695174962423e-05,
      "loss": 0.9854691028594971,
      "step": 1504
    },
    {
      "epoch": 0.7247336423282971,
      "grad_norm": 4.56036901473999,
      "learning_rate": 3.7355288784904116e-05,
      "loss": 0.7997022271156311,
      "step": 1505
    },
    {
      "epoch": 0.7252151929212063,
      "grad_norm": 2.630113363265991,
      "learning_rate": 3.7233778248319176e-05,
      "loss": 0.6918640732765198,
      "step": 1506
    },
    {
      "epoch": 0.7256967435141155,
      "grad_norm": 2.0350565910339355,
      "learning_rate": 3.7112420435530845e-05,
      "loss": 0.5981045961380005,
      "step": 1507
    },
    {
      "epoch": 0.7261782941070246,
      "grad_norm": 2.4248499870300293,
      "learning_rate": 3.69912156418289e-05,
      "loss": 0.8709573745727539,
      "step": 1508
    },
    {
      "epoch": 0.7266598446999338,
      "grad_norm": 1.0892748832702637,
      "learning_rate": 3.687016416213084e-05,
      "loss": 0.43704891204833984,
      "step": 1509
    },
    {
      "epoch": 0.7271413952928429,
      "grad_norm": 2.2195944786071777,
      "learning_rate": 3.674926629098113e-05,
      "loss": 1.0598080158233643,
      "step": 1510
    },
    {
      "epoch": 0.7276229458857522,
      "grad_norm": 2.175382375717163,
      "learning_rate": 3.6628522322550394e-05,
      "loss": 0.6169099807739258,
      "step": 1511
    },
    {
      "epoch": 0.7281044964786613,
      "grad_norm": 1.8716521263122559,
      "learning_rate": 3.6507932550634846e-05,
      "loss": 0.7334168553352356,
      "step": 1512
    },
    {
      "epoch": 0.7285860470715705,
      "grad_norm": 2.5147321224212646,
      "learning_rate": 3.638749726865552e-05,
      "loss": 0.5663888454437256,
      "step": 1513
    },
    {
      "epoch": 0.7290675976644796,
      "grad_norm": 2.403719425201416,
      "learning_rate": 3.6267216769657485e-05,
      "loss": 0.8791754245758057,
      "step": 1514
    },
    {
      "epoch": 0.7295491482573888,
      "grad_norm": 1.8424488306045532,
      "learning_rate": 3.6147091346309224e-05,
      "loss": 0.9261001944541931,
      "step": 1515
    },
    {
      "epoch": 0.7300306988502979,
      "grad_norm": 3.36635684967041,
      "learning_rate": 3.602712129090189e-05,
      "loss": 0.5302868485450745,
      "step": 1516
    },
    {
      "epoch": 0.7305122494432071,
      "grad_norm": 1.8761022090911865,
      "learning_rate": 3.590730689534857e-05,
      "loss": 0.44378042221069336,
      "step": 1517
    },
    {
      "epoch": 0.7309938000361162,
      "grad_norm": 2.5761327743530273,
      "learning_rate": 3.578764845118362e-05,
      "loss": 0.9266976714134216,
      "step": 1518
    },
    {
      "epoch": 0.7314753506290255,
      "grad_norm": 1.0269430875778198,
      "learning_rate": 3.566814624956194e-05,
      "loss": 0.6342741847038269,
      "step": 1519
    },
    {
      "epoch": 0.7319569012219346,
      "grad_norm": 3.7801918983459473,
      "learning_rate": 3.554880058125819e-05,
      "loss": 0.9923316240310669,
      "step": 1520
    },
    {
      "epoch": 0.7324384518148438,
      "grad_norm": 1.366782784461975,
      "learning_rate": 3.5429611736666235e-05,
      "loss": 0.3871268630027771,
      "step": 1521
    },
    {
      "epoch": 0.732920002407753,
      "grad_norm": 2.8801584243774414,
      "learning_rate": 3.53105800057983e-05,
      "loss": 0.46662548184394836,
      "step": 1522
    },
    {
      "epoch": 0.7334015530006621,
      "grad_norm": 1.9568599462509155,
      "learning_rate": 3.519170567828435e-05,
      "loss": 0.8205584287643433,
      "step": 1523
    },
    {
      "epoch": 0.7338831035935713,
      "grad_norm": 2.8359971046447754,
      "learning_rate": 3.507298904337134e-05,
      "loss": 0.646842896938324,
      "step": 1524
    },
    {
      "epoch": 0.7343646541864804,
      "grad_norm": 1.3017363548278809,
      "learning_rate": 3.495443038992253e-05,
      "loss": 0.34668368101119995,
      "step": 1525
    },
    {
      "epoch": 0.7348462047793897,
      "grad_norm": 8.52542781829834,
      "learning_rate": 3.4836030006416775e-05,
      "loss": 0.6620774269104004,
      "step": 1526
    },
    {
      "epoch": 0.7353277553722988,
      "grad_norm": 2.0775399208068848,
      "learning_rate": 3.471778818094785e-05,
      "loss": 0.7691874504089355,
      "step": 1527
    },
    {
      "epoch": 0.735809305965208,
      "grad_norm": 2.1365134716033936,
      "learning_rate": 3.459970520122364e-05,
      "loss": 0.7845427393913269,
      "step": 1528
    },
    {
      "epoch": 0.7362908565581171,
      "grad_norm": 2.4064700603485107,
      "learning_rate": 3.44817813545656e-05,
      "loss": 0.4174688160419464,
      "step": 1529
    },
    {
      "epoch": 0.7367724071510263,
      "grad_norm": 2.0866596698760986,
      "learning_rate": 3.4364016927907974e-05,
      "loss": 0.7920362949371338,
      "step": 1530
    },
    {
      "epoch": 0.7372539577439354,
      "grad_norm": 2.026822090148926,
      "learning_rate": 3.424641220779711e-05,
      "loss": 0.9459177851676941,
      "step": 1531
    },
    {
      "epoch": 0.7377355083368446,
      "grad_norm": 2.546722888946533,
      "learning_rate": 3.412896748039067e-05,
      "loss": 0.6286554932594299,
      "step": 1532
    },
    {
      "epoch": 0.7382170589297538,
      "grad_norm": 1.5018588304519653,
      "learning_rate": 3.401168303145713e-05,
      "loss": 1.2848738431930542,
      "step": 1533
    },
    {
      "epoch": 0.738698609522663,
      "grad_norm": 6.512143611907959,
      "learning_rate": 3.3894559146374924e-05,
      "loss": 0.7121766805648804,
      "step": 1534
    },
    {
      "epoch": 0.7391801601155722,
      "grad_norm": 1.072776198387146,
      "learning_rate": 3.37775961101318e-05,
      "loss": 0.36095431447029114,
      "step": 1535
    },
    {
      "epoch": 0.7396617107084813,
      "grad_norm": 3.9811277389526367,
      "learning_rate": 3.366079420732413e-05,
      "loss": 0.6103596687316895,
      "step": 1536
    },
    {
      "epoch": 0.7401432613013905,
      "grad_norm": 4.248976230621338,
      "learning_rate": 3.3544153722156216e-05,
      "loss": 0.6014701128005981,
      "step": 1537
    },
    {
      "epoch": 0.7406248118942996,
      "grad_norm": 1.3992478847503662,
      "learning_rate": 3.3427674938439594e-05,
      "loss": 0.911087155342102,
      "step": 1538
    },
    {
      "epoch": 0.7411063624872088,
      "grad_norm": 2.1802425384521484,
      "learning_rate": 3.3311358139592317e-05,
      "loss": 0.8243244886398315,
      "step": 1539
    },
    {
      "epoch": 0.741587913080118,
      "grad_norm": 2.9368648529052734,
      "learning_rate": 3.319520360863837e-05,
      "loss": 0.566057026386261,
      "step": 1540
    },
    {
      "epoch": 0.7420694636730272,
      "grad_norm": 2.241831064224243,
      "learning_rate": 3.3079211628206854e-05,
      "loss": 0.6615368723869324,
      "step": 1541
    },
    {
      "epoch": 0.7425510142659363,
      "grad_norm": 2.107577085494995,
      "learning_rate": 3.296338248053129e-05,
      "loss": 0.7434890270233154,
      "step": 1542
    },
    {
      "epoch": 0.7430325648588455,
      "grad_norm": 3.612004280090332,
      "learning_rate": 3.2847716447449096e-05,
      "loss": 0.7647389769554138,
      "step": 1543
    },
    {
      "epoch": 0.7435141154517546,
      "grad_norm": 2.2122089862823486,
      "learning_rate": 3.2732213810400745e-05,
      "loss": 0.8760401010513306,
      "step": 1544
    },
    {
      "epoch": 0.7439956660446638,
      "grad_norm": 2.319847822189331,
      "learning_rate": 3.261687485042915e-05,
      "loss": 0.33415693044662476,
      "step": 1545
    },
    {
      "epoch": 0.7444772166375729,
      "grad_norm": 1.9580154418945312,
      "learning_rate": 3.250169984817897e-05,
      "loss": 0.5143443942070007,
      "step": 1546
    },
    {
      "epoch": 0.7449587672304822,
      "grad_norm": 1.7028541564941406,
      "learning_rate": 3.238668908389586e-05,
      "loss": 0.9557589888572693,
      "step": 1547
    },
    {
      "epoch": 0.7454403178233914,
      "grad_norm": 2.973677635192871,
      "learning_rate": 3.227184283742591e-05,
      "loss": 0.9735615849494934,
      "step": 1548
    },
    {
      "epoch": 0.7459218684163005,
      "grad_norm": 1.7476050853729248,
      "learning_rate": 3.215716138821488e-05,
      "loss": 0.6330348253250122,
      "step": 1549
    },
    {
      "epoch": 0.7464034190092097,
      "grad_norm": 1.5159329175949097,
      "learning_rate": 3.204264501530756e-05,
      "loss": 0.8052382469177246,
      "step": 1550
    },
    {
      "epoch": 0.7468849696021188,
      "grad_norm": 1.9966834783554077,
      "learning_rate": 3.192829399734706e-05,
      "loss": 0.6953385472297668,
      "step": 1551
    },
    {
      "epoch": 0.747366520195028,
      "grad_norm": 1.1455020904541016,
      "learning_rate": 3.181410861257413e-05,
      "loss": 0.886067807674408,
      "step": 1552
    },
    {
      "epoch": 0.7478480707879371,
      "grad_norm": 1.7974690198898315,
      "learning_rate": 3.170008913882656e-05,
      "loss": 0.7926231622695923,
      "step": 1553
    },
    {
      "epoch": 0.7483296213808464,
      "grad_norm": 2.3232922554016113,
      "learning_rate": 3.1586235853538325e-05,
      "loss": 0.6949499845504761,
      "step": 1554
    },
    {
      "epoch": 0.7488111719737555,
      "grad_norm": 3.2760112285614014,
      "learning_rate": 3.1472549033739126e-05,
      "loss": 0.875335693359375,
      "step": 1555
    },
    {
      "epoch": 0.7492927225666647,
      "grad_norm": 1.650658130645752,
      "learning_rate": 3.1359028956053615e-05,
      "loss": 0.511374294757843,
      "step": 1556
    },
    {
      "epoch": 0.7497742731595738,
      "grad_norm": 1.5674755573272705,
      "learning_rate": 3.1245675896700685e-05,
      "loss": 0.8766633868217468,
      "step": 1557
    },
    {
      "epoch": 0.750255823752483,
      "grad_norm": 2.0909409523010254,
      "learning_rate": 3.113249013149284e-05,
      "loss": 0.6018157005310059,
      "step": 1558
    },
    {
      "epoch": 0.7507373743453921,
      "grad_norm": 2.340400457382202,
      "learning_rate": 3.101947193583557e-05,
      "loss": 1.5338536500930786,
      "step": 1559
    },
    {
      "epoch": 0.7512189249383013,
      "grad_norm": 2.725926637649536,
      "learning_rate": 3.0906621584726546e-05,
      "loss": 0.9796492457389832,
      "step": 1560
    },
    {
      "epoch": 0.7517004755312104,
      "grad_norm": 1.3635221719741821,
      "learning_rate": 3.079393935275513e-05,
      "loss": 0.554710328578949,
      "step": 1561
    },
    {
      "epoch": 0.7521820261241197,
      "grad_norm": 2.1766269207000732,
      "learning_rate": 3.068142551410155e-05,
      "loss": 0.46962887048721313,
      "step": 1562
    },
    {
      "epoch": 0.7526635767170289,
      "grad_norm": 1.7623589038848877,
      "learning_rate": 3.0569080342536347e-05,
      "loss": 0.7526323795318604,
      "step": 1563
    },
    {
      "epoch": 0.753145127309938,
      "grad_norm": 2.600128412246704,
      "learning_rate": 3.0456904111419572e-05,
      "loss": 0.618369996547699,
      "step": 1564
    },
    {
      "epoch": 0.7536266779028472,
      "grad_norm": 3.250112533569336,
      "learning_rate": 3.034489709370033e-05,
      "loss": 1.1798557043075562,
      "step": 1565
    },
    {
      "epoch": 0.7541082284957563,
      "grad_norm": 3.693274736404419,
      "learning_rate": 3.0233059561915855e-05,
      "loss": 0.6116794347763062,
      "step": 1566
    },
    {
      "epoch": 0.7545897790886655,
      "grad_norm": 2.126807928085327,
      "learning_rate": 3.01213917881911e-05,
      "loss": 0.7569847106933594,
      "step": 1567
    },
    {
      "epoch": 0.7550713296815746,
      "grad_norm": 2.0772931575775146,
      "learning_rate": 3.0009894044237907e-05,
      "loss": 0.7947670817375183,
      "step": 1568
    },
    {
      "epoch": 0.7555528802744839,
      "grad_norm": 3.616492509841919,
      "learning_rate": 2.9898566601354418e-05,
      "loss": 1.052865982055664,
      "step": 1569
    },
    {
      "epoch": 0.756034430867393,
      "grad_norm": 2.3350512981414795,
      "learning_rate": 2.9787409730424374e-05,
      "loss": 0.9928959012031555,
      "step": 1570
    },
    {
      "epoch": 0.7565159814603022,
      "grad_norm": 2.0888309478759766,
      "learning_rate": 2.96764237019165e-05,
      "loss": 1.0255086421966553,
      "step": 1571
    },
    {
      "epoch": 0.7569975320532113,
      "grad_norm": 2.032150983810425,
      "learning_rate": 2.9565608785883815e-05,
      "loss": 0.9266300797462463,
      "step": 1572
    },
    {
      "epoch": 0.7574790826461205,
      "grad_norm": 1.7835674285888672,
      "learning_rate": 2.9454965251962973e-05,
      "loss": 1.0870325565338135,
      "step": 1573
    },
    {
      "epoch": 0.7579606332390296,
      "grad_norm": 2.513573169708252,
      "learning_rate": 2.9344493369373637e-05,
      "loss": 0.8469479084014893,
      "step": 1574
    },
    {
      "epoch": 0.7584421838319388,
      "grad_norm": 1.5565505027770996,
      "learning_rate": 2.9234193406917833e-05,
      "loss": 0.7374131679534912,
      "step": 1575
    },
    {
      "epoch": 0.7589237344248481,
      "grad_norm": 4.1210408210754395,
      "learning_rate": 2.912406563297916e-05,
      "loss": 0.677290678024292,
      "step": 1576
    },
    {
      "epoch": 0.7594052850177572,
      "grad_norm": 2.406564950942993,
      "learning_rate": 2.901411031552236e-05,
      "loss": 0.6457397937774658,
      "step": 1577
    },
    {
      "epoch": 0.7598868356106664,
      "grad_norm": 0.6361212730407715,
      "learning_rate": 2.8904327722092495e-05,
      "loss": 0.4490494728088379,
      "step": 1578
    },
    {
      "epoch": 0.7603683862035755,
      "grad_norm": 2.1413636207580566,
      "learning_rate": 2.879471811981437e-05,
      "loss": 0.9692299365997314,
      "step": 1579
    },
    {
      "epoch": 0.7608499367964847,
      "grad_norm": 2.651672840118408,
      "learning_rate": 2.868528177539187e-05,
      "loss": 0.9879105687141418,
      "step": 1580
    },
    {
      "epoch": 0.7613314873893938,
      "grad_norm": 1.2825648784637451,
      "learning_rate": 2.8576018955107285e-05,
      "loss": 0.8283687829971313,
      "step": 1581
    },
    {
      "epoch": 0.761813037982303,
      "grad_norm": 2.5643911361694336,
      "learning_rate": 2.8466929924820705e-05,
      "loss": 1.06561279296875,
      "step": 1582
    },
    {
      "epoch": 0.7622945885752122,
      "grad_norm": 2.0799810886383057,
      "learning_rate": 2.8358014949969334e-05,
      "loss": 0.8342494964599609,
      "step": 1583
    },
    {
      "epoch": 0.7627761391681214,
      "grad_norm": 3.2869009971618652,
      "learning_rate": 2.8249274295566864e-05,
      "loss": 0.7339048981666565,
      "step": 1584
    },
    {
      "epoch": 0.7632576897610305,
      "grad_norm": 3.5507149696350098,
      "learning_rate": 2.8140708226202884e-05,
      "loss": 0.783934473991394,
      "step": 1585
    },
    {
      "epoch": 0.7637392403539397,
      "grad_norm": 3.1774659156799316,
      "learning_rate": 2.803231700604204e-05,
      "loss": 0.6622165441513062,
      "step": 1586
    },
    {
      "epoch": 0.7642207909468488,
      "grad_norm": 1.691411018371582,
      "learning_rate": 2.7924100898823702e-05,
      "loss": 0.37145382165908813,
      "step": 1587
    },
    {
      "epoch": 0.764702341539758,
      "grad_norm": 1.7954684495925903,
      "learning_rate": 2.7816060167861002e-05,
      "loss": 1.132523536682129,
      "step": 1588
    },
    {
      "epoch": 0.7651838921326672,
      "grad_norm": 1.7289997339248657,
      "learning_rate": 2.7708195076040445e-05,
      "loss": 0.5267021656036377,
      "step": 1589
    },
    {
      "epoch": 0.7656654427255764,
      "grad_norm": 2.7406585216522217,
      "learning_rate": 2.760050588582114e-05,
      "loss": 0.7358564138412476,
      "step": 1590
    },
    {
      "epoch": 0.7661469933184856,
      "grad_norm": 0.7409334778785706,
      "learning_rate": 2.749299285923417e-05,
      "loss": 0.5098696947097778,
      "step": 1591
    },
    {
      "epoch": 0.7666285439113947,
      "grad_norm": 1.8043601512908936,
      "learning_rate": 2.7385656257881997e-05,
      "loss": 0.958987295627594,
      "step": 1592
    },
    {
      "epoch": 0.7671100945043039,
      "grad_norm": 3.150944232940674,
      "learning_rate": 2.7278496342937788e-05,
      "loss": 0.7128248810768127,
      "step": 1593
    },
    {
      "epoch": 0.767591645097213,
      "grad_norm": 1.7216988801956177,
      "learning_rate": 2.717151337514482e-05,
      "loss": 0.716182291507721,
      "step": 1594
    },
    {
      "epoch": 0.7680731956901222,
      "grad_norm": 1.5065703392028809,
      "learning_rate": 2.7064707614815776e-05,
      "loss": 0.5126557946205139,
      "step": 1595
    },
    {
      "epoch": 0.7685547462830313,
      "grad_norm": 2.709780693054199,
      "learning_rate": 2.6958079321832185e-05,
      "loss": 0.7355021238327026,
      "step": 1596
    },
    {
      "epoch": 0.7690362968759406,
      "grad_norm": 2.993232488632202,
      "learning_rate": 2.6851628755643776e-05,
      "loss": 0.6603368520736694,
      "step": 1597
    },
    {
      "epoch": 0.7695178474688497,
      "grad_norm": 1.8010609149932861,
      "learning_rate": 2.6745356175267765e-05,
      "loss": 0.47278061509132385,
      "step": 1598
    },
    {
      "epoch": 0.7699993980617589,
      "grad_norm": 2.456239938735962,
      "learning_rate": 2.6639261839288343e-05,
      "loss": 0.7185255289077759,
      "step": 1599
    },
    {
      "epoch": 0.770480948654668,
      "grad_norm": 2.0750105381011963,
      "learning_rate": 2.6533346005855987e-05,
      "loss": 0.6045894622802734,
      "step": 1600
    },
    {
      "epoch": 0.7709624992475772,
      "grad_norm": 1.652189016342163,
      "learning_rate": 2.6427608932686843e-05,
      "loss": 0.7198444604873657,
      "step": 1601
    },
    {
      "epoch": 0.7714440498404863,
      "grad_norm": 1.5234538316726685,
      "learning_rate": 2.6322050877062064e-05,
      "loss": 0.48733392357826233,
      "step": 1602
    },
    {
      "epoch": 0.7719256004333955,
      "grad_norm": 2.0894863605499268,
      "learning_rate": 2.6216672095827266e-05,
      "loss": 0.5220861434936523,
      "step": 1603
    },
    {
      "epoch": 0.7724071510263048,
      "grad_norm": 6.180787563323975,
      "learning_rate": 2.6111472845391827e-05,
      "loss": 1.098311424255371,
      "step": 1604
    },
    {
      "epoch": 0.7728887016192139,
      "grad_norm": 2.3121771812438965,
      "learning_rate": 2.6006453381728236e-05,
      "loss": 0.5790321230888367,
      "step": 1605
    },
    {
      "epoch": 0.7733702522121231,
      "grad_norm": 2.610154867172241,
      "learning_rate": 2.5901613960371585e-05,
      "loss": 0.7664477825164795,
      "step": 1606
    },
    {
      "epoch": 0.7738518028050322,
      "grad_norm": 2.02698016166687,
      "learning_rate": 2.5796954836418884e-05,
      "loss": 0.6946688294410706,
      "step": 1607
    },
    {
      "epoch": 0.7743333533979414,
      "grad_norm": 2.8777859210968018,
      "learning_rate": 2.569247626452842e-05,
      "loss": 0.4291987717151642,
      "step": 1608
    },
    {
      "epoch": 0.7748149039908505,
      "grad_norm": 2.4330689907073975,
      "learning_rate": 2.558817849891918e-05,
      "loss": 1.0675934553146362,
      "step": 1609
    },
    {
      "epoch": 0.7752964545837597,
      "grad_norm": 2.020845890045166,
      "learning_rate": 2.548406179337015e-05,
      "loss": 0.7503652572631836,
      "step": 1610
    },
    {
      "epoch": 0.7757780051766688,
      "grad_norm": 2.804804801940918,
      "learning_rate": 2.5380126401219807e-05,
      "loss": 0.8434704542160034,
      "step": 1611
    },
    {
      "epoch": 0.7762595557695781,
      "grad_norm": 1.912571668624878,
      "learning_rate": 2.527637257536547e-05,
      "loss": 0.45627743005752563,
      "step": 1612
    },
    {
      "epoch": 0.7767411063624872,
      "grad_norm": 2.859957456588745,
      "learning_rate": 2.517280056826262e-05,
      "loss": 0.7084078192710876,
      "step": 1613
    },
    {
      "epoch": 0.7772226569553964,
      "grad_norm": 2.7481119632720947,
      "learning_rate": 2.5069410631924385e-05,
      "loss": 1.0676352977752686,
      "step": 1614
    },
    {
      "epoch": 0.7777042075483055,
      "grad_norm": 1.3579630851745605,
      "learning_rate": 2.4966203017920818e-05,
      "loss": 0.5575138926506042,
      "step": 1615
    },
    {
      "epoch": 0.7781857581412147,
      "grad_norm": 1.937870979309082,
      "learning_rate": 2.4863177977378392e-05,
      "loss": 0.6939438581466675,
      "step": 1616
    },
    {
      "epoch": 0.7786673087341239,
      "grad_norm": 1.653530478477478,
      "learning_rate": 2.4760335760979312e-05,
      "loss": 0.7483892440795898,
      "step": 1617
    },
    {
      "epoch": 0.779148859327033,
      "grad_norm": 3.4126288890838623,
      "learning_rate": 2.4657676618960944e-05,
      "loss": 0.7157730460166931,
      "step": 1618
    },
    {
      "epoch": 0.7796304099199423,
      "grad_norm": 2.5448575019836426,
      "learning_rate": 2.455520080111522e-05,
      "loss": 0.8743493556976318,
      "step": 1619
    },
    {
      "epoch": 0.7801119605128514,
      "grad_norm": 1.0214663743972778,
      "learning_rate": 2.4452908556787912e-05,
      "loss": 0.44510790705680847,
      "step": 1620
    },
    {
      "epoch": 0.7805935111057606,
      "grad_norm": 1.8694747686386108,
      "learning_rate": 2.4350800134878203e-05,
      "loss": 1.040855050086975,
      "step": 1621
    },
    {
      "epoch": 0.7810750616986697,
      "grad_norm": 2.1782407760620117,
      "learning_rate": 2.4248875783837987e-05,
      "loss": 0.4617324471473694,
      "step": 1622
    },
    {
      "epoch": 0.7815566122915789,
      "grad_norm": 2.144824504852295,
      "learning_rate": 2.414713575167129e-05,
      "loss": 0.4272627532482147,
      "step": 1623
    },
    {
      "epoch": 0.782038162884488,
      "grad_norm": 1.950473666191101,
      "learning_rate": 2.4045580285933557e-05,
      "loss": 0.35621756315231323,
      "step": 1624
    },
    {
      "epoch": 0.7825197134773972,
      "grad_norm": 1.6698945760726929,
      "learning_rate": 2.3944209633731242e-05,
      "loss": 0.8533354997634888,
      "step": 1625
    },
    {
      "epoch": 0.7830012640703063,
      "grad_norm": 2.3374836444854736,
      "learning_rate": 2.3843024041721053e-05,
      "loss": 0.7404719591140747,
      "step": 1626
    },
    {
      "epoch": 0.7834828146632156,
      "grad_norm": 2.028017044067383,
      "learning_rate": 2.3742023756109456e-05,
      "loss": 0.4868413209915161,
      "step": 1627
    },
    {
      "epoch": 0.7839643652561247,
      "grad_norm": 2.3631255626678467,
      "learning_rate": 2.3641209022651976e-05,
      "loss": 0.9500097632408142,
      "step": 1628
    },
    {
      "epoch": 0.7844459158490339,
      "grad_norm": 1.6879633665084839,
      "learning_rate": 2.3540580086652675e-05,
      "loss": 0.7224256992340088,
      "step": 1629
    },
    {
      "epoch": 0.7849274664419431,
      "grad_norm": 4.225047588348389,
      "learning_rate": 2.344013719296353e-05,
      "loss": 0.5835891962051392,
      "step": 1630
    },
    {
      "epoch": 0.7854090170348522,
      "grad_norm": 4.6645002365112305,
      "learning_rate": 2.3339880585983842e-05,
      "loss": 1.103783369064331,
      "step": 1631
    },
    {
      "epoch": 0.7858905676277614,
      "grad_norm": 2.486330270767212,
      "learning_rate": 2.3239810509659597e-05,
      "loss": 1.0096263885498047,
      "step": 1632
    },
    {
      "epoch": 0.7863721182206705,
      "grad_norm": 1.6724703311920166,
      "learning_rate": 2.313992720748295e-05,
      "loss": 0.7161979079246521,
      "step": 1633
    },
    {
      "epoch": 0.7868536688135798,
      "grad_norm": 1.9089106321334839,
      "learning_rate": 2.304023092249159e-05,
      "loss": 0.8530615568161011,
      "step": 1634
    },
    {
      "epoch": 0.7873352194064889,
      "grad_norm": 1.4664883613586426,
      "learning_rate": 2.2940721897268136e-05,
      "loss": 0.5826014280319214,
      "step": 1635
    },
    {
      "epoch": 0.7878167699993981,
      "grad_norm": 3.302122116088867,
      "learning_rate": 2.2841400373939592e-05,
      "loss": 0.8968521356582642,
      "step": 1636
    },
    {
      "epoch": 0.7882983205923072,
      "grad_norm": 2.6423821449279785,
      "learning_rate": 2.274226659417671e-05,
      "loss": 1.3045952320098877,
      "step": 1637
    },
    {
      "epoch": 0.7887798711852164,
      "grad_norm": 0.8947679996490479,
      "learning_rate": 2.2643320799193402e-05,
      "loss": 0.7163830995559692,
      "step": 1638
    },
    {
      "epoch": 0.7892614217781255,
      "grad_norm": 1.641568899154663,
      "learning_rate": 2.2544563229746218e-05,
      "loss": 0.5801525712013245,
      "step": 1639
    },
    {
      "epoch": 0.7897429723710347,
      "grad_norm": 1.639611005783081,
      "learning_rate": 2.2445994126133708e-05,
      "loss": 0.8533473610877991,
      "step": 1640
    },
    {
      "epoch": 0.7902245229639439,
      "grad_norm": 2.2356441020965576,
      "learning_rate": 2.234761372819577e-05,
      "loss": 0.7684714794158936,
      "step": 1641
    },
    {
      "epoch": 0.7907060735568531,
      "grad_norm": 1.257577657699585,
      "learning_rate": 2.2249422275313214e-05,
      "loss": 0.5451156497001648,
      "step": 1642
    },
    {
      "epoch": 0.7911876241497622,
      "grad_norm": 2.461451292037964,
      "learning_rate": 2.215142000640714e-05,
      "loss": 0.5267204642295837,
      "step": 1643
    },
    {
      "epoch": 0.7916691747426714,
      "grad_norm": 2.0822408199310303,
      "learning_rate": 2.2053607159938195e-05,
      "loss": 0.5773230195045471,
      "step": 1644
    },
    {
      "epoch": 0.7921507253355806,
      "grad_norm": 2.2660553455352783,
      "learning_rate": 2.1955983973906236e-05,
      "loss": 0.4283539056777954,
      "step": 1645
    },
    {
      "epoch": 0.7926322759284897,
      "grad_norm": 2.4742932319641113,
      "learning_rate": 2.1858550685849578e-05,
      "loss": 0.5106106996536255,
      "step": 1646
    },
    {
      "epoch": 0.793113826521399,
      "grad_norm": 1.9923046827316284,
      "learning_rate": 2.17613075328445e-05,
      "loss": 0.6857454776763916,
      "step": 1647
    },
    {
      "epoch": 0.793595377114308,
      "grad_norm": 2.169079065322876,
      "learning_rate": 2.1664254751504642e-05,
      "loss": 0.703352689743042,
      "step": 1648
    },
    {
      "epoch": 0.7940769277072173,
      "grad_norm": 1.056151270866394,
      "learning_rate": 2.1567392577980393e-05,
      "loss": 0.32297348976135254,
      "step": 1649
    },
    {
      "epoch": 0.7945584783001264,
      "grad_norm": 4.235377311706543,
      "learning_rate": 2.1470721247958404e-05,
      "loss": 0.7060508728027344,
      "step": 1650
    },
    {
      "epoch": 0.7950400288930356,
      "grad_norm": 1.5977118015289307,
      "learning_rate": 2.137424099666091e-05,
      "loss": 0.578859806060791,
      "step": 1651
    },
    {
      "epoch": 0.7955215794859447,
      "grad_norm": 5.629222393035889,
      "learning_rate": 2.1277952058845284e-05,
      "loss": 0.7520133256912231,
      "step": 1652
    },
    {
      "epoch": 0.7960031300788539,
      "grad_norm": 1.7176508903503418,
      "learning_rate": 2.118185466880327e-05,
      "loss": 0.561086118221283,
      "step": 1653
    },
    {
      "epoch": 0.796484680671763,
      "grad_norm": 2.4769234657287598,
      "learning_rate": 2.1085949060360654e-05,
      "loss": 0.8051009178161621,
      "step": 1654
    },
    {
      "epoch": 0.7969662312646723,
      "grad_norm": 2.3826639652252197,
      "learning_rate": 2.0990235466876517e-05,
      "loss": 0.763968825340271,
      "step": 1655
    },
    {
      "epoch": 0.7974477818575814,
      "grad_norm": 1.617227554321289,
      "learning_rate": 2.089471412124274e-05,
      "loss": 0.5491857528686523,
      "step": 1656
    },
    {
      "epoch": 0.7979293324504906,
      "grad_norm": 1.4059133529663086,
      "learning_rate": 2.079938525588342e-05,
      "loss": 0.5116761922836304,
      "step": 1657
    },
    {
      "epoch": 0.7984108830433998,
      "grad_norm": 1.7358564138412476,
      "learning_rate": 2.0704249102754324e-05,
      "loss": 0.6177835464477539,
      "step": 1658
    },
    {
      "epoch": 0.7988924336363089,
      "grad_norm": 2.2434935569763184,
      "learning_rate": 2.0609305893342278e-05,
      "loss": 0.6524842381477356,
      "step": 1659
    },
    {
      "epoch": 0.7993739842292181,
      "grad_norm": 3.120727777481079,
      "learning_rate": 2.0514555858664663e-05,
      "loss": 0.6625393629074097,
      "step": 1660
    },
    {
      "epoch": 0.7998555348221272,
      "grad_norm": 2.596045732498169,
      "learning_rate": 2.0419999229268805e-05,
      "loss": 0.5787920355796814,
      "step": 1661
    },
    {
      "epoch": 0.8003370854150365,
      "grad_norm": 4.540925025939941,
      "learning_rate": 2.032563623523147e-05,
      "loss": 1.3179080486297607,
      "step": 1662
    },
    {
      "epoch": 0.8008186360079456,
      "grad_norm": 1.5051778554916382,
      "learning_rate": 2.0231467106158186e-05,
      "loss": 0.536230742931366,
      "step": 1663
    },
    {
      "epoch": 0.8013001866008548,
      "grad_norm": 2.297910451889038,
      "learning_rate": 2.0137492071182863e-05,
      "loss": 0.6389889717102051,
      "step": 1664
    },
    {
      "epoch": 0.8017817371937639,
      "grad_norm": 2.963632583618164,
      "learning_rate": 2.0043711358967043e-05,
      "loss": 1.2386908531188965,
      "step": 1665
    },
    {
      "epoch": 0.8022632877866731,
      "grad_norm": 2.144054889678955,
      "learning_rate": 1.9950125197699508e-05,
      "loss": 0.6220329403877258,
      "step": 1666
    },
    {
      "epoch": 0.8027448383795822,
      "grad_norm": 1.0514551401138306,
      "learning_rate": 1.985673381509565e-05,
      "loss": 0.6354718208312988,
      "step": 1667
    },
    {
      "epoch": 0.8032263889724914,
      "grad_norm": 3.7318215370178223,
      "learning_rate": 1.9763537438396894e-05,
      "loss": 0.7919983863830566,
      "step": 1668
    },
    {
      "epoch": 0.8037079395654005,
      "grad_norm": 1.3612358570098877,
      "learning_rate": 1.96705362943702e-05,
      "loss": 0.5753484964370728,
      "step": 1669
    },
    {
      "epoch": 0.8041894901583098,
      "grad_norm": 3.6093125343322754,
      "learning_rate": 1.9577730609307454e-05,
      "loss": 0.8563456535339355,
      "step": 1670
    },
    {
      "epoch": 0.804671040751219,
      "grad_norm": 3.1442453861236572,
      "learning_rate": 1.9485120609024975e-05,
      "loss": 0.7687146067619324,
      "step": 1671
    },
    {
      "epoch": 0.8051525913441281,
      "grad_norm": 2.042131185531616,
      "learning_rate": 1.9392706518862935e-05,
      "loss": 0.690903902053833,
      "step": 1672
    },
    {
      "epoch": 0.8056341419370373,
      "grad_norm": 1.2224483489990234,
      "learning_rate": 1.9300488563684804e-05,
      "loss": 0.5825680494308472,
      "step": 1673
    },
    {
      "epoch": 0.8061156925299464,
      "grad_norm": 1.7768957614898682,
      "learning_rate": 1.920846696787684e-05,
      "loss": 0.4168967604637146,
      "step": 1674
    },
    {
      "epoch": 0.8065972431228556,
      "grad_norm": 2.0154497623443604,
      "learning_rate": 1.9116641955347446e-05,
      "loss": 0.5095771551132202,
      "step": 1675
    },
    {
      "epoch": 0.8070787937157647,
      "grad_norm": 2.258610248565674,
      "learning_rate": 1.9025013749526767e-05,
      "loss": 0.31659185886383057,
      "step": 1676
    },
    {
      "epoch": 0.807560344308674,
      "grad_norm": 1.6206918954849243,
      "learning_rate": 1.8933582573366036e-05,
      "loss": 0.7182268500328064,
      "step": 1677
    },
    {
      "epoch": 0.8080418949015831,
      "grad_norm": 2.3676815032958984,
      "learning_rate": 1.8842348649337116e-05,
      "loss": 0.9320851564407349,
      "step": 1678
    },
    {
      "epoch": 0.8085234454944923,
      "grad_norm": 1.4210022687911987,
      "learning_rate": 1.875131219943187e-05,
      "loss": 1.0072102546691895,
      "step": 1679
    },
    {
      "epoch": 0.8090049960874014,
      "grad_norm": 1.3253257274627686,
      "learning_rate": 1.8660473445161663e-05,
      "loss": 0.40254199504852295,
      "step": 1680
    },
    {
      "epoch": 0.8094865466803106,
      "grad_norm": 1.92241632938385,
      "learning_rate": 1.856983260755686e-05,
      "loss": 0.8012986779212952,
      "step": 1681
    },
    {
      "epoch": 0.8099680972732197,
      "grad_norm": 1.6853246688842773,
      "learning_rate": 1.8479389907166223e-05,
      "loss": 1.0961699485778809,
      "step": 1682
    },
    {
      "epoch": 0.8104496478661289,
      "grad_norm": 2.6904683113098145,
      "learning_rate": 1.8389145564056387e-05,
      "loss": 0.9050538539886475,
      "step": 1683
    },
    {
      "epoch": 0.810931198459038,
      "grad_norm": 2.022174119949341,
      "learning_rate": 1.829909979781137e-05,
      "loss": 0.4262884855270386,
      "step": 1684
    },
    {
      "epoch": 0.8114127490519473,
      "grad_norm": 1.5452542304992676,
      "learning_rate": 1.820925282753201e-05,
      "loss": 0.4250182509422302,
      "step": 1685
    },
    {
      "epoch": 0.8118942996448565,
      "grad_norm": 1.5475248098373413,
      "learning_rate": 1.8119604871835437e-05,
      "loss": 0.4848100543022156,
      "step": 1686
    },
    {
      "epoch": 0.8123758502377656,
      "grad_norm": 1.6167923212051392,
      "learning_rate": 1.8030156148854492e-05,
      "loss": 0.5589006543159485,
      "step": 1687
    },
    {
      "epoch": 0.8128574008306748,
      "grad_norm": 1.0089949369430542,
      "learning_rate": 1.7940906876237284e-05,
      "loss": 0.44827237725257874,
      "step": 1688
    },
    {
      "epoch": 0.8133389514235839,
      "grad_norm": 1.3869147300720215,
      "learning_rate": 1.78518572711466e-05,
      "loss": 0.41181278228759766,
      "step": 1689
    },
    {
      "epoch": 0.8138205020164931,
      "grad_norm": 2.3763556480407715,
      "learning_rate": 1.776300755025939e-05,
      "loss": 0.8653163909912109,
      "step": 1690
    },
    {
      "epoch": 0.8143020526094022,
      "grad_norm": 2.422149896621704,
      "learning_rate": 1.767435792976626e-05,
      "loss": 0.9010547995567322,
      "step": 1691
    },
    {
      "epoch": 0.8147836032023115,
      "grad_norm": 1.9990559816360474,
      "learning_rate": 1.7585908625370905e-05,
      "loss": 1.023693323135376,
      "step": 1692
    },
    {
      "epoch": 0.8152651537952206,
      "grad_norm": 1.2901694774627686,
      "learning_rate": 1.749765985228963e-05,
      "loss": 0.3572543263435364,
      "step": 1693
    },
    {
      "epoch": 0.8157467043881298,
      "grad_norm": 2.3446764945983887,
      "learning_rate": 1.740961182525077e-05,
      "loss": 0.41018784046173096,
      "step": 1694
    },
    {
      "epoch": 0.8162282549810389,
      "grad_norm": 1.5795892477035522,
      "learning_rate": 1.7321764758494252e-05,
      "loss": 1.1125493049621582,
      "step": 1695
    },
    {
      "epoch": 0.8167098055739481,
      "grad_norm": 1.3904321193695068,
      "learning_rate": 1.7234118865770987e-05,
      "loss": 0.94490647315979,
      "step": 1696
    },
    {
      "epoch": 0.8171913561668572,
      "grad_norm": 2.2052767276763916,
      "learning_rate": 1.7146674360342373e-05,
      "loss": 1.1280611753463745,
      "step": 1697
    },
    {
      "epoch": 0.8176729067597664,
      "grad_norm": 1.408920407295227,
      "learning_rate": 1.7059431454979824e-05,
      "loss": 1.5280137062072754,
      "step": 1698
    },
    {
      "epoch": 0.8181544573526757,
      "grad_norm": 2.0590484142303467,
      "learning_rate": 1.6972390361964195e-05,
      "loss": 0.9288832545280457,
      "step": 1699
    },
    {
      "epoch": 0.8186360079455848,
      "grad_norm": 19.42177963256836,
      "learning_rate": 1.688555129308531e-05,
      "loss": 1.2027837038040161,
      "step": 1700
    },
    {
      "epoch": 0.819117558538494,
      "grad_norm": 1.4646050930023193,
      "learning_rate": 1.6798914459641434e-05,
      "loss": 0.5764758586883545,
      "step": 1701
    },
    {
      "epoch": 0.8195991091314031,
      "grad_norm": 0.9640679955482483,
      "learning_rate": 1.6712480072438662e-05,
      "loss": 0.7724522948265076,
      "step": 1702
    },
    {
      "epoch": 0.8200806597243123,
      "grad_norm": 1.5659940242767334,
      "learning_rate": 1.6626248341790596e-05,
      "loss": 0.8127111196517944,
      "step": 1703
    },
    {
      "epoch": 0.8205622103172214,
      "grad_norm": 2.3534438610076904,
      "learning_rate": 1.6540219477517684e-05,
      "loss": 1.1205682754516602,
      "step": 1704
    },
    {
      "epoch": 0.8210437609101306,
      "grad_norm": 2.3394439220428467,
      "learning_rate": 1.6454393688946767e-05,
      "loss": 0.7127081155776978,
      "step": 1705
    },
    {
      "epoch": 0.8215253115030398,
      "grad_norm": 2.2475192546844482,
      "learning_rate": 1.6368771184910557e-05,
      "loss": 0.9197447896003723,
      "step": 1706
    },
    {
      "epoch": 0.822006862095949,
      "grad_norm": 3.1100456714630127,
      "learning_rate": 1.6283352173747145e-05,
      "loss": 1.0595561265945435,
      "step": 1707
    },
    {
      "epoch": 0.8224884126888581,
      "grad_norm": 2.4789206981658936,
      "learning_rate": 1.619813686329946e-05,
      "loss": 1.2160650491714478,
      "step": 1708
    },
    {
      "epoch": 0.8229699632817673,
      "grad_norm": 2.9909067153930664,
      "learning_rate": 1.611312546091476e-05,
      "loss": 1.1965662240982056,
      "step": 1709
    },
    {
      "epoch": 0.8234515138746764,
      "grad_norm": 1.8327099084854126,
      "learning_rate": 1.6028318173444202e-05,
      "loss": 0.8763675689697266,
      "step": 1710
    },
    {
      "epoch": 0.8239330644675856,
      "grad_norm": 1.6109602451324463,
      "learning_rate": 1.594371520724226e-05,
      "loss": 0.6016088128089905,
      "step": 1711
    },
    {
      "epoch": 0.8244146150604948,
      "grad_norm": 4.269639492034912,
      "learning_rate": 1.5859316768166244e-05,
      "loss": 0.6507865190505981,
      "step": 1712
    },
    {
      "epoch": 0.824896165653404,
      "grad_norm": 2.422987699508667,
      "learning_rate": 1.5775123061575836e-05,
      "loss": 1.5406382083892822,
      "step": 1713
    },
    {
      "epoch": 0.8253777162463132,
      "grad_norm": 1.0873609781265259,
      "learning_rate": 1.569113429233252e-05,
      "loss": 0.5942772030830383,
      "step": 1714
    },
    {
      "epoch": 0.8258592668392223,
      "grad_norm": 1.4857779741287231,
      "learning_rate": 1.5607350664799157e-05,
      "loss": 0.8704441785812378,
      "step": 1715
    },
    {
      "epoch": 0.8263408174321315,
      "grad_norm": 1.7813067436218262,
      "learning_rate": 1.552377238283943e-05,
      "loss": 0.905272901058197,
      "step": 1716
    },
    {
      "epoch": 0.8268223680250406,
      "grad_norm": 1.9900977611541748,
      "learning_rate": 1.5440399649817385e-05,
      "loss": 0.7601683139801025,
      "step": 1717
    },
    {
      "epoch": 0.8273039186179498,
      "grad_norm": 1.113451600074768,
      "learning_rate": 1.5357232668596933e-05,
      "loss": 0.9406479597091675,
      "step": 1718
    },
    {
      "epoch": 0.8277854692108589,
      "grad_norm": 2.3224246501922607,
      "learning_rate": 1.5274271641541295e-05,
      "loss": 0.5435302257537842,
      "step": 1719
    },
    {
      "epoch": 0.8282670198037682,
      "grad_norm": 3.29193115234375,
      "learning_rate": 1.5191516770512649e-05,
      "loss": 0.8528803586959839,
      "step": 1720
    },
    {
      "epoch": 0.8287485703966773,
      "grad_norm": 1.13429594039917,
      "learning_rate": 1.5108968256871437e-05,
      "loss": 0.6358800530433655,
      "step": 1721
    },
    {
      "epoch": 0.8292301209895865,
      "grad_norm": 1.9766297340393066,
      "learning_rate": 1.5026626301476087e-05,
      "loss": 0.7182720899581909,
      "step": 1722
    },
    {
      "epoch": 0.8297116715824956,
      "grad_norm": 2.5203065872192383,
      "learning_rate": 1.4944491104682379e-05,
      "loss": 0.36321550607681274,
      "step": 1723
    },
    {
      "epoch": 0.8301932221754048,
      "grad_norm": 2.697087287902832,
      "learning_rate": 1.4862562866343034e-05,
      "loss": 1.0022884607315063,
      "step": 1724
    },
    {
      "epoch": 0.8306747727683139,
      "grad_norm": 2.0794317722320557,
      "learning_rate": 1.4780841785807164e-05,
      "loss": 0.5318996906280518,
      "step": 1725
    },
    {
      "epoch": 0.8311563233612231,
      "grad_norm": 2.7635414600372314,
      "learning_rate": 1.4699328061919848e-05,
      "loss": 0.8253240585327148,
      "step": 1726
    },
    {
      "epoch": 0.8316378739541324,
      "grad_norm": 3.875403881072998,
      "learning_rate": 1.4618021893021605e-05,
      "loss": 0.5761056542396545,
      "step": 1727
    },
    {
      "epoch": 0.8321194245470415,
      "grad_norm": 1.8315447568893433,
      "learning_rate": 1.453692347694794e-05,
      "loss": 0.7333091497421265,
      "step": 1728
    },
    {
      "epoch": 0.8326009751399507,
      "grad_norm": 1.3995144367218018,
      "learning_rate": 1.4456033011028835e-05,
      "loss": 0.49542075395584106,
      "step": 1729
    },
    {
      "epoch": 0.8330825257328598,
      "grad_norm": 2.613591194152832,
      "learning_rate": 1.437535069208833e-05,
      "loss": 0.836378276348114,
      "step": 1730
    },
    {
      "epoch": 0.833564076325769,
      "grad_norm": 2.2486417293548584,
      "learning_rate": 1.4294876716443906e-05,
      "loss": 0.6846655011177063,
      "step": 1731
    },
    {
      "epoch": 0.8340456269186781,
      "grad_norm": 1.138690710067749,
      "learning_rate": 1.4214611279906187e-05,
      "loss": 0.31126007437705994,
      "step": 1732
    },
    {
      "epoch": 0.8345271775115873,
      "grad_norm": 1.8550935983657837,
      "learning_rate": 1.4134554577778337e-05,
      "loss": 0.981239378452301,
      "step": 1733
    },
    {
      "epoch": 0.8350087281044964,
      "grad_norm": 3.284966468811035,
      "learning_rate": 1.4054706804855634e-05,
      "loss": 0.6755277514457703,
      "step": 1734
    },
    {
      "epoch": 0.8354902786974057,
      "grad_norm": 2.5506527423858643,
      "learning_rate": 1.3975068155424976e-05,
      "loss": 0.716273844242096,
      "step": 1735
    },
    {
      "epoch": 0.8359718292903148,
      "grad_norm": 1.9901787042617798,
      "learning_rate": 1.3895638823264446e-05,
      "loss": 0.8489405512809753,
      "step": 1736
    },
    {
      "epoch": 0.836453379883224,
      "grad_norm": 5.520270347595215,
      "learning_rate": 1.3816419001642777e-05,
      "loss": 0.8489415049552917,
      "step": 1737
    },
    {
      "epoch": 0.8369349304761331,
      "grad_norm": 1.5447607040405273,
      "learning_rate": 1.3737408883318948e-05,
      "loss": 0.9109627604484558,
      "step": 1738
    },
    {
      "epoch": 0.8374164810690423,
      "grad_norm": 1.8914209604263306,
      "learning_rate": 1.365860866054165e-05,
      "loss": 0.720345139503479,
      "step": 1739
    },
    {
      "epoch": 0.8378980316619515,
      "grad_norm": 2.1682732105255127,
      "learning_rate": 1.358001852504891e-05,
      "loss": 0.7833454608917236,
      "step": 1740
    },
    {
      "epoch": 0.8383795822548606,
      "grad_norm": 3.2370963096618652,
      "learning_rate": 1.3501638668067485e-05,
      "loss": 1.0938481092453003,
      "step": 1741
    },
    {
      "epoch": 0.8388611328477699,
      "grad_norm": 6.025616645812988,
      "learning_rate": 1.3423469280312562e-05,
      "loss": 1.1536976099014282,
      "step": 1742
    },
    {
      "epoch": 0.839342683440679,
      "grad_norm": 2.0176846981048584,
      "learning_rate": 1.3345510551987128e-05,
      "loss": 0.3541467785835266,
      "step": 1743
    },
    {
      "epoch": 0.8398242340335882,
      "grad_norm": 5.713888645172119,
      "learning_rate": 1.326776267278167e-05,
      "loss": 0.4998454749584198,
      "step": 1744
    },
    {
      "epoch": 0.8403057846264973,
      "grad_norm": 1.0268326997756958,
      "learning_rate": 1.3190225831873581e-05,
      "loss": 0.6474172472953796,
      "step": 1745
    },
    {
      "epoch": 0.8407873352194065,
      "grad_norm": 1.2071791887283325,
      "learning_rate": 1.3112900217926782e-05,
      "loss": 0.7902032732963562,
      "step": 1746
    },
    {
      "epoch": 0.8412688858123156,
      "grad_norm": 1.2479091882705688,
      "learning_rate": 1.3035786019091223e-05,
      "loss": 0.7959186434745789,
      "step": 1747
    },
    {
      "epoch": 0.8417504364052248,
      "grad_norm": 1.2748886346817017,
      "learning_rate": 1.2958883423002422e-05,
      "loss": 0.8399845361709595,
      "step": 1748
    },
    {
      "epoch": 0.842231986998134,
      "grad_norm": 2.3503594398498535,
      "learning_rate": 1.288219261678103e-05,
      "loss": 1.2467331886291504,
      "step": 1749
    },
    {
      "epoch": 0.8427135375910432,
      "grad_norm": 3.5420756340026855,
      "learning_rate": 1.2805713787032381e-05,
      "loss": 0.9873737692832947,
      "step": 1750
    },
    {
      "epoch": 0.8431950881839523,
      "grad_norm": 2.0347323417663574,
      "learning_rate": 1.2729447119846016e-05,
      "loss": 0.4646882712841034,
      "step": 1751
    },
    {
      "epoch": 0.8436766387768615,
      "grad_norm": 1.6277610063552856,
      "learning_rate": 1.265339280079525e-05,
      "loss": 0.6238312721252441,
      "step": 1752
    },
    {
      "epoch": 0.8441581893697706,
      "grad_norm": 1.5678956508636475,
      "learning_rate": 1.257755101493665e-05,
      "loss": 0.9950130581855774,
      "step": 1753
    },
    {
      "epoch": 0.8446397399626798,
      "grad_norm": 1.9874587059020996,
      "learning_rate": 1.2501921946809714e-05,
      "loss": 0.7601380348205566,
      "step": 1754
    },
    {
      "epoch": 0.845121290555589,
      "grad_norm": 3.3925981521606445,
      "learning_rate": 1.2426505780436326e-05,
      "loss": 0.6927961111068726,
      "step": 1755
    },
    {
      "epoch": 0.8456028411484982,
      "grad_norm": 1.9708420038223267,
      "learning_rate": 1.2351302699320332e-05,
      "loss": 0.7333077192306519,
      "step": 1756
    },
    {
      "epoch": 0.8460843917414074,
      "grad_norm": 2.6311519145965576,
      "learning_rate": 1.2276312886447106e-05,
      "loss": 0.5576064586639404,
      "step": 1757
    },
    {
      "epoch": 0.8465659423343165,
      "grad_norm": 1.8699443340301514,
      "learning_rate": 1.2201536524283074e-05,
      "loss": 0.5901805758476257,
      "step": 1758
    },
    {
      "epoch": 0.8470474929272257,
      "grad_norm": 4.415778160095215,
      "learning_rate": 1.2126973794775343e-05,
      "loss": 0.9481086730957031,
      "step": 1759
    },
    {
      "epoch": 0.8475290435201348,
      "grad_norm": 3.505995988845825,
      "learning_rate": 1.2052624879351104e-05,
      "loss": 0.9161257147789001,
      "step": 1760
    },
    {
      "epoch": 0.848010594113044,
      "grad_norm": 2.775193214416504,
      "learning_rate": 1.1978489958917382e-05,
      "loss": 0.9492869973182678,
      "step": 1761
    },
    {
      "epoch": 0.8484921447059531,
      "grad_norm": 2.8440182209014893,
      "learning_rate": 1.1904569213860472e-05,
      "loss": 0.8795627355575562,
      "step": 1762
    },
    {
      "epoch": 0.8489736952988624,
      "grad_norm": 2.1821465492248535,
      "learning_rate": 1.1830862824045552e-05,
      "loss": 0.9482166767120361,
      "step": 1763
    },
    {
      "epoch": 0.8494552458917715,
      "grad_norm": 2.951324462890625,
      "learning_rate": 1.1757370968816217e-05,
      "loss": 0.3014884293079376,
      "step": 1764
    },
    {
      "epoch": 0.8499367964846807,
      "grad_norm": 1.9869091510772705,
      "learning_rate": 1.1684093826994024e-05,
      "loss": 0.33685925602912903,
      "step": 1765
    },
    {
      "epoch": 0.8504183470775898,
      "grad_norm": 2.138092279434204,
      "learning_rate": 1.1611031576878117e-05,
      "loss": 0.5400164723396301,
      "step": 1766
    },
    {
      "epoch": 0.850899897670499,
      "grad_norm": 1.6487535238265991,
      "learning_rate": 1.1538184396244778e-05,
      "loss": 0.46052929759025574,
      "step": 1767
    },
    {
      "epoch": 0.8513814482634082,
      "grad_norm": 2.558133602142334,
      "learning_rate": 1.146555246234694e-05,
      "loss": 0.7569980621337891,
      "step": 1768
    },
    {
      "epoch": 0.8518629988563173,
      "grad_norm": 1.7085074186325073,
      "learning_rate": 1.1393135951913824e-05,
      "loss": 0.5074298977851868,
      "step": 1769
    },
    {
      "epoch": 0.8523445494492266,
      "grad_norm": 1.5966845750808716,
      "learning_rate": 1.132093504115046e-05,
      "loss": 0.7343792915344238,
      "step": 1770
    },
    {
      "epoch": 0.8528261000421357,
      "grad_norm": 2.096651077270508,
      "learning_rate": 1.1248949905737283e-05,
      "loss": 0.6334657669067383,
      "step": 1771
    },
    {
      "epoch": 0.8533076506350449,
      "grad_norm": 3.897386312484741,
      "learning_rate": 1.1177180720829694e-05,
      "loss": 0.7838603854179382,
      "step": 1772
    },
    {
      "epoch": 0.853789201227954,
      "grad_norm": 3.6394762992858887,
      "learning_rate": 1.1105627661057671e-05,
      "loss": 0.6444329619407654,
      "step": 1773
    },
    {
      "epoch": 0.8542707518208632,
      "grad_norm": 4.107766628265381,
      "learning_rate": 1.103429090052528e-05,
      "loss": 1.1960633993148804,
      "step": 1774
    },
    {
      "epoch": 0.8547523024137723,
      "grad_norm": 2.154711961746216,
      "learning_rate": 1.096317061281027e-05,
      "loss": 0.6288588047027588,
      "step": 1775
    },
    {
      "epoch": 0.8552338530066815,
      "grad_norm": 2.2184550762176514,
      "learning_rate": 1.0892266970963704e-05,
      "loss": 1.009575366973877,
      "step": 1776
    },
    {
      "epoch": 0.8557154035995906,
      "grad_norm": 1.5103213787078857,
      "learning_rate": 1.082158014750948e-05,
      "loss": 0.5736418962478638,
      "step": 1777
    },
    {
      "epoch": 0.8561969541924999,
      "grad_norm": 1.3792436122894287,
      "learning_rate": 1.0751110314443958e-05,
      "loss": 0.9216289520263672,
      "step": 1778
    },
    {
      "epoch": 0.856678504785409,
      "grad_norm": 1.840848684310913,
      "learning_rate": 1.0680857643235431e-05,
      "loss": 0.9239728450775146,
      "step": 1779
    },
    {
      "epoch": 0.8571600553783182,
      "grad_norm": 3.2897086143493652,
      "learning_rate": 1.0610822304823887e-05,
      "loss": 1.194601058959961,
      "step": 1780
    },
    {
      "epoch": 0.8576416059712274,
      "grad_norm": 1.8633419275283813,
      "learning_rate": 1.0541004469620452e-05,
      "loss": 0.8140504360198975,
      "step": 1781
    },
    {
      "epoch": 0.8581231565641365,
      "grad_norm": 1.3140840530395508,
      "learning_rate": 1.0471404307507016e-05,
      "loss": 1.0204516649246216,
      "step": 1782
    },
    {
      "epoch": 0.8586047071570457,
      "grad_norm": 2.459521532058716,
      "learning_rate": 1.0402021987835831e-05,
      "loss": 0.45632699131965637,
      "step": 1783
    },
    {
      "epoch": 0.8590862577499548,
      "grad_norm": 3.6389317512512207,
      "learning_rate": 1.0332857679429098e-05,
      "loss": 1.2126121520996094,
      "step": 1784
    },
    {
      "epoch": 0.8595678083428641,
      "grad_norm": 1.8589650392532349,
      "learning_rate": 1.0263911550578531e-05,
      "loss": 0.32376155257225037,
      "step": 1785
    },
    {
      "epoch": 0.8600493589357732,
      "grad_norm": 2.837494134902954,
      "learning_rate": 1.0195183769045013e-05,
      "loss": 0.6162177324295044,
      "step": 1786
    },
    {
      "epoch": 0.8605309095286824,
      "grad_norm": 3.5470738410949707,
      "learning_rate": 1.0126674502058054e-05,
      "loss": 1.1724510192871094,
      "step": 1787
    },
    {
      "epoch": 0.8610124601215915,
      "grad_norm": 2.9405713081359863,
      "learning_rate": 1.005838391631555e-05,
      "loss": 1.1177942752838135,
      "step": 1788
    },
    {
      "epoch": 0.8614940107145007,
      "grad_norm": 1.1091272830963135,
      "learning_rate": 9.990312177983263e-06,
      "loss": 0.8158119916915894,
      "step": 1789
    },
    {
      "epoch": 0.8619755613074098,
      "grad_norm": 1.494266390800476,
      "learning_rate": 9.922459452694466e-06,
      "loss": 1.0165987014770508,
      "step": 1790
    },
    {
      "epoch": 0.862457111900319,
      "grad_norm": 2.0390355587005615,
      "learning_rate": 9.854825905549503e-06,
      "loss": 0.603755533695221,
      "step": 1791
    },
    {
      "epoch": 0.8629386624932281,
      "grad_norm": 1.519302248954773,
      "learning_rate": 9.787411701115456e-06,
      "loss": 0.7159218192100525,
      "step": 1792
    },
    {
      "epoch": 0.8634202130861374,
      "grad_norm": 2.14306640625,
      "learning_rate": 9.720217003425647e-06,
      "loss": 0.9232797622680664,
      "step": 1793
    },
    {
      "epoch": 0.8639017636790465,
      "grad_norm": 2.8411998748779297,
      "learning_rate": 9.65324197597931e-06,
      "loss": 0.6438859105110168,
      "step": 1794
    },
    {
      "epoch": 0.8643833142719557,
      "grad_norm": 2.4847517013549805,
      "learning_rate": 9.58648678174121e-06,
      "loss": 0.6084542274475098,
      "step": 1795
    },
    {
      "epoch": 0.8648648648648649,
      "grad_norm": 6.788503646850586,
      "learning_rate": 9.51995158314113e-06,
      "loss": 0.8708192110061646,
      "step": 1796
    },
    {
      "epoch": 0.865346415457774,
      "grad_norm": 1.060307502746582,
      "learning_rate": 9.45363654207363e-06,
      "loss": 0.6474368572235107,
      "step": 1797
    },
    {
      "epoch": 0.8658279660506832,
      "grad_norm": 2.5565807819366455,
      "learning_rate": 9.387541819897549e-06,
      "loss": 0.6963127255439758,
      "step": 1798
    },
    {
      "epoch": 0.8663095166435923,
      "grad_norm": 2.2074174880981445,
      "learning_rate": 9.321667577435634e-06,
      "loss": 0.5291393399238586,
      "step": 1799
    },
    {
      "epoch": 0.8667910672365016,
      "grad_norm": 4.0830888748168945,
      "learning_rate": 9.256013974974175e-06,
      "loss": 0.6379281282424927,
      "step": 1800
    },
    {
      "epoch": 0.8672726178294107,
      "grad_norm": 2.9662184715270996,
      "learning_rate": 9.19058117226258e-06,
      "loss": 0.740398645401001,
      "step": 1801
    },
    {
      "epoch": 0.8677541684223199,
      "grad_norm": 1.3723981380462646,
      "learning_rate": 9.125369328513034e-06,
      "loss": 0.6717973351478577,
      "step": 1802
    },
    {
      "epoch": 0.868235719015229,
      "grad_norm": 1.6995066404342651,
      "learning_rate": 9.060378602400054e-06,
      "loss": 0.4252944588661194,
      "step": 1803
    },
    {
      "epoch": 0.8687172696081382,
      "grad_norm": 2.00872802734375,
      "learning_rate": 8.995609152060136e-06,
      "loss": 0.767527163028717,
      "step": 1804
    },
    {
      "epoch": 0.8691988202010473,
      "grad_norm": 2.280327320098877,
      "learning_rate": 8.931061135091357e-06,
      "loss": 1.1668963432312012,
      "step": 1805
    },
    {
      "epoch": 0.8696803707939565,
      "grad_norm": 1.8072197437286377,
      "learning_rate": 8.866734708553015e-06,
      "loss": 1.0749906301498413,
      "step": 1806
    },
    {
      "epoch": 0.8701619213868657,
      "grad_norm": 3.3288466930389404,
      "learning_rate": 8.802630028965242e-06,
      "loss": 0.5301840901374817,
      "step": 1807
    },
    {
      "epoch": 0.8706434719797749,
      "grad_norm": 1.5904195308685303,
      "learning_rate": 8.738747252308555e-06,
      "loss": 0.3822922706604004,
      "step": 1808
    },
    {
      "epoch": 0.8711250225726841,
      "grad_norm": 2.9779927730560303,
      "learning_rate": 8.675086534023591e-06,
      "loss": 0.8622705936431885,
      "step": 1809
    },
    {
      "epoch": 0.8716065731655932,
      "grad_norm": 10.029919624328613,
      "learning_rate": 8.611648029010643e-06,
      "loss": 0.5127744078636169,
      "step": 1810
    },
    {
      "epoch": 0.8720881237585024,
      "grad_norm": 2.638657808303833,
      "learning_rate": 8.548431891629316e-06,
      "loss": 0.7983468174934387,
      "step": 1811
    },
    {
      "epoch": 0.8725696743514115,
      "grad_norm": 2.734145402908325,
      "learning_rate": 8.485438275698154e-06,
      "loss": 0.5578221678733826,
      "step": 1812
    },
    {
      "epoch": 0.8730512249443207,
      "grad_norm": 2.210209846496582,
      "learning_rate": 8.422667334494249e-06,
      "loss": 0.6434561610221863,
      "step": 1813
    },
    {
      "epoch": 0.8735327755372299,
      "grad_norm": 2.7448275089263916,
      "learning_rate": 8.360119220752893e-06,
      "loss": 0.6519855856895447,
      "step": 1814
    },
    {
      "epoch": 0.8740143261301391,
      "grad_norm": 3.012415885925293,
      "learning_rate": 8.297794086667165e-06,
      "loss": 0.735754132270813,
      "step": 1815
    },
    {
      "epoch": 0.8744958767230482,
      "grad_norm": 4.281802654266357,
      "learning_rate": 8.235692083887613e-06,
      "loss": 0.6815641522407532,
      "step": 1816
    },
    {
      "epoch": 0.8749774273159574,
      "grad_norm": 1.540742039680481,
      "learning_rate": 8.173813363521843e-06,
      "loss": 0.881055474281311,
      "step": 1817
    },
    {
      "epoch": 0.8754589779088665,
      "grad_norm": 2.7494571208953857,
      "learning_rate": 8.112158076134157e-06,
      "loss": 1.2117091417312622,
      "step": 1818
    },
    {
      "epoch": 0.8759405285017757,
      "grad_norm": 1.928873062133789,
      "learning_rate": 8.05072637174522e-06,
      "loss": 0.39002010226249695,
      "step": 1819
    },
    {
      "epoch": 0.8764220790946848,
      "grad_norm": 1.5758920907974243,
      "learning_rate": 7.989518399831641e-06,
      "loss": 1.0210663080215454,
      "step": 1820
    },
    {
      "epoch": 0.876903629687594,
      "grad_norm": 1.3454341888427734,
      "learning_rate": 7.928534309325675e-06,
      "loss": 0.9133199453353882,
      "step": 1821
    },
    {
      "epoch": 0.8773851802805033,
      "grad_norm": 4.219327449798584,
      "learning_rate": 7.8677742486148e-06,
      "loss": 1.0876537561416626,
      "step": 1822
    },
    {
      "epoch": 0.8778667308734124,
      "grad_norm": 1.5010168552398682,
      "learning_rate": 7.807238365541391e-06,
      "loss": 0.6519786715507507,
      "step": 1823
    },
    {
      "epoch": 0.8783482814663216,
      "grad_norm": 1.7740635871887207,
      "learning_rate": 7.746926807402344e-06,
      "loss": 0.855210542678833,
      "step": 1824
    },
    {
      "epoch": 0.8788298320592307,
      "grad_norm": 1.7274343967437744,
      "learning_rate": 7.686839720948736e-06,
      "loss": 0.9496231079101562,
      "step": 1825
    },
    {
      "epoch": 0.8793113826521399,
      "grad_norm": 1.9287621974945068,
      "learning_rate": 7.6269772523854365e-06,
      "loss": 0.6469759941101074,
      "step": 1826
    },
    {
      "epoch": 0.879792933245049,
      "grad_norm": 1.6034387350082397,
      "learning_rate": 7.567339547370789e-06,
      "loss": 0.518660843372345,
      "step": 1827
    },
    {
      "epoch": 0.8802744838379583,
      "grad_norm": 1.885786771774292,
      "learning_rate": 7.507926751016248e-06,
      "loss": 0.9674926400184631,
      "step": 1828
    },
    {
      "epoch": 0.8807560344308674,
      "grad_norm": 2.0882320404052734,
      "learning_rate": 7.4487390078859855e-06,
      "loss": 0.839196503162384,
      "step": 1829
    },
    {
      "epoch": 0.8812375850237766,
      "grad_norm": 2.8181467056274414,
      "learning_rate": 7.389776461996578e-06,
      "loss": 0.7034145593643188,
      "step": 1830
    },
    {
      "epoch": 0.8817191356166857,
      "grad_norm": 1.4634780883789062,
      "learning_rate": 7.331039256816663e-06,
      "loss": 0.8236523270606995,
      "step": 1831
    },
    {
      "epoch": 0.8822006862095949,
      "grad_norm": 1.9718531370162964,
      "learning_rate": 7.27252753526656e-06,
      "loss": 0.8954199552536011,
      "step": 1832
    },
    {
      "epoch": 0.882682236802504,
      "grad_norm": 3.329256534576416,
      "learning_rate": 7.214241439717962e-06,
      "loss": 0.6557578444480896,
      "step": 1833
    },
    {
      "epoch": 0.8831637873954132,
      "grad_norm": 1.5268913507461548,
      "learning_rate": 7.1561811119935425e-06,
      "loss": 0.43837663531303406,
      "step": 1834
    },
    {
      "epoch": 0.8836453379883223,
      "grad_norm": 2.164134979248047,
      "learning_rate": 7.098346693366642e-06,
      "loss": 0.7235320210456848,
      "step": 1835
    },
    {
      "epoch": 0.8841268885812316,
      "grad_norm": 1.822515845298767,
      "learning_rate": 7.0407383245609136e-06,
      "loss": 0.8851585984230042,
      "step": 1836
    },
    {
      "epoch": 0.8846084391741408,
      "grad_norm": 1.9905312061309814,
      "learning_rate": 6.983356145749975e-06,
      "loss": 0.6312957406044006,
      "step": 1837
    },
    {
      "epoch": 0.8850899897670499,
      "grad_norm": 2.0450570583343506,
      "learning_rate": 6.9262002965570835e-06,
      "loss": 0.6580808758735657,
      "step": 1838
    },
    {
      "epoch": 0.8855715403599591,
      "grad_norm": 1.2863441705703735,
      "learning_rate": 6.869270916054782e-06,
      "loss": 0.188202366232872,
      "step": 1839
    },
    {
      "epoch": 0.8860530909528682,
      "grad_norm": 4.214168071746826,
      "learning_rate": 6.812568142764575e-06,
      "loss": 0.9264922142028809,
      "step": 1840
    },
    {
      "epoch": 0.8865346415457774,
      "grad_norm": 3.0248055458068848,
      "learning_rate": 6.756092114656587e-06,
      "loss": 0.8124273419380188,
      "step": 1841
    },
    {
      "epoch": 0.8870161921386865,
      "grad_norm": 2.9074318408966064,
      "learning_rate": 6.699842969149195e-06,
      "loss": 0.6676780581474304,
      "step": 1842
    },
    {
      "epoch": 0.8874977427315958,
      "grad_norm": 1.5832053422927856,
      "learning_rate": 6.64382084310875e-06,
      "loss": 0.4962480068206787,
      "step": 1843
    },
    {
      "epoch": 0.8879792933245049,
      "grad_norm": 6.354259014129639,
      "learning_rate": 6.5880258728491905e-06,
      "loss": 0.5115213990211487,
      "step": 1844
    },
    {
      "epoch": 0.8884608439174141,
      "grad_norm": 1.8550896644592285,
      "learning_rate": 6.532458194131763e-06,
      "loss": 0.6252191066741943,
      "step": 1845
    },
    {
      "epoch": 0.8889423945103232,
      "grad_norm": 5.327465534210205,
      "learning_rate": 6.477117942164657e-06,
      "loss": 0.8996787071228027,
      "step": 1846
    },
    {
      "epoch": 0.8894239451032324,
      "grad_norm": 1.7860552072525024,
      "learning_rate": 6.422005251602658e-06,
      "loss": 0.5533450245857239,
      "step": 1847
    },
    {
      "epoch": 0.8899054956961415,
      "grad_norm": 1.8348726034164429,
      "learning_rate": 6.367120256546888e-06,
      "loss": 0.4685377776622772,
      "step": 1848
    },
    {
      "epoch": 0.8903870462890507,
      "grad_norm": 1.460806965827942,
      "learning_rate": 6.312463090544396e-06,
      "loss": 0.6288714408874512,
      "step": 1849
    },
    {
      "epoch": 0.89086859688196,
      "grad_norm": 2.359792470932007,
      "learning_rate": 6.258033886587911e-06,
      "loss": 1.1805469989776611,
      "step": 1850
    },
    {
      "epoch": 0.8913501474748691,
      "grad_norm": 0.8285127878189087,
      "learning_rate": 6.2038327771154485e-06,
      "loss": 0.45282289385795593,
      "step": 1851
    },
    {
      "epoch": 0.8918316980677783,
      "grad_norm": 4.482977867126465,
      "learning_rate": 6.1498598940100346e-06,
      "loss": 0.883598804473877,
      "step": 1852
    },
    {
      "epoch": 0.8923132486606874,
      "grad_norm": 2.294842481613159,
      "learning_rate": 6.0961153685993646e-06,
      "loss": 0.8555234670639038,
      "step": 1853
    },
    {
      "epoch": 0.8927947992535966,
      "grad_norm": 2.8804733753204346,
      "learning_rate": 6.0425993316554965e-06,
      "loss": 0.6862808465957642,
      "step": 1854
    },
    {
      "epoch": 0.8932763498465057,
      "grad_norm": 1.6861966848373413,
      "learning_rate": 5.989311913394546e-06,
      "loss": 0.5246079564094543,
      "step": 1855
    },
    {
      "epoch": 0.8937579004394149,
      "grad_norm": 2.2151832580566406,
      "learning_rate": 5.93625324347632e-06,
      "loss": 0.3396415412425995,
      "step": 1856
    },
    {
      "epoch": 0.894239451032324,
      "grad_norm": 1.164566159248352,
      "learning_rate": 5.8834234510040335e-06,
      "loss": 0.521573007106781,
      "step": 1857
    },
    {
      "epoch": 0.8947210016252333,
      "grad_norm": 5.0740180015563965,
      "learning_rate": 5.830822664523994e-06,
      "loss": 0.8860540986061096,
      "step": 1858
    },
    {
      "epoch": 0.8952025522181424,
      "grad_norm": 1.9234141111373901,
      "learning_rate": 5.77845101202531e-06,
      "loss": 0.7049664258956909,
      "step": 1859
    },
    {
      "epoch": 0.8956841028110516,
      "grad_norm": 1.4048093557357788,
      "learning_rate": 5.726308620939536e-06,
      "loss": 0.5329050421714783,
      "step": 1860
    },
    {
      "epoch": 0.8961656534039607,
      "grad_norm": 2.861079216003418,
      "learning_rate": 5.674395618140393e-06,
      "loss": 0.5589011907577515,
      "step": 1861
    },
    {
      "epoch": 0.8966472039968699,
      "grad_norm": 1.5645968914031982,
      "learning_rate": 5.622712129943453e-06,
      "loss": 0.6571087837219238,
      "step": 1862
    },
    {
      "epoch": 0.8971287545897791,
      "grad_norm": 2.1817469596862793,
      "learning_rate": 5.571258282105829e-06,
      "loss": 0.6116480231285095,
      "step": 1863
    },
    {
      "epoch": 0.8976103051826883,
      "grad_norm": 1.0763663053512573,
      "learning_rate": 5.520034199825841e-06,
      "loss": 1.1373647451400757,
      "step": 1864
    },
    {
      "epoch": 0.8980918557755975,
      "grad_norm": 1.6418788433074951,
      "learning_rate": 5.469040007742776e-06,
      "loss": 0.5137871503829956,
      "step": 1865
    },
    {
      "epoch": 0.8985734063685066,
      "grad_norm": 2.5534069538116455,
      "learning_rate": 5.418275829936537e-06,
      "loss": 1.2674925327301025,
      "step": 1866
    },
    {
      "epoch": 0.8990549569614158,
      "grad_norm": 1.7408275604248047,
      "learning_rate": 5.36774178992735e-06,
      "loss": 0.6076784729957581,
      "step": 1867
    },
    {
      "epoch": 0.8995365075543249,
      "grad_norm": 2.683398962020874,
      "learning_rate": 5.317438010675469e-06,
      "loss": 0.7806070446968079,
      "step": 1868
    },
    {
      "epoch": 0.9000180581472341,
      "grad_norm": 2.030315637588501,
      "learning_rate": 5.267364614580861e-06,
      "loss": 0.8753706216812134,
      "step": 1869
    },
    {
      "epoch": 0.9004996087401432,
      "grad_norm": 1.5364534854888916,
      "learning_rate": 5.217521723482943e-06,
      "loss": 0.5193024277687073,
      "step": 1870
    },
    {
      "epoch": 0.9009811593330525,
      "grad_norm": 2.9137372970581055,
      "learning_rate": 5.167909458660258e-06,
      "loss": 0.598779022693634,
      "step": 1871
    },
    {
      "epoch": 0.9014627099259616,
      "grad_norm": 1.8630374670028687,
      "learning_rate": 5.118527940830165e-06,
      "loss": 0.8497840762138367,
      "step": 1872
    },
    {
      "epoch": 0.9019442605188708,
      "grad_norm": 2.4683010578155518,
      "learning_rate": 5.069377290148602e-06,
      "loss": 0.5051344037055969,
      "step": 1873
    },
    {
      "epoch": 0.9024258111117799,
      "grad_norm": 1.7138243913650513,
      "learning_rate": 5.020457626209707e-06,
      "loss": 0.4720628559589386,
      "step": 1874
    },
    {
      "epoch": 0.9029073617046891,
      "grad_norm": 2.553943395614624,
      "learning_rate": 4.971769068045628e-06,
      "loss": 0.6861895322799683,
      "step": 1875
    },
    {
      "epoch": 0.9033889122975982,
      "grad_norm": 1.2986865043640137,
      "learning_rate": 4.923311734126135e-06,
      "loss": 0.7743862867355347,
      "step": 1876
    },
    {
      "epoch": 0.9038704628905074,
      "grad_norm": 2.346980333328247,
      "learning_rate": 4.875085742358432e-06,
      "loss": 1.2708526849746704,
      "step": 1877
    },
    {
      "epoch": 0.9043520134834167,
      "grad_norm": 2.2848098278045654,
      "learning_rate": 4.827091210086776e-06,
      "loss": 0.7205824255943298,
      "step": 1878
    },
    {
      "epoch": 0.9048335640763258,
      "grad_norm": 1.9337440729141235,
      "learning_rate": 4.779328254092252e-06,
      "loss": 1.1107653379440308,
      "step": 1879
    },
    {
      "epoch": 0.905315114669235,
      "grad_norm": 4.126399517059326,
      "learning_rate": 4.731796990592452e-06,
      "loss": 0.5711522102355957,
      "step": 1880
    },
    {
      "epoch": 0.9057966652621441,
      "grad_norm": 1.1134138107299805,
      "learning_rate": 4.68449753524125e-06,
      "loss": 0.8676082491874695,
      "step": 1881
    },
    {
      "epoch": 0.9062782158550533,
      "grad_norm": 2.09855318069458,
      "learning_rate": 4.637430003128429e-06,
      "loss": 0.7797391414642334,
      "step": 1882
    },
    {
      "epoch": 0.9067597664479624,
      "grad_norm": 2.563356876373291,
      "learning_rate": 4.5905945087794996e-06,
      "loss": 0.763664722442627,
      "step": 1883
    },
    {
      "epoch": 0.9072413170408716,
      "grad_norm": 1.7145023345947266,
      "learning_rate": 4.543991166155337e-06,
      "loss": 0.9080654978752136,
      "step": 1884
    },
    {
      "epoch": 0.9077228676337807,
      "grad_norm": 3.032350778579712,
      "learning_rate": 4.497620088651966e-06,
      "loss": 0.8827487826347351,
      "step": 1885
    },
    {
      "epoch": 0.90820441822669,
      "grad_norm": 2.354414939880371,
      "learning_rate": 4.451481389100232e-06,
      "loss": 0.8509942889213562,
      "step": 1886
    },
    {
      "epoch": 0.9086859688195991,
      "grad_norm": 3.49049711227417,
      "learning_rate": 4.405575179765586e-06,
      "loss": 0.835801362991333,
      "step": 1887
    },
    {
      "epoch": 0.9091675194125083,
      "grad_norm": 1.9126769304275513,
      "learning_rate": 4.359901572347758e-06,
      "loss": 0.6399697065353394,
      "step": 1888
    },
    {
      "epoch": 0.9096490700054174,
      "grad_norm": 0.941493570804596,
      "learning_rate": 4.314460677980537e-06,
      "loss": 0.3583628535270691,
      "step": 1889
    },
    {
      "epoch": 0.9101306205983266,
      "grad_norm": 1.6237558126449585,
      "learning_rate": 4.269252607231422e-06,
      "loss": 0.5488873720169067,
      "step": 1890
    },
    {
      "epoch": 0.9106121711912358,
      "grad_norm": 1.799116611480713,
      "learning_rate": 4.224277470101445e-06,
      "loss": 1.1100224256515503,
      "step": 1891
    },
    {
      "epoch": 0.9110937217841449,
      "grad_norm": 5.039967060089111,
      "learning_rate": 4.179535376024857e-06,
      "loss": 1.0829538106918335,
      "step": 1892
    },
    {
      "epoch": 0.9115752723770542,
      "grad_norm": 2.527869462966919,
      "learning_rate": 4.135026433868827e-06,
      "loss": 0.9230284690856934,
      "step": 1893
    },
    {
      "epoch": 0.9120568229699633,
      "grad_norm": 1.2518911361694336,
      "learning_rate": 4.090750751933248e-06,
      "loss": 0.887606143951416,
      "step": 1894
    },
    {
      "epoch": 0.9125383735628725,
      "grad_norm": 1.0996094942092896,
      "learning_rate": 4.046708437950464e-06,
      "loss": 0.5563896894454956,
      "step": 1895
    },
    {
      "epoch": 0.9130199241557816,
      "grad_norm": 2.3735291957855225,
      "learning_rate": 4.0028995990849084e-06,
      "loss": 0.4965125322341919,
      "step": 1896
    },
    {
      "epoch": 0.9135014747486908,
      "grad_norm": 1.6580549478530884,
      "learning_rate": 3.95932434193299e-06,
      "loss": 0.3492845892906189,
      "step": 1897
    },
    {
      "epoch": 0.9139830253415999,
      "grad_norm": 2.7380456924438477,
      "learning_rate": 3.915982772522719e-06,
      "loss": 0.6443988084793091,
      "step": 1898
    },
    {
      "epoch": 0.9144645759345091,
      "grad_norm": 2.208221435546875,
      "learning_rate": 3.872874996313513e-06,
      "loss": 0.5126693248748779,
      "step": 1899
    },
    {
      "epoch": 0.9149461265274182,
      "grad_norm": 2.2725961208343506,
      "learning_rate": 3.830001118195936e-06,
      "loss": 0.6966872811317444,
      "step": 1900
    },
    {
      "epoch": 0.9154276771203275,
      "grad_norm": 2.0959906578063965,
      "learning_rate": 3.787361242491394e-06,
      "loss": 0.5551361441612244,
      "step": 1901
    },
    {
      "epoch": 0.9159092277132366,
      "grad_norm": 3.0877785682678223,
      "learning_rate": 3.744955472951928e-06,
      "loss": 0.8979296684265137,
      "step": 1902
    },
    {
      "epoch": 0.9163907783061458,
      "grad_norm": 2.749868869781494,
      "learning_rate": 3.702783912759955e-06,
      "loss": 0.9359360337257385,
      "step": 1903
    },
    {
      "epoch": 0.916872328899055,
      "grad_norm": 1.5884201526641846,
      "learning_rate": 3.660846664528006e-06,
      "loss": 0.33591821789741516,
      "step": 1904
    },
    {
      "epoch": 0.9173538794919641,
      "grad_norm": 2.4273812770843506,
      "learning_rate": 3.6191438302984772e-06,
      "loss": 0.7867469191551208,
      "step": 1905
    },
    {
      "epoch": 0.9178354300848733,
      "grad_norm": 1.8173773288726807,
      "learning_rate": 3.577675511543388e-06,
      "loss": 0.5689818859100342,
      "step": 1906
    },
    {
      "epoch": 0.9183169806777824,
      "grad_norm": 3.2586236000061035,
      "learning_rate": 3.5364418091641373e-06,
      "loss": 0.8921031951904297,
      "step": 1907
    },
    {
      "epoch": 0.9187985312706917,
      "grad_norm": 3.260666608810425,
      "learning_rate": 3.495442823491224e-06,
      "loss": 0.7662656903266907,
      "step": 1908
    },
    {
      "epoch": 0.9192800818636008,
      "grad_norm": 1.342635989189148,
      "learning_rate": 3.4546786542840605e-06,
      "loss": 0.6159889698028564,
      "step": 1909
    },
    {
      "epoch": 0.91976163245651,
      "grad_norm": 4.522103786468506,
      "learning_rate": 3.4141494007306816e-06,
      "loss": 0.640151858329773,
      "step": 1910
    },
    {
      "epoch": 0.9202431830494191,
      "grad_norm": 1.2931874990463257,
      "learning_rate": 3.373855161447548e-06,
      "loss": 0.7301903367042542,
      "step": 1911
    },
    {
      "epoch": 0.9207247336423283,
      "grad_norm": 2.4328622817993164,
      "learning_rate": 3.333796034479242e-06,
      "loss": 0.841766893863678,
      "step": 1912
    },
    {
      "epoch": 0.9212062842352374,
      "grad_norm": 1.8867064714431763,
      "learning_rate": 3.293972117298294e-06,
      "loss": 0.7451145648956299,
      "step": 1913
    },
    {
      "epoch": 0.9216878348281466,
      "grad_norm": 2.7373173236846924,
      "learning_rate": 3.2543835068049255e-06,
      "loss": 0.5494900345802307,
      "step": 1914
    },
    {
      "epoch": 0.9221693854210558,
      "grad_norm": 1.9295862913131714,
      "learning_rate": 3.21503029932676e-06,
      "loss": 0.7260355949401855,
      "step": 1915
    },
    {
      "epoch": 0.922650936013965,
      "grad_norm": 6.219395160675049,
      "learning_rate": 3.1759125906186793e-06,
      "loss": 0.5387226939201355,
      "step": 1916
    },
    {
      "epoch": 0.9231324866068741,
      "grad_norm": 2.4589684009552,
      "learning_rate": 3.137030475862535e-06,
      "loss": 0.864848256111145,
      "step": 1917
    },
    {
      "epoch": 0.9236140371997833,
      "grad_norm": 3.0752179622650146,
      "learning_rate": 3.098384049666925e-06,
      "loss": 0.3438478708267212,
      "step": 1918
    },
    {
      "epoch": 0.9240955877926925,
      "grad_norm": 2.0400497913360596,
      "learning_rate": 3.059973406066963e-06,
      "loss": 0.6857901811599731,
      "step": 1919
    },
    {
      "epoch": 0.9245771383856016,
      "grad_norm": 1.6464850902557373,
      "learning_rate": 3.0217986385240537e-06,
      "loss": 0.4986359179019928,
      "step": 1920
    },
    {
      "epoch": 0.9250586889785108,
      "grad_norm": 1.5397393703460693,
      "learning_rate": 2.983859839925662e-06,
      "loss": 0.7675492763519287,
      "step": 1921
    },
    {
      "epoch": 0.92554023957142,
      "grad_norm": 2.370856285095215,
      "learning_rate": 2.94615710258509e-06,
      "loss": 0.8671760559082031,
      "step": 1922
    },
    {
      "epoch": 0.9260217901643292,
      "grad_norm": 1.815647006034851,
      "learning_rate": 2.908690518241275e-06,
      "loss": 0.6704913973808289,
      "step": 1923
    },
    {
      "epoch": 0.9265033407572383,
      "grad_norm": 1.424587368965149,
      "learning_rate": 2.8714601780584937e-06,
      "loss": 0.6523185968399048,
      "step": 1924
    },
    {
      "epoch": 0.9269848913501475,
      "grad_norm": 2.761028528213501,
      "learning_rate": 2.834466172626238e-06,
      "loss": 0.4928123354911804,
      "step": 1925
    },
    {
      "epoch": 0.9274664419430566,
      "grad_norm": 1.2203357219696045,
      "learning_rate": 2.7977085919589254e-06,
      "loss": 0.3803952932357788,
      "step": 1926
    },
    {
      "epoch": 0.9279479925359658,
      "grad_norm": 4.186097621917725,
      "learning_rate": 2.76118752549569e-06,
      "loss": 0.9912437200546265,
      "step": 1927
    },
    {
      "epoch": 0.9284295431288749,
      "grad_norm": 2.9175939559936523,
      "learning_rate": 2.7249030621001924e-06,
      "loss": 0.7338413000106812,
      "step": 1928
    },
    {
      "epoch": 0.9289110937217842,
      "grad_norm": 3.231977939605713,
      "learning_rate": 2.688855290060399e-06,
      "loss": 0.8454397916793823,
      "step": 1929
    },
    {
      "epoch": 0.9293926443146933,
      "grad_norm": 1.274575114250183,
      "learning_rate": 2.653044297088314e-06,
      "loss": 0.5701355934143066,
      "step": 1930
    },
    {
      "epoch": 0.9298741949076025,
      "grad_norm": 2.101989984512329,
      "learning_rate": 2.6174701703198468e-06,
      "loss": 0.5113231539726257,
      "step": 1931
    },
    {
      "epoch": 0.9303557455005117,
      "grad_norm": 3.5023107528686523,
      "learning_rate": 2.5821329963145347e-06,
      "loss": 0.8258587121963501,
      "step": 1932
    },
    {
      "epoch": 0.9308372960934208,
      "grad_norm": 4.5869035720825195,
      "learning_rate": 2.547032861055376e-06,
      "loss": 0.5059298872947693,
      "step": 1933
    },
    {
      "epoch": 0.93131884668633,
      "grad_norm": 2.850600242614746,
      "learning_rate": 2.5121698499485757e-06,
      "loss": 0.851980447769165,
      "step": 1934
    },
    {
      "epoch": 0.9318003972792391,
      "grad_norm": 2.8298370838165283,
      "learning_rate": 2.4775440478233993e-06,
      "loss": 0.8488339185714722,
      "step": 1935
    },
    {
      "epoch": 0.9322819478721484,
      "grad_norm": 1.7519915103912354,
      "learning_rate": 2.4431555389319074e-06,
      "loss": 0.6096922159194946,
      "step": 1936
    },
    {
      "epoch": 0.9327634984650575,
      "grad_norm": 1.2772984504699707,
      "learning_rate": 2.4090044069487784e-06,
      "loss": 0.5907295942306519,
      "step": 1937
    },
    {
      "epoch": 0.9332450490579667,
      "grad_norm": 1.3821014165878296,
      "learning_rate": 2.3750907349711084e-06,
      "loss": 0.6068422794342041,
      "step": 1938
    },
    {
      "epoch": 0.9337265996508758,
      "grad_norm": 1.3913278579711914,
      "learning_rate": 2.3414146055182106e-06,
      "loss": 0.8689660429954529,
      "step": 1939
    },
    {
      "epoch": 0.934208150243785,
      "grad_norm": 3.146981716156006,
      "learning_rate": 2.307976100531384e-06,
      "loss": 1.1123387813568115,
      "step": 1940
    },
    {
      "epoch": 0.9346897008366941,
      "grad_norm": 1.5993727445602417,
      "learning_rate": 2.274775301373744e-06,
      "loss": 0.7586812973022461,
      "step": 1941
    },
    {
      "epoch": 0.9351712514296033,
      "grad_norm": 2.7153189182281494,
      "learning_rate": 2.241812288830003e-06,
      "loss": 0.8807139992713928,
      "step": 1942
    },
    {
      "epoch": 0.9356528020225124,
      "grad_norm": 4.2512006759643555,
      "learning_rate": 2.2090871431063253e-06,
      "loss": 0.9558701515197754,
      "step": 1943
    },
    {
      "epoch": 0.9361343526154217,
      "grad_norm": 3.2264180183410645,
      "learning_rate": 2.176599943830071e-06,
      "loss": 1.0251234769821167,
      "step": 1944
    },
    {
      "epoch": 0.9366159032083309,
      "grad_norm": 1.9459080696105957,
      "learning_rate": 2.144350770049597e-06,
      "loss": 1.0112384557724,
      "step": 1945
    },
    {
      "epoch": 0.93709745380124,
      "grad_norm": 2.175313711166382,
      "learning_rate": 2.112339700234156e-06,
      "loss": 1.147891879081726,
      "step": 1946
    },
    {
      "epoch": 0.9375790043941492,
      "grad_norm": 2.3391759395599365,
      "learning_rate": 2.0805668122735767e-06,
      "loss": 1.0167261362075806,
      "step": 1947
    },
    {
      "epoch": 0.9380605549870583,
      "grad_norm": 3.4028477668762207,
      "learning_rate": 2.0490321834781833e-06,
      "loss": 0.832416296005249,
      "step": 1948
    },
    {
      "epoch": 0.9385421055799675,
      "grad_norm": 2.0738070011138916,
      "learning_rate": 2.0177358905785537e-06,
      "loss": 0.5863141417503357,
      "step": 1949
    },
    {
      "epoch": 0.9390236561728766,
      "grad_norm": 1.6731641292572021,
      "learning_rate": 1.986678009725329e-06,
      "loss": 1.0583198070526123,
      "step": 1950
    },
    {
      "epoch": 0.9395052067657859,
      "grad_norm": 1.1223913431167603,
      "learning_rate": 1.955858616489059e-06,
      "loss": 1.071386694908142,
      "step": 1951
    },
    {
      "epoch": 0.939986757358695,
      "grad_norm": 2.1744823455810547,
      "learning_rate": 1.9252777858599915e-06,
      "loss": 0.4888773560523987,
      "step": 1952
    },
    {
      "epoch": 0.9404683079516042,
      "grad_norm": 2.54960298538208,
      "learning_rate": 1.8949355922479151e-06,
      "loss": 0.8104636669158936,
      "step": 1953
    },
    {
      "epoch": 0.9409498585445133,
      "grad_norm": 2.3711905479431152,
      "learning_rate": 1.8648321094819287e-06,
      "loss": 0.7300409078598022,
      "step": 1954
    },
    {
      "epoch": 0.9414314091374225,
      "grad_norm": 1.4486585855484009,
      "learning_rate": 1.8349674108103288e-06,
      "loss": 0.9925588369369507,
      "step": 1955
    },
    {
      "epoch": 0.9419129597303316,
      "grad_norm": 2.337092638015747,
      "learning_rate": 1.8053415689003872e-06,
      "loss": 0.4967162311077118,
      "step": 1956
    },
    {
      "epoch": 0.9423945103232408,
      "grad_norm": 2.186246633529663,
      "learning_rate": 1.7759546558381967e-06,
      "loss": 0.5689997673034668,
      "step": 1957
    },
    {
      "epoch": 0.94287606091615,
      "grad_norm": 1.8951377868652344,
      "learning_rate": 1.7468067431284707e-06,
      "loss": 0.8406978249549866,
      "step": 1958
    },
    {
      "epoch": 0.9433576115090592,
      "grad_norm": 1.8798428773880005,
      "learning_rate": 1.7178979016943764e-06,
      "loss": 1.2161320447921753,
      "step": 1959
    },
    {
      "epoch": 0.9438391621019684,
      "grad_norm": 3.048610210418701,
      "learning_rate": 1.6892282018773908e-06,
      "loss": 0.8621540069580078,
      "step": 1960
    },
    {
      "epoch": 0.9443207126948775,
      "grad_norm": 4.253155708312988,
      "learning_rate": 1.6607977134370789e-06,
      "loss": 1.3255707025527954,
      "step": 1961
    },
    {
      "epoch": 0.9448022632877867,
      "grad_norm": 2.788092851638794,
      "learning_rate": 1.6326065055510043e-06,
      "loss": 0.9271857142448425,
      "step": 1962
    },
    {
      "epoch": 0.9452838138806958,
      "grad_norm": 4.708611965179443,
      "learning_rate": 1.6046546468144407e-06,
      "loss": 0.9531923532485962,
      "step": 1963
    },
    {
      "epoch": 0.945765364473605,
      "grad_norm": 2.289548635482788,
      "learning_rate": 1.576942205240317e-06,
      "loss": 0.8942689895629883,
      "step": 1964
    },
    {
      "epoch": 0.9462469150665141,
      "grad_norm": 1.728692889213562,
      "learning_rate": 1.5494692482590057e-06,
      "loss": 0.4066816568374634,
      "step": 1965
    },
    {
      "epoch": 0.9467284656594234,
      "grad_norm": 0.8854557871818542,
      "learning_rate": 1.522235842718156e-06,
      "loss": 0.34332430362701416,
      "step": 1966
    },
    {
      "epoch": 0.9472100162523325,
      "grad_norm": 2.4654204845428467,
      "learning_rate": 1.4952420548825285e-06,
      "loss": 1.0930558443069458,
      "step": 1967
    },
    {
      "epoch": 0.9476915668452417,
      "grad_norm": 1.3775367736816406,
      "learning_rate": 1.468487950433839e-06,
      "loss": 0.4075854420661926,
      "step": 1968
    },
    {
      "epoch": 0.9481731174381508,
      "grad_norm": 2.0057859420776367,
      "learning_rate": 1.441973594470636e-06,
      "loss": 0.6738939881324768,
      "step": 1969
    },
    {
      "epoch": 0.94865466803106,
      "grad_norm": 1.5300246477127075,
      "learning_rate": 1.415699051508068e-06,
      "loss": 0.43441304564476013,
      "step": 1970
    },
    {
      "epoch": 0.9491362186239691,
      "grad_norm": 3.4095311164855957,
      "learning_rate": 1.3896643854777847e-06,
      "loss": 1.0773264169692993,
      "step": 1971
    },
    {
      "epoch": 0.9496177692168783,
      "grad_norm": 4.122169494628906,
      "learning_rate": 1.3638696597277679e-06,
      "loss": 0.7260087132453918,
      "step": 1972
    },
    {
      "epoch": 0.9500993198097876,
      "grad_norm": 1.653339147567749,
      "learning_rate": 1.3383149370221449e-06,
      "loss": 0.5981525182723999,
      "step": 1973
    },
    {
      "epoch": 0.9505808704026967,
      "grad_norm": 1.117030382156372,
      "learning_rate": 1.313000279541121e-06,
      "loss": 0.41399234533309937,
      "step": 1974
    },
    {
      "epoch": 0.9510624209956059,
      "grad_norm": 2.5914013385772705,
      "learning_rate": 1.287925748880703e-06,
      "loss": 0.6134347915649414,
      "step": 1975
    },
    {
      "epoch": 0.951543971588515,
      "grad_norm": 3.830418348312378,
      "learning_rate": 1.2630914060526522e-06,
      "loss": 1.0437018871307373,
      "step": 1976
    },
    {
      "epoch": 0.9520255221814242,
      "grad_norm": 1.8955632448196411,
      "learning_rate": 1.2384973114843101e-06,
      "loss": 0.9949756860733032,
      "step": 1977
    },
    {
      "epoch": 0.9525070727743333,
      "grad_norm": 2.570497751235962,
      "learning_rate": 1.2141435250184185e-06,
      "loss": 0.4804862141609192,
      "step": 1978
    },
    {
      "epoch": 0.9529886233672425,
      "grad_norm": 2.0977566242218018,
      "learning_rate": 1.1900301059130093e-06,
      "loss": 0.6975605487823486,
      "step": 1979
    },
    {
      "epoch": 0.9534701739601517,
      "grad_norm": 1.847624659538269,
      "learning_rate": 1.1661571128412596e-06,
      "loss": 0.6540684700012207,
      "step": 1980
    },
    {
      "epoch": 0.9539517245530609,
      "grad_norm": 4.380690097808838,
      "learning_rate": 1.142524603891315e-06,
      "loss": 0.6058237552642822,
      "step": 1981
    },
    {
      "epoch": 0.95443327514597,
      "grad_norm": 1.211367130279541,
      "learning_rate": 1.1191326365661892e-06,
      "loss": 0.46046242117881775,
      "step": 1982
    },
    {
      "epoch": 0.9549148257388792,
      "grad_norm": 2.106376886367798,
      "learning_rate": 1.0959812677835968e-06,
      "loss": 0.7920570373535156,
      "step": 1983
    },
    {
      "epoch": 0.9553963763317883,
      "grad_norm": 1.3183211088180542,
      "learning_rate": 1.0730705538758322e-06,
      "loss": 0.6143832802772522,
      "step": 1984
    },
    {
      "epoch": 0.9558779269246975,
      "grad_norm": 1.6042373180389404,
      "learning_rate": 1.0504005505896141e-06,
      "loss": 0.5574995279312134,
      "step": 1985
    },
    {
      "epoch": 0.9563594775176067,
      "grad_norm": 2.6498067378997803,
      "learning_rate": 1.0279713130859514e-06,
      "loss": 0.7216347455978394,
      "step": 1986
    },
    {
      "epoch": 0.9568410281105159,
      "grad_norm": 2.2287399768829346,
      "learning_rate": 1.005782895940055e-06,
      "loss": 0.8709151148796082,
      "step": 1987
    },
    {
      "epoch": 0.9573225787034251,
      "grad_norm": 2.855205774307251,
      "learning_rate": 9.838353531411272e-07,
      "loss": 0.4653676450252533,
      "step": 1988
    },
    {
      "epoch": 0.9578041292963342,
      "grad_norm": 2.1802797317504883,
      "learning_rate": 9.62128738092294e-07,
      "loss": 0.5144358277320862,
      "step": 1989
    },
    {
      "epoch": 0.9582856798892434,
      "grad_norm": 1.9923255443572998,
      "learning_rate": 9.406631036104508e-07,
      "loss": 0.6633358001708984,
      "step": 1990
    },
    {
      "epoch": 0.9587672304821525,
      "grad_norm": 0.779755711555481,
      "learning_rate": 9.194385019261287e-07,
      "loss": 0.5907173156738281,
      "step": 1991
    },
    {
      "epoch": 0.9592487810750617,
      "grad_norm": 3.1367998123168945,
      "learning_rate": 8.984549846833612e-07,
      "loss": 0.9462472796440125,
      "step": 1992
    },
    {
      "epoch": 0.9597303316679708,
      "grad_norm": 1.8132468461990356,
      "learning_rate": 8.777126029396065e-07,
      "loss": 0.48685222864151,
      "step": 1993
    },
    {
      "epoch": 0.9602118822608801,
      "grad_norm": 2.1201539039611816,
      "learning_rate": 8.572114071655479e-07,
      "loss": 0.36985668540000916,
      "step": 1994
    },
    {
      "epoch": 0.9606934328537892,
      "grad_norm": 1.8523855209350586,
      "learning_rate": 8.369514472450379e-07,
      "loss": 0.5595872402191162,
      "step": 1995
    },
    {
      "epoch": 0.9611749834466984,
      "grad_norm": 1.3683743476867676,
      "learning_rate": 8.169327724749543e-07,
      "loss": 0.6902904510498047,
      "step": 1996
    },
    {
      "epoch": 0.9616565340396075,
      "grad_norm": 1.5133596658706665,
      "learning_rate": 7.971554315650442e-07,
      "loss": 0.7271482348442078,
      "step": 1997
    },
    {
      "epoch": 0.9621380846325167,
      "grad_norm": 3.1602718830108643,
      "learning_rate": 7.776194726378583e-07,
      "loss": 0.9070822596549988,
      "step": 1998
    },
    {
      "epoch": 0.9626196352254258,
      "grad_norm": 1.2385915517807007,
      "learning_rate": 7.583249432286277e-07,
      "loss": 0.6294569373130798,
      "step": 1999
    },
    {
      "epoch": 0.963101185818335,
      "grad_norm": 1.7936358451843262,
      "learning_rate": 7.392718902850981e-07,
      "loss": 0.7525774240493774,
      "step": 2000
    },
    {
      "epoch": 0.9635827364112443,
      "grad_norm": 3.0448832511901855,
      "learning_rate": 7.204603601674853e-07,
      "loss": 0.8182637095451355,
      "step": 2001
    },
    {
      "epoch": 0.9640642870041534,
      "grad_norm": 2.527158260345459,
      "learning_rate": 7.018903986483083e-07,
      "loss": 0.6195785403251648,
      "step": 2002
    },
    {
      "epoch": 0.9645458375970626,
      "grad_norm": 2.3781354427337646,
      "learning_rate": 6.835620509122897e-07,
      "loss": 1.2343189716339111,
      "step": 2003
    },
    {
      "epoch": 0.9650273881899717,
      "grad_norm": 3.125363826751709,
      "learning_rate": 6.65475361556267e-07,
      "loss": 0.9276353716850281,
      "step": 2004
    },
    {
      "epoch": 0.9655089387828809,
      "grad_norm": 2.6775107383728027,
      "learning_rate": 6.47630374589081e-07,
      "loss": 1.2648056745529175,
      "step": 2005
    },
    {
      "epoch": 0.96599048937579,
      "grad_norm": 1.5956145524978638,
      "learning_rate": 6.300271334314434e-07,
      "loss": 0.3427439033985138,
      "step": 2006
    },
    {
      "epoch": 0.9664720399686992,
      "grad_norm": 2.645883083343506,
      "learning_rate": 6.126656809158359e-07,
      "loss": 0.7873553037643433,
      "step": 2007
    },
    {
      "epoch": 0.9669535905616083,
      "grad_norm": 1.4090005159378052,
      "learning_rate": 5.955460592864337e-07,
      "loss": 0.6955510377883911,
      "step": 2008
    },
    {
      "epoch": 0.9674351411545176,
      "grad_norm": 1.4704909324645996,
      "learning_rate": 5.78668310198982e-07,
      "loss": 0.7139987945556641,
      "step": 2009
    },
    {
      "epoch": 0.9679166917474267,
      "grad_norm": 4.128110885620117,
      "learning_rate": 5.620324747207084e-07,
      "loss": 0.9668685793876648,
      "step": 2010
    },
    {
      "epoch": 0.9683982423403359,
      "grad_norm": 3.407973527908325,
      "learning_rate": 5.456385933301777e-07,
      "loss": 0.628740668296814,
      "step": 2011
    },
    {
      "epoch": 0.968879792933245,
      "grad_norm": 2.4768881797790527,
      "learning_rate": 5.294867059172592e-07,
      "loss": 0.8109857439994812,
      "step": 2012
    },
    {
      "epoch": 0.9693613435261542,
      "grad_norm": 4.210770130157471,
      "learning_rate": 5.135768517829819e-07,
      "loss": 0.8884713649749756,
      "step": 2013
    },
    {
      "epoch": 0.9698428941190634,
      "grad_norm": 2.278441905975342,
      "learning_rate": 4.979090696394795e-07,
      "loss": 0.757031261920929,
      "step": 2014
    },
    {
      "epoch": 0.9703244447119725,
      "grad_norm": 1.517447590827942,
      "learning_rate": 4.824833976098453e-07,
      "loss": 0.6675227284431458,
      "step": 2015
    },
    {
      "epoch": 0.9708059953048818,
      "grad_norm": 1.8914839029312134,
      "learning_rate": 4.6729987322807757e-07,
      "loss": 0.6843055486679077,
      "step": 2016
    },
    {
      "epoch": 0.9712875458977909,
      "grad_norm": 1.053194522857666,
      "learning_rate": 4.523585334389679e-07,
      "loss": 0.6639283299446106,
      "step": 2017
    },
    {
      "epoch": 0.9717690964907001,
      "grad_norm": 1.5420119762420654,
      "learning_rate": 4.3765941459804614e-07,
      "loss": 0.8280912041664124,
      "step": 2018
    },
    {
      "epoch": 0.9722506470836092,
      "grad_norm": 3.3444290161132812,
      "learning_rate": 4.232025524714356e-07,
      "loss": 0.9113718271255493,
      "step": 2019
    },
    {
      "epoch": 0.9727321976765184,
      "grad_norm": 1.6752758026123047,
      "learning_rate": 4.0898798223582e-07,
      "loss": 0.9070107936859131,
      "step": 2020
    },
    {
      "epoch": 0.9732137482694275,
      "grad_norm": 1.4064198732376099,
      "learning_rate": 3.950157384783104e-07,
      "loss": 0.5857382416725159,
      "step": 2021
    },
    {
      "epoch": 0.9736952988623367,
      "grad_norm": 1.6179803609848022,
      "learning_rate": 3.8128585519640046e-07,
      "loss": 0.5793299674987793,
      "step": 2022
    },
    {
      "epoch": 0.9741768494552459,
      "grad_norm": 1.9372811317443848,
      "learning_rate": 3.677983657978779e-07,
      "loss": 0.7690396904945374,
      "step": 2023
    },
    {
      "epoch": 0.9746584000481551,
      "grad_norm": 1.3107908964157104,
      "learning_rate": 3.545533031007131e-07,
      "loss": 0.3828981816768646,
      "step": 2024
    },
    {
      "epoch": 0.9751399506410642,
      "grad_norm": 2.332566499710083,
      "learning_rate": 3.415506993330153e-07,
      "loss": 0.7227606773376465,
      "step": 2025
    },
    {
      "epoch": 0.9756215012339734,
      "grad_norm": 3.585488796234131,
      "learning_rate": 3.2879058613292105e-07,
      "loss": 0.9689480662345886,
      "step": 2026
    },
    {
      "epoch": 0.9761030518268826,
      "grad_norm": 1.7523847818374634,
      "learning_rate": 3.1627299454856095e-07,
      "loss": 0.5732117891311646,
      "step": 2027
    },
    {
      "epoch": 0.9765846024197917,
      "grad_norm": 1.3317168951034546,
      "learning_rate": 3.0399795503793793e-07,
      "loss": 0.5991673469543457,
      "step": 2028
    },
    {
      "epoch": 0.9770661530127009,
      "grad_norm": 2.6973512172698975,
      "learning_rate": 2.9196549746888235e-07,
      "loss": 0.7186727523803711,
      "step": 2029
    },
    {
      "epoch": 0.97754770360561,
      "grad_norm": 2.531564474105835,
      "learning_rate": 2.801756511189524e-07,
      "loss": 0.55071622133255,
      "step": 2030
    },
    {
      "epoch": 0.9780292541985193,
      "grad_norm": 2.438011884689331,
      "learning_rate": 2.686284446754006e-07,
      "loss": 0.9811122417449951,
      "step": 2031
    },
    {
      "epoch": 0.9785108047914284,
      "grad_norm": 3.77339768409729,
      "learning_rate": 2.573239062350963e-07,
      "loss": 1.080123782157898,
      "step": 2032
    },
    {
      "epoch": 0.9789923553843376,
      "grad_norm": 2.75046443939209,
      "learning_rate": 2.4626206330440326e-07,
      "loss": 0.7446631193161011,
      "step": 2033
    },
    {
      "epoch": 0.9794739059772467,
      "grad_norm": 1.7935954332351685,
      "learning_rate": 2.3544294279918e-07,
      "loss": 0.723296582698822,
      "step": 2034
    },
    {
      "epoch": 0.9799554565701559,
      "grad_norm": 2.720088481903076,
      "learning_rate": 2.2486657104471286e-07,
      "loss": 0.7790317535400391,
      "step": 2035
    },
    {
      "epoch": 0.980437007163065,
      "grad_norm": 3.0413575172424316,
      "learning_rate": 2.1453297377557191e-07,
      "loss": 0.5200697779655457,
      "step": 2036
    },
    {
      "epoch": 0.9809185577559743,
      "grad_norm": 1.7779182195663452,
      "learning_rate": 2.044421761356552e-07,
      "loss": 0.6045622229576111,
      "step": 2037
    },
    {
      "epoch": 0.9814001083488834,
      "grad_norm": 1.672035813331604,
      "learning_rate": 1.9459420267804452e-07,
      "loss": 0.8883457779884338,
      "step": 2038
    },
    {
      "epoch": 0.9818816589417926,
      "grad_norm": 1.9397308826446533,
      "learning_rate": 1.8498907736499426e-07,
      "loss": 0.773249089717865,
      "step": 2039
    },
    {
      "epoch": 0.9823632095347017,
      "grad_norm": 2.310973644256592,
      "learning_rate": 1.7562682356786487e-07,
      "loss": 1.4396414756774902,
      "step": 2040
    },
    {
      "epoch": 0.9828447601276109,
      "grad_norm": 1.1470699310302734,
      "learning_rate": 1.665074640670228e-07,
      "loss": 0.9253256320953369,
      "step": 2041
    },
    {
      "epoch": 0.9833263107205201,
      "grad_norm": 1.9057748317718506,
      "learning_rate": 1.576310210518517e-07,
      "loss": 0.9289815425872803,
      "step": 2042
    },
    {
      "epoch": 0.9838078613134292,
      "grad_norm": 2.2500340938568115,
      "learning_rate": 1.489975161206636e-07,
      "loss": 0.7012436985969543,
      "step": 2043
    },
    {
      "epoch": 0.9842894119063385,
      "grad_norm": 1.6075319051742554,
      "learning_rate": 1.406069702806323e-07,
      "loss": 0.4521413743495941,
      "step": 2044
    },
    {
      "epoch": 0.9847709624992476,
      "grad_norm": 1.4256826639175415,
      "learning_rate": 1.324594039477822e-07,
      "loss": 0.6309791803359985,
      "step": 2045
    },
    {
      "epoch": 0.9852525130921568,
      "grad_norm": 1.9217804670333862,
      "learning_rate": 1.2455483694689962e-07,
      "loss": 0.9839752912521362,
      "step": 2046
    },
    {
      "epoch": 0.9857340636850659,
      "grad_norm": 1.0898535251617432,
      "learning_rate": 1.1689328851151038e-07,
      "loss": 0.7070698142051697,
      "step": 2047
    },
    {
      "epoch": 0.9862156142779751,
      "grad_norm": 1.4987190961837769,
      "learning_rate": 1.0947477728381339e-07,
      "loss": 0.9827098846435547,
      "step": 2048
    },
    {
      "epoch": 0.9866971648708842,
      "grad_norm": 3.4413888454437256,
      "learning_rate": 1.0229932131465836e-07,
      "loss": 0.9514762759208679,
      "step": 2049
    },
    {
      "epoch": 0.9871787154637934,
      "grad_norm": 1.5854613780975342,
      "learning_rate": 9.536693806347919e-08,
      "loss": 1.075221061706543,
      "step": 2050
    },
    {
      "epoch": 0.9876602660567025,
      "grad_norm": 1.3114711046218872,
      "learning_rate": 8.867764439826065e-08,
      "loss": 0.3786975145339966,
      "step": 2051
    },
    {
      "epoch": 0.9881418166496118,
      "grad_norm": 1.548056960105896,
      "learning_rate": 8.223145659550513e-08,
      "loss": 0.5599035620689392,
      "step": 2052
    },
    {
      "epoch": 0.9886233672425209,
      "grad_norm": 2.5505032539367676,
      "learning_rate": 7.602839034017706e-08,
      "loss": 0.9582996368408203,
      "step": 2053
    },
    {
      "epoch": 0.9891049178354301,
      "grad_norm": 1.95079505443573,
      "learning_rate": 7.006846072568074e-08,
      "loss": 0.5530710220336914,
      "step": 2054
    },
    {
      "epoch": 0.9895864684283393,
      "grad_norm": 2.545119524002075,
      "learning_rate": 6.435168225381594e-08,
      "loss": 0.5636353492736816,
      "step": 2055
    },
    {
      "epoch": 0.9900680190212484,
      "grad_norm": 2.5665462017059326,
      "learning_rate": 5.887806883474456e-08,
      "loss": 0.5712226629257202,
      "step": 2056
    },
    {
      "epoch": 0.9905495696141576,
      "grad_norm": 2.021754026412964,
      "learning_rate": 5.364763378694626e-08,
      "loss": 0.9338845610618591,
      "step": 2057
    },
    {
      "epoch": 0.9910311202070667,
      "grad_norm": 1.2843495607376099,
      "learning_rate": 4.8660389837207334e-08,
      "loss": 0.45808300375938416,
      "step": 2058
    },
    {
      "epoch": 0.991512670799976,
      "grad_norm": 4.009743690490723,
      "learning_rate": 4.391634912056519e-08,
      "loss": 0.8420942425727844,
      "step": 2059
    },
    {
      "epoch": 0.9919942213928851,
      "grad_norm": 2.67016339302063,
      "learning_rate": 3.9415523180297286e-08,
      "loss": 0.5695664882659912,
      "step": 2060
    },
    {
      "epoch": 0.9924757719857943,
      "grad_norm": 2.872577428817749,
      "learning_rate": 3.515792296789888e-08,
      "loss": 0.9617867469787598,
      "step": 2061
    },
    {
      "epoch": 0.9929573225787034,
      "grad_norm": 2.436815023422241,
      "learning_rate": 3.114355884301645e-08,
      "loss": 0.8231787085533142,
      "step": 2062
    },
    {
      "epoch": 0.9934388731716126,
      "grad_norm": 2.016432285308838,
      "learning_rate": 2.7372440573469883e-08,
      "loss": 0.7411775588989258,
      "step": 2063
    },
    {
      "epoch": 0.9939204237645217,
      "grad_norm": 2.515350580215454,
      "learning_rate": 2.384457733520806e-08,
      "loss": 1.08511483669281,
      "step": 2064
    },
    {
      "epoch": 0.9944019743574309,
      "grad_norm": 3.87332820892334,
      "learning_rate": 2.0559977712297785e-08,
      "loss": 0.4204919934272766,
      "step": 2065
    },
    {
      "epoch": 0.99488352495034,
      "grad_norm": 1.948048710823059,
      "learning_rate": 1.7518649696857126e-08,
      "loss": 0.6953433156013489,
      "step": 2066
    },
    {
      "epoch": 0.9953650755432493,
      "grad_norm": 1.5868656635284424,
      "learning_rate": 1.4720600689110963e-08,
      "loss": 0.8035791516304016,
      "step": 2067
    },
    {
      "epoch": 0.9958466261361584,
      "grad_norm": 2.0449914932250977,
      "learning_rate": 1.216583749731326e-08,
      "loss": 1.0284188985824585,
      "step": 2068
    },
    {
      "epoch": 0.9963281767290676,
      "grad_norm": 1.3479809761047363,
      "learning_rate": 9.854366337758159e-09,
      "loss": 0.6502044200897217,
      "step": 2069
    },
    {
      "epoch": 0.9968097273219768,
      "grad_norm": 1.4428173303604126,
      "learning_rate": 7.786192834746686e-09,
      "loss": 0.5858941674232483,
      "step": 2070
    },
    {
      "epoch": 0.9972912779148859,
      "grad_norm": 3.1851589679718018,
      "learning_rate": 5.961322020608951e-09,
      "loss": 0.8516849279403687,
      "step": 2071
    },
    {
      "epoch": 0.9977728285077951,
      "grad_norm": 3.6162655353546143,
      "learning_rate": 4.3797583356264275e-09,
      "loss": 0.6134951114654541,
      "step": 2072
    },
    {
      "epoch": 0.9982543791007042,
      "grad_norm": 1.2383058071136475,
      "learning_rate": 3.0415056281096755e-09,
      "loss": 0.49473655223846436,
      "step": 2073
    },
    {
      "epoch": 0.9987359296936135,
      "grad_norm": 2.9997646808624268,
      "learning_rate": 1.9465671543095197e-09,
      "loss": 0.7891368269920349,
      "step": 2074
    },
    {
      "epoch": 0.9992174802865226,
      "grad_norm": 1.883669137954712,
      "learning_rate": 1.094945578439255e-09,
      "loss": 1.0797771215438843,
      "step": 2075
    },
    {
      "epoch": 0.9996990308794318,
      "grad_norm": 2.7585933208465576,
      "learning_rate": 4.866429726857469e-10,
      "loss": 1.0907106399536133,
      "step": 2076
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.393275022506714,
      "learning_rate": 1.2166081717612797e-10,
      "loss": 0.7034765481948853,
      "step": 2077
    },
    {
      "epoch": 1.0,
      "step": 2077,
      "total_flos": 7.016730599121224e+17,
      "train_loss": 0.8948814488398552,
      "train_runtime": 5064.9207,
      "train_samples_per_second": 6.56,
      "train_steps_per_second": 0.41
    }
  ],
  "logging_steps": 1,
  "max_steps": 2077,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2400000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.016730599121224e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}