{
  "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": 321.7454528808594,
      "learning_rate": 0.0,
      "loss": 2.68013858795166,
      "step": 1
    },
    {
      "epoch": 0.0009631011858183351,
      "grad_norm": 67.41497802734375,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 2.275019884109497,
      "step": 2
    },
    {
      "epoch": 0.0014446517787275025,
      "grad_norm": 85.2185287475586,
      "learning_rate": 6.349206349206349e-06,
      "loss": 2.2506096363067627,
      "step": 3
    },
    {
      "epoch": 0.0019262023716366701,
      "grad_norm": 37.25349044799805,
      "learning_rate": 9.523809523809523e-06,
      "loss": 2.663640022277832,
      "step": 4
    },
    {
      "epoch": 0.0024077529645458377,
      "grad_norm": 52.38661193847656,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 2.0638294219970703,
      "step": 5
    },
    {
      "epoch": 0.002889303557455005,
      "grad_norm": 78.00179290771484,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 2.3762643337249756,
      "step": 6
    },
    {
      "epoch": 0.0033708541503641725,
      "grad_norm": 168.51290893554688,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 2.3485310077667236,
      "step": 7
    },
    {
      "epoch": 0.0038524047432733403,
      "grad_norm": 48.69268035888672,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 2.120938777923584,
      "step": 8
    },
    {
      "epoch": 0.004333955336182508,
      "grad_norm": 39.41934585571289,
      "learning_rate": 2.5396825396825397e-05,
      "loss": 2.098552942276001,
      "step": 9
    },
    {
      "epoch": 0.004815505929091675,
      "grad_norm": 26.102609634399414,
      "learning_rate": 2.857142857142857e-05,
      "loss": 2.173189640045166,
      "step": 10
    },
    {
      "epoch": 0.005297056522000842,
      "grad_norm": 86.58882904052734,
      "learning_rate": 3.1746031746031745e-05,
      "loss": 2.512845516204834,
      "step": 11
    },
    {
      "epoch": 0.00577860711491001,
      "grad_norm": 221.00045776367188,
      "learning_rate": 3.492063492063492e-05,
      "loss": 2.7334773540496826,
      "step": 12
    },
    {
      "epoch": 0.006260157707819178,
      "grad_norm": 30.29314422607422,
      "learning_rate": 3.809523809523809e-05,
      "loss": 2.190080165863037,
      "step": 13
    },
    {
      "epoch": 0.006741708300728345,
      "grad_norm": 245.29042053222656,
      "learning_rate": 4.126984126984127e-05,
      "loss": 2.1272945404052734,
      "step": 14
    },
    {
      "epoch": 0.007223258893637513,
      "grad_norm": 49.61991500854492,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 2.0472631454467773,
      "step": 15
    },
    {
      "epoch": 0.0077048094865466805,
      "grad_norm": 613.6011352539062,
      "learning_rate": 4.761904761904762e-05,
      "loss": 2.6586735248565674,
      "step": 16
    },
    {
      "epoch": 0.008186360079455848,
      "grad_norm": 49.33387756347656,
      "learning_rate": 5.0793650793650794e-05,
      "loss": 2.21207594871521,
      "step": 17
    },
    {
      "epoch": 0.008667910672365015,
      "grad_norm": 76.43783569335938,
      "learning_rate": 5.396825396825397e-05,
      "loss": 2.291227102279663,
      "step": 18
    },
    {
      "epoch": 0.009149461265274182,
      "grad_norm": 58.18196487426758,
      "learning_rate": 5.714285714285714e-05,
      "loss": 2.3672409057617188,
      "step": 19
    },
    {
      "epoch": 0.00963101185818335,
      "grad_norm": 114.43791198730469,
      "learning_rate": 6.0317460317460316e-05,
      "loss": 2.505011796951294,
      "step": 20
    },
    {
      "epoch": 0.010112562451092518,
      "grad_norm": 41.25947189331055,
      "learning_rate": 6.349206349206349e-05,
      "loss": 1.8590681552886963,
      "step": 21
    },
    {
      "epoch": 0.010594113044001685,
      "grad_norm": 83.9918212890625,
      "learning_rate": 6.666666666666667e-05,
      "loss": 2.3784050941467285,
      "step": 22
    },
    {
      "epoch": 0.011075663636910853,
      "grad_norm": 238.9562530517578,
      "learning_rate": 6.984126984126984e-05,
      "loss": 2.5110580921173096,
      "step": 23
    },
    {
      "epoch": 0.01155721422982002,
      "grad_norm": 63.45235061645508,
      "learning_rate": 7.301587301587302e-05,
      "loss": 2.249647378921509,
      "step": 24
    },
    {
      "epoch": 0.012038764822729187,
      "grad_norm": 100.17549133300781,
      "learning_rate": 7.619047619047618e-05,
      "loss": 2.2895896434783936,
      "step": 25
    },
    {
      "epoch": 0.012520315415638356,
      "grad_norm": 132.5685272216797,
      "learning_rate": 7.936507936507937e-05,
      "loss": 2.0196285247802734,
      "step": 26
    },
    {
      "epoch": 0.013001866008547523,
      "grad_norm": 52.019126892089844,
      "learning_rate": 8.253968253968255e-05,
      "loss": 1.8349158763885498,
      "step": 27
    },
    {
      "epoch": 0.01348341660145669,
      "grad_norm": 39.554996490478516,
      "learning_rate": 8.571428571428571e-05,
      "loss": 2.197366952896118,
      "step": 28
    },
    {
      "epoch": 0.013964967194365858,
      "grad_norm": 157.50970458984375,
      "learning_rate": 8.888888888888889e-05,
      "loss": 2.35345458984375,
      "step": 29
    },
    {
      "epoch": 0.014446517787275025,
      "grad_norm": 44.91498565673828,
      "learning_rate": 9.206349206349206e-05,
      "loss": 1.953705072402954,
      "step": 30
    },
    {
      "epoch": 0.014928068380184192,
      "grad_norm": 103.84178161621094,
      "learning_rate": 9.523809523809524e-05,
      "loss": 2.610750913619995,
      "step": 31
    },
    {
      "epoch": 0.015409618973093361,
      "grad_norm": 34.198402404785156,
      "learning_rate": 9.841269841269841e-05,
      "loss": 2.6935482025146484,
      "step": 32
    },
    {
      "epoch": 0.01589116956600253,
      "grad_norm": 233.34249877929688,
      "learning_rate": 0.00010158730158730159,
      "loss": 1.9780076742172241,
      "step": 33
    },
    {
      "epoch": 0.016372720158911697,
      "grad_norm": 26.665225982666016,
      "learning_rate": 0.00010476190476190477,
      "loss": 2.2433481216430664,
      "step": 34
    },
    {
      "epoch": 0.016854270751820864,
      "grad_norm": 75.23570251464844,
      "learning_rate": 0.00010793650793650794,
      "loss": 2.486795425415039,
      "step": 35
    },
    {
      "epoch": 0.01733582134473003,
      "grad_norm": 124.65180969238281,
      "learning_rate": 0.00011111111111111112,
      "loss": 2.4075000286102295,
      "step": 36
    },
    {
      "epoch": 0.017817371937639197,
      "grad_norm": 220.2021942138672,
      "learning_rate": 0.00011428571428571428,
      "loss": 2.5082898139953613,
      "step": 37
    },
    {
      "epoch": 0.018298922530548364,
      "grad_norm": 665.28759765625,
      "learning_rate": 0.00011746031746031746,
      "loss": 2.283926010131836,
      "step": 38
    },
    {
      "epoch": 0.018780473123457535,
      "grad_norm": 94.30010986328125,
      "learning_rate": 0.00012063492063492063,
      "loss": 2.164767265319824,
      "step": 39
    },
    {
      "epoch": 0.0192620237163667,
      "grad_norm": 61.93984603881836,
      "learning_rate": 0.0001238095238095238,
      "loss": 2.453726291656494,
      "step": 40
    },
    {
      "epoch": 0.01974357430927587,
      "grad_norm": 84.58031463623047,
      "learning_rate": 0.00012698412698412698,
      "loss": 2.3571081161499023,
      "step": 41
    },
    {
      "epoch": 0.020225124902185036,
      "grad_norm": 132.82244873046875,
      "learning_rate": 0.00013015873015873017,
      "loss": 2.0443947315216064,
      "step": 42
    },
    {
      "epoch": 0.020706675495094203,
      "grad_norm": 886.052001953125,
      "learning_rate": 0.00013333333333333334,
      "loss": 2.1403098106384277,
      "step": 43
    },
    {
      "epoch": 0.02118822608800337,
      "grad_norm": 187.56796264648438,
      "learning_rate": 0.0001365079365079365,
      "loss": 2.8176968097686768,
      "step": 44
    },
    {
      "epoch": 0.02166977668091254,
      "grad_norm": 72.96453857421875,
      "learning_rate": 0.00013968253968253967,
      "loss": 2.4982311725616455,
      "step": 45
    },
    {
      "epoch": 0.022151327273821707,
      "grad_norm": 500.8731384277344,
      "learning_rate": 0.00014285714285714287,
      "loss": 2.5446393489837646,
      "step": 46
    },
    {
      "epoch": 0.022632877866730874,
      "grad_norm": 74.3232650756836,
      "learning_rate": 0.00014603174603174603,
      "loss": 2.6876707077026367,
      "step": 47
    },
    {
      "epoch": 0.02311442845964004,
      "grad_norm": 298.7872619628906,
      "learning_rate": 0.00014920634920634923,
      "loss": 2.645242929458618,
      "step": 48
    },
    {
      "epoch": 0.023595979052549208,
      "grad_norm": 73.13448333740234,
      "learning_rate": 0.00015238095238095237,
      "loss": 2.8086609840393066,
      "step": 49
    },
    {
      "epoch": 0.024077529645458375,
      "grad_norm": 10569.1982421875,
      "learning_rate": 0.00015555555555555556,
      "loss": 2.462059736251831,
      "step": 50
    },
    {
      "epoch": 0.024559080238367545,
      "grad_norm": 159.7171173095703,
      "learning_rate": 0.00015873015873015873,
      "loss": 2.2481374740600586,
      "step": 51
    },
    {
      "epoch": 0.025040630831276712,
      "grad_norm": 31.103609085083008,
      "learning_rate": 0.00016190476190476192,
      "loss": 2.2522459030151367,
      "step": 52
    },
    {
      "epoch": 0.02552218142418588,
      "grad_norm": 69.71514892578125,
      "learning_rate": 0.0001650793650793651,
      "loss": 2.1398792266845703,
      "step": 53
    },
    {
      "epoch": 0.026003732017095046,
      "grad_norm": 55.43810272216797,
      "learning_rate": 0.00016825396825396826,
      "loss": 2.314161539077759,
      "step": 54
    },
    {
      "epoch": 0.026485282610004213,
      "grad_norm": 213.0353240966797,
      "learning_rate": 0.00017142857142857143,
      "loss": 2.636957883834839,
      "step": 55
    },
    {
      "epoch": 0.02696683320291338,
      "grad_norm": 55.90201950073242,
      "learning_rate": 0.00017460317460317462,
      "loss": 2.024723768234253,
      "step": 56
    },
    {
      "epoch": 0.02744838379582255,
      "grad_norm": 22.28119659423828,
      "learning_rate": 0.00017777777777777779,
      "loss": 2.180311679840088,
      "step": 57
    },
    {
      "epoch": 0.027929934388731717,
      "grad_norm": 52.47058868408203,
      "learning_rate": 0.00018095238095238095,
      "loss": 2.132675886154175,
      "step": 58
    },
    {
      "epoch": 0.028411484981640884,
      "grad_norm": 148.48812866210938,
      "learning_rate": 0.00018412698412698412,
      "loss": 2.3559181690216064,
      "step": 59
    },
    {
      "epoch": 0.02889303557455005,
      "grad_norm": 48.93524932861328,
      "learning_rate": 0.00018730158730158731,
      "loss": 1.958573818206787,
      "step": 60
    },
    {
      "epoch": 0.029374586167459218,
      "grad_norm": 228.8285675048828,
      "learning_rate": 0.00019047619047619048,
      "loss": 2.456819772720337,
      "step": 61
    },
    {
      "epoch": 0.029856136760368385,
      "grad_norm": 110.63797760009766,
      "learning_rate": 0.00019365079365079365,
      "loss": 2.1174046993255615,
      "step": 62
    },
    {
      "epoch": 0.030337687353277555,
      "grad_norm": 325.382080078125,
      "learning_rate": 0.00019682539682539682,
      "loss": 2.404069185256958,
      "step": 63
    },
    {
      "epoch": 0.030819237946186722,
      "grad_norm": 62.22050476074219,
      "learning_rate": 0.0002,
      "loss": 2.187788724899292,
      "step": 64
    },
    {
      "epoch": 0.03130078853909589,
      "grad_norm": 117.30904388427734,
      "learning_rate": 0.00019999987833918285,
      "loss": 2.2458250522613525,
      "step": 65
    },
    {
      "epoch": 0.03178233913200506,
      "grad_norm": 107.20811462402344,
      "learning_rate": 0.00019999951335702735,
      "loss": 2.6168434619903564,
      "step": 66
    },
    {
      "epoch": 0.03226388972491422,
      "grad_norm": 346.3019714355469,
      "learning_rate": 0.00019999890505442158,
      "loss": 2.321580648422241,
      "step": 67
    },
    {
      "epoch": 0.03274544031782339,
      "grad_norm": 538.304443359375,
      "learning_rate": 0.0001999980534328457,
      "loss": 2.2181715965270996,
      "step": 68
    },
    {
      "epoch": 0.03322699091073256,
      "grad_norm": 42.598697662353516,
      "learning_rate": 0.0001999969584943719,
      "loss": 2.6818625926971436,
      "step": 69
    },
    {
      "epoch": 0.03370854150364173,
      "grad_norm": 58.62196731567383,
      "learning_rate": 0.00019999562024166438,
      "loss": 2.6244213581085205,
      "step": 70
    },
    {
      "epoch": 0.03419009209655089,
      "grad_norm": 1473.3076171875,
      "learning_rate": 0.0001999940386779794,
      "loss": 2.300719976425171,
      "step": 71
    },
    {
      "epoch": 0.03467164268946006,
      "grad_norm": 181.53558349609375,
      "learning_rate": 0.00019999221380716527,
      "loss": 2.9540743827819824,
      "step": 72
    },
    {
      "epoch": 0.03515319328236923,
      "grad_norm": 33.675357818603516,
      "learning_rate": 0.00019999014563366226,
      "loss": 2.26188588142395,
      "step": 73
    },
    {
      "epoch": 0.035634743875278395,
      "grad_norm": 73.54495239257812,
      "learning_rate": 0.00019998783416250268,
      "loss": 2.581285238265991,
      "step": 74
    },
    {
      "epoch": 0.036116294468187565,
      "grad_norm": 63.8938102722168,
      "learning_rate": 0.0001999852793993109,
      "loss": 2.5267906188964844,
      "step": 75
    },
    {
      "epoch": 0.03659784506109673,
      "grad_norm": 222.57411193847656,
      "learning_rate": 0.00019998248135030315,
      "loss": 2.3622689247131348,
      "step": 76
    },
    {
      "epoch": 0.0370793956540059,
      "grad_norm": 53.57831954956055,
      "learning_rate": 0.00019997944002228774,
      "loss": 2.0972423553466797,
      "step": 77
    },
    {
      "epoch": 0.03756094624691507,
      "grad_norm": 135.51170349121094,
      "learning_rate": 0.00019997615542266482,
      "loss": 2.23136830329895,
      "step": 78
    },
    {
      "epoch": 0.03804249683982423,
      "grad_norm": 73.25692749023438,
      "learning_rate": 0.00019997262755942655,
      "loss": 2.7493536472320557,
      "step": 79
    },
    {
      "epoch": 0.0385240474327334,
      "grad_norm": 69.76377868652344,
      "learning_rate": 0.000199968856441157,
      "loss": 2.688168525695801,
      "step": 80
    },
    {
      "epoch": 0.03900559802564257,
      "grad_norm": 612.4141235351562,
      "learning_rate": 0.0001999648420770321,
      "loss": 2.573000192642212,
      "step": 81
    },
    {
      "epoch": 0.03948714861855174,
      "grad_norm": 62.284610748291016,
      "learning_rate": 0.0001999605844768197,
      "loss": 2.42378568649292,
      "step": 82
    },
    {
      "epoch": 0.0399686992114609,
      "grad_norm": 40.93765640258789,
      "learning_rate": 0.00019995608365087946,
      "loss": 2.4993321895599365,
      "step": 83
    },
    {
      "epoch": 0.04045024980437007,
      "grad_norm": 567.7521362304688,
      "learning_rate": 0.0001999513396101628,
      "loss": 1.890772819519043,
      "step": 84
    },
    {
      "epoch": 0.04093180039727924,
      "grad_norm": 163.58010864257812,
      "learning_rate": 0.00019994635236621306,
      "loss": 2.644516706466675,
      "step": 85
    },
    {
      "epoch": 0.041413350990188405,
      "grad_norm": 57.1867561340332,
      "learning_rate": 0.00019994112193116528,
      "loss": 2.514359951019287,
      "step": 86
    },
    {
      "epoch": 0.041894901583097575,
      "grad_norm": 42.296714782714844,
      "learning_rate": 0.00019993564831774618,
      "loss": 2.2242441177368164,
      "step": 87
    },
    {
      "epoch": 0.04237645217600674,
      "grad_norm": 493.85675048828125,
      "learning_rate": 0.00019992993153927432,
      "loss": 2.708584785461426,
      "step": 88
    },
    {
      "epoch": 0.04285800276891591,
      "grad_norm": 176.70016479492188,
      "learning_rate": 0.00019992397160965982,
      "loss": 2.3863372802734375,
      "step": 89
    },
    {
      "epoch": 0.04333955336182508,
      "grad_norm": 39.418819427490234,
      "learning_rate": 0.0001999177685434045,
      "loss": 2.3180670738220215,
      "step": 90
    },
    {
      "epoch": 0.04382110395473424,
      "grad_norm": 102.74596405029297,
      "learning_rate": 0.00019991132235560176,
      "loss": 2.3300135135650635,
      "step": 91
    },
    {
      "epoch": 0.044302654547643414,
      "grad_norm": 21.294551849365234,
      "learning_rate": 0.00019990463306193652,
      "loss": 2.324836492538452,
      "step": 92
    },
    {
      "epoch": 0.04478420514055258,
      "grad_norm": 113.57337951660156,
      "learning_rate": 0.00019989770067868533,
      "loss": 2.278836488723755,
      "step": 93
    },
    {
      "epoch": 0.04526575573346175,
      "grad_norm": 53.526851654052734,
      "learning_rate": 0.00019989052522271622,
      "loss": 2.3314788341522217,
      "step": 94
    },
    {
      "epoch": 0.04574730632637091,
      "grad_norm": 226.02947998046875,
      "learning_rate": 0.00019988310671148848,
      "loss": 2.870971441268921,
      "step": 95
    },
    {
      "epoch": 0.04622885691928008,
      "grad_norm": 77.19041442871094,
      "learning_rate": 0.00019987544516305311,
      "loss": 2.255547523498535,
      "step": 96
    },
    {
      "epoch": 0.04671040751218925,
      "grad_norm": 62.48857498168945,
      "learning_rate": 0.00019986754059605222,
      "loss": 2.1827213764190674,
      "step": 97
    },
    {
      "epoch": 0.047191958105098415,
      "grad_norm": 131.2965850830078,
      "learning_rate": 0.00019985939302971938,
      "loss": 2.742527723312378,
      "step": 98
    },
    {
      "epoch": 0.047673508698007586,
      "grad_norm": 96.15840148925781,
      "learning_rate": 0.00019985100248387933,
      "loss": 2.3680310249328613,
      "step": 99
    },
    {
      "epoch": 0.04815505929091675,
      "grad_norm": 57.02033615112305,
      "learning_rate": 0.00019984236897894816,
      "loss": 2.721726417541504,
      "step": 100
    },
    {
      "epoch": 0.04863660988382592,
      "grad_norm": 41.55272674560547,
      "learning_rate": 0.000199833492535933,
      "loss": 2.256671190261841,
      "step": 101
    },
    {
      "epoch": 0.04911816047673509,
      "grad_norm": 102.20014190673828,
      "learning_rate": 0.00019982437317643217,
      "loss": 2.308504343032837,
      "step": 102
    },
    {
      "epoch": 0.04959971106964425,
      "grad_norm": 52.46580123901367,
      "learning_rate": 0.00019981501092263503,
      "loss": 2.4225263595581055,
      "step": 103
    },
    {
      "epoch": 0.050081261662553424,
      "grad_norm": 170.1755828857422,
      "learning_rate": 0.00019980540579732196,
      "loss": 2.40871524810791,
      "step": 104
    },
    {
      "epoch": 0.05056281225546259,
      "grad_norm": 388.2884826660156,
      "learning_rate": 0.00019979555782386434,
      "loss": 2.158656120300293,
      "step": 105
    },
    {
      "epoch": 0.05104436284837176,
      "grad_norm": 56.656585693359375,
      "learning_rate": 0.00019978546702622443,
      "loss": 2.3279011249542236,
      "step": 106
    },
    {
      "epoch": 0.05152591344128093,
      "grad_norm": 67.324951171875,
      "learning_rate": 0.00019977513342895532,
      "loss": 2.5031135082244873,
      "step": 107
    },
    {
      "epoch": 0.05200746403419009,
      "grad_norm": 136.13160705566406,
      "learning_rate": 0.00019976455705720083,
      "loss": 2.4979424476623535,
      "step": 108
    },
    {
      "epoch": 0.05248901462709926,
      "grad_norm": 467.31304931640625,
      "learning_rate": 0.0001997537379366956,
      "loss": 2.2675154209136963,
      "step": 109
    },
    {
      "epoch": 0.052970565220008425,
      "grad_norm": 67.81368255615234,
      "learning_rate": 0.00019974267609376494,
      "loss": 2.101496458053589,
      "step": 110
    },
    {
      "epoch": 0.053452115812917596,
      "grad_norm": 128.62509155273438,
      "learning_rate": 0.00019973137155532462,
      "loss": 2.4258201122283936,
      "step": 111
    },
    {
      "epoch": 0.05393366640582676,
      "grad_norm": 62.502769470214844,
      "learning_rate": 0.00019971982434888107,
      "loss": 2.259277820587158,
      "step": 112
    },
    {
      "epoch": 0.05441521699873593,
      "grad_norm": 34.0765266418457,
      "learning_rate": 0.00019970803450253114,
      "loss": 2.2929930686950684,
      "step": 113
    },
    {
      "epoch": 0.0548967675916451,
      "grad_norm": 86.18826293945312,
      "learning_rate": 0.0001996960020449621,
      "loss": 2.483612298965454,
      "step": 114
    },
    {
      "epoch": 0.055378318184554264,
      "grad_norm": 31.546741485595703,
      "learning_rate": 0.00019968372700545145,
      "loss": 2.143441677093506,
      "step": 115
    },
    {
      "epoch": 0.055859868777463434,
      "grad_norm": 92.97311401367188,
      "learning_rate": 0.00019967120941386709,
      "loss": 2.3038527965545654,
      "step": 116
    },
    {
      "epoch": 0.0563414193703726,
      "grad_norm": 886.7743530273438,
      "learning_rate": 0.000199658449300667,
      "loss": 2.2860076427459717,
      "step": 117
    },
    {
      "epoch": 0.05682296996328177,
      "grad_norm": 126.09291076660156,
      "learning_rate": 0.00019964544669689928,
      "loss": 2.479112386703491,
      "step": 118
    },
    {
      "epoch": 0.05730452055619094,
      "grad_norm": 53.912899017333984,
      "learning_rate": 0.00019963220163420214,
      "loss": 2.902468681335449,
      "step": 119
    },
    {
      "epoch": 0.0577860711491001,
      "grad_norm": 84.75859832763672,
      "learning_rate": 0.0001996187141448036,
      "loss": 2.6681010723114014,
      "step": 120
    },
    {
      "epoch": 0.05826762174200927,
      "grad_norm": 118.74105834960938,
      "learning_rate": 0.0001996049842615217,
      "loss": 2.3271281719207764,
      "step": 121
    },
    {
      "epoch": 0.058749172334918436,
      "grad_norm": 105.51011657714844,
      "learning_rate": 0.0001995910120177642,
      "loss": 2.174503803253174,
      "step": 122
    },
    {
      "epoch": 0.059230722927827606,
      "grad_norm": 71.1513671875,
      "learning_rate": 0.00019957679744752859,
      "loss": 2.479018211364746,
      "step": 123
    },
    {
      "epoch": 0.05971227352073677,
      "grad_norm": 41.507179260253906,
      "learning_rate": 0.00019956234058540195,
      "loss": 2.2723090648651123,
      "step": 124
    },
    {
      "epoch": 0.06019382411364594,
      "grad_norm": 1894.11083984375,
      "learning_rate": 0.00019954764146656105,
      "loss": 2.6347084045410156,
      "step": 125
    },
    {
      "epoch": 0.06067537470655511,
      "grad_norm": 57.88983154296875,
      "learning_rate": 0.00019953270012677195,
      "loss": 2.3583426475524902,
      "step": 126
    },
    {
      "epoch": 0.061156925299464274,
      "grad_norm": 49.12632369995117,
      "learning_rate": 0.00019951751660239015,
      "loss": 2.317436456680298,
      "step": 127
    },
    {
      "epoch": 0.061638475892373444,
      "grad_norm": 75.92652130126953,
      "learning_rate": 0.00019950209093036052,
      "loss": 2.3678700923919678,
      "step": 128
    },
    {
      "epoch": 0.06212002648528261,
      "grad_norm": 77.8193130493164,
      "learning_rate": 0.000199486423148217,
      "loss": 2.627444267272949,
      "step": 129
    },
    {
      "epoch": 0.06260157707819178,
      "grad_norm": 58.36602783203125,
      "learning_rate": 0.00019947051329408276,
      "loss": 2.4588303565979004,
      "step": 130
    },
    {
      "epoch": 0.06308312767110094,
      "grad_norm": 125.43273162841797,
      "learning_rate": 0.00019945436140666981,
      "loss": 2.4778406620025635,
      "step": 131
    },
    {
      "epoch": 0.06356467826401012,
      "grad_norm": 186.53868103027344,
      "learning_rate": 0.0001994379675252793,
      "loss": 2.44643497467041,
      "step": 132
    },
    {
      "epoch": 0.06404622885691928,
      "grad_norm": 47.93274688720703,
      "learning_rate": 0.00019942133168980103,
      "loss": 2.655207633972168,
      "step": 133
    },
    {
      "epoch": 0.06452777944982845,
      "grad_norm": 127.77775573730469,
      "learning_rate": 0.00019940445394071355,
      "loss": 2.423912763595581,
      "step": 134
    },
    {
      "epoch": 0.06500933004273761,
      "grad_norm": 5075.32080078125,
      "learning_rate": 0.0001993873343190842,
      "loss": 2.35478138923645,
      "step": 135
    },
    {
      "epoch": 0.06549088063564679,
      "grad_norm": 75.9028549194336,
      "learning_rate": 0.00019936997286656855,
      "loss": 2.3184895515441895,
      "step": 136
    },
    {
      "epoch": 0.06597243122855595,
      "grad_norm": 98.64381408691406,
      "learning_rate": 0.00019935236962541092,
      "loss": 2.939502000808716,
      "step": 137
    },
    {
      "epoch": 0.06645398182146511,
      "grad_norm": 50.403648376464844,
      "learning_rate": 0.00019933452463844376,
      "loss": 2.5624372959136963,
      "step": 138
    },
    {
      "epoch": 0.06693553241437429,
      "grad_norm": 57.519020080566406,
      "learning_rate": 0.00019931643794908772,
      "loss": 2.4487826824188232,
      "step": 139
    },
    {
      "epoch": 0.06741708300728345,
      "grad_norm": 38.554595947265625,
      "learning_rate": 0.00019929810960135172,
      "loss": 2.5147571563720703,
      "step": 140
    },
    {
      "epoch": 0.06789863360019262,
      "grad_norm": 76.87574768066406,
      "learning_rate": 0.00019927953963983254,
      "loss": 2.37192440032959,
      "step": 141
    },
    {
      "epoch": 0.06838018419310178,
      "grad_norm": 29.697391510009766,
      "learning_rate": 0.00019926072810971492,
      "loss": 2.109170913696289,
      "step": 142
    },
    {
      "epoch": 0.06886173478601096,
      "grad_norm": 80.23985290527344,
      "learning_rate": 0.00019924167505677137,
      "loss": 2.8158271312713623,
      "step": 143
    },
    {
      "epoch": 0.06934328537892012,
      "grad_norm": 88.70072937011719,
      "learning_rate": 0.00019922238052736215,
      "loss": 2.4240474700927734,
      "step": 144
    },
    {
      "epoch": 0.06982483597182929,
      "grad_norm": 49.948097229003906,
      "learning_rate": 0.00019920284456843498,
      "loss": 2.288271188735962,
      "step": 145
    },
    {
      "epoch": 0.07030638656473846,
      "grad_norm": 74.19551849365234,
      "learning_rate": 0.00019918306722752505,
      "loss": 2.364229202270508,
      "step": 146
    },
    {
      "epoch": 0.07078793715764763,
      "grad_norm": 220.5440673828125,
      "learning_rate": 0.00019916304855275497,
      "loss": 2.8707008361816406,
      "step": 147
    },
    {
      "epoch": 0.07126948775055679,
      "grad_norm": 219.0526123046875,
      "learning_rate": 0.00019914278859283445,
      "loss": 3.1257731914520264,
      "step": 148
    },
    {
      "epoch": 0.07175103834346597,
      "grad_norm": 385.242919921875,
      "learning_rate": 0.0001991222873970604,
      "loss": 2.6987204551696777,
      "step": 149
    },
    {
      "epoch": 0.07223258893637513,
      "grad_norm": 237.09764099121094,
      "learning_rate": 0.00019910154501531663,
      "loss": 2.4596710205078125,
      "step": 150
    },
    {
      "epoch": 0.0727141395292843,
      "grad_norm": 71.6876220703125,
      "learning_rate": 0.0001990805614980739,
      "loss": 2.165684938430786,
      "step": 151
    },
    {
      "epoch": 0.07319569012219346,
      "grad_norm": 52.48270797729492,
      "learning_rate": 0.00019905933689638955,
      "loss": 2.5384480953216553,
      "step": 152
    },
    {
      "epoch": 0.07367724071510263,
      "grad_norm": 234.6460418701172,
      "learning_rate": 0.00019903787126190772,
      "loss": 2.8752388954162598,
      "step": 153
    },
    {
      "epoch": 0.0741587913080118,
      "grad_norm": 112.19245147705078,
      "learning_rate": 0.00019901616464685888,
      "loss": 2.8501789569854736,
      "step": 154
    },
    {
      "epoch": 0.07464034190092096,
      "grad_norm": 693.4070434570312,
      "learning_rate": 0.00019899421710405996,
      "loss": 3.326218366622925,
      "step": 155
    },
    {
      "epoch": 0.07512189249383014,
      "grad_norm": 180.6138153076172,
      "learning_rate": 0.00019897202868691407,
      "loss": 3.1607935428619385,
      "step": 156
    },
    {
      "epoch": 0.0756034430867393,
      "grad_norm": 110.9269790649414,
      "learning_rate": 0.00019894959944941038,
      "loss": 2.333319902420044,
      "step": 157
    },
    {
      "epoch": 0.07608499367964847,
      "grad_norm": 329.3002624511719,
      "learning_rate": 0.0001989269294461242,
      "loss": 2.459348678588867,
      "step": 158
    },
    {
      "epoch": 0.07656654427255763,
      "grad_norm": 215.49281311035156,
      "learning_rate": 0.0001989040187322164,
      "loss": 2.876829147338867,
      "step": 159
    },
    {
      "epoch": 0.0770480948654668,
      "grad_norm": 311.6734313964844,
      "learning_rate": 0.00019888086736343384,
      "loss": 2.3887791633605957,
      "step": 160
    },
    {
      "epoch": 0.07752964545837597,
      "grad_norm": 147.3771514892578,
      "learning_rate": 0.0001988574753961087,
      "loss": 3.4750375747680664,
      "step": 161
    },
    {
      "epoch": 0.07801119605128513,
      "grad_norm": 92.67632293701172,
      "learning_rate": 0.00019883384288715874,
      "loss": 2.9961111545562744,
      "step": 162
    },
    {
      "epoch": 0.07849274664419431,
      "grad_norm": 930.2900390625,
      "learning_rate": 0.000198809969894087,
      "loss": 2.592177391052246,
      "step": 163
    },
    {
      "epoch": 0.07897429723710347,
      "grad_norm": 64.02899169921875,
      "learning_rate": 0.0001987858564749816,
      "loss": 2.8908722400665283,
      "step": 164
    },
    {
      "epoch": 0.07945584783001264,
      "grad_norm": 45.44175338745117,
      "learning_rate": 0.00019876150268851572,
      "loss": 2.3945958614349365,
      "step": 165
    },
    {
      "epoch": 0.0799373984229218,
      "grad_norm": 718.85546875,
      "learning_rate": 0.00019873690859394737,
      "loss": 2.3570098876953125,
      "step": 166
    },
    {
      "epoch": 0.08041894901583098,
      "grad_norm": 69.55239868164062,
      "learning_rate": 0.0001987120742511193,
      "loss": 2.653460741043091,
      "step": 167
    },
    {
      "epoch": 0.08090049960874014,
      "grad_norm": 369.64404296875,
      "learning_rate": 0.0001986869997204589,
      "loss": 2.779702663421631,
      "step": 168
    },
    {
      "epoch": 0.0813820502016493,
      "grad_norm": 99.79022216796875,
      "learning_rate": 0.00019866168506297788,
      "loss": 2.9537105560302734,
      "step": 169
    },
    {
      "epoch": 0.08186360079455848,
      "grad_norm": 344.4448547363281,
      "learning_rate": 0.00019863613034027224,
      "loss": 3.2134697437286377,
      "step": 170
    },
    {
      "epoch": 0.08234515138746765,
      "grad_norm": 95.29148864746094,
      "learning_rate": 0.00019861033561452223,
      "loss": 2.7037200927734375,
      "step": 171
    },
    {
      "epoch": 0.08282670198037681,
      "grad_norm": 294.568115234375,
      "learning_rate": 0.00019858430094849195,
      "loss": 3.050241470336914,
      "step": 172
    },
    {
      "epoch": 0.08330825257328599,
      "grad_norm": 54.81675720214844,
      "learning_rate": 0.0001985580264055294,
      "loss": 2.581057548522949,
      "step": 173
    },
    {
      "epoch": 0.08378980316619515,
      "grad_norm": 75.01016998291016,
      "learning_rate": 0.00019853151204956616,
      "loss": 3.2000341415405273,
      "step": 174
    },
    {
      "epoch": 0.08427135375910431,
      "grad_norm": 59.06689453125,
      "learning_rate": 0.00019850475794511749,
      "loss": 3.7143819332122803,
      "step": 175
    },
    {
      "epoch": 0.08475290435201348,
      "grad_norm": 231.8850555419922,
      "learning_rate": 0.00019847776415728185,
      "loss": 3.169750690460205,
      "step": 176
    },
    {
      "epoch": 0.08523445494492266,
      "grad_norm": 468.8564453125,
      "learning_rate": 0.000198450530751741,
      "loss": 3.052363157272339,
      "step": 177
    },
    {
      "epoch": 0.08571600553783182,
      "grad_norm": 135.98670959472656,
      "learning_rate": 0.00019842305779475968,
      "loss": 3.4236221313476562,
      "step": 178
    },
    {
      "epoch": 0.08619755613074098,
      "grad_norm": 2636.9462890625,
      "learning_rate": 0.00019839534535318558,
      "loss": 3.00744891166687,
      "step": 179
    },
    {
      "epoch": 0.08667910672365016,
      "grad_norm": 199.4244842529297,
      "learning_rate": 0.00019836739349444899,
      "loss": 2.7622032165527344,
      "step": 180
    },
    {
      "epoch": 0.08716065731655932,
      "grad_norm": 173.1476593017578,
      "learning_rate": 0.00019833920228656292,
      "loss": 3.1958789825439453,
      "step": 181
    },
    {
      "epoch": 0.08764220790946849,
      "grad_norm": 74.40721130371094,
      "learning_rate": 0.0001983107717981226,
      "loss": 3.347377300262451,
      "step": 182
    },
    {
      "epoch": 0.08812375850237765,
      "grad_norm": 71.2850112915039,
      "learning_rate": 0.00019828210209830562,
      "loss": 2.889693021774292,
      "step": 183
    },
    {
      "epoch": 0.08860530909528683,
      "grad_norm": 98.93665313720703,
      "learning_rate": 0.00019825319325687154,
      "loss": 3.283125162124634,
      "step": 184
    },
    {
      "epoch": 0.08908685968819599,
      "grad_norm": 266.1285095214844,
      "learning_rate": 0.00019822404534416182,
      "loss": 3.6556084156036377,
      "step": 185
    },
    {
      "epoch": 0.08956841028110515,
      "grad_norm": 73.89322662353516,
      "learning_rate": 0.00019819465843109963,
      "loss": 3.265658378601074,
      "step": 186
    },
    {
      "epoch": 0.09004996087401433,
      "grad_norm": 541.5470581054688,
      "learning_rate": 0.00019816503258918969,
      "loss": 3.329449415206909,
      "step": 187
    },
    {
      "epoch": 0.0905315114669235,
      "grad_norm": 83.84955596923828,
      "learning_rate": 0.00019813516789051808,
      "loss": 3.066371440887451,
      "step": 188
    },
    {
      "epoch": 0.09101306205983266,
      "grad_norm": 82.78404998779297,
      "learning_rate": 0.0001981050644077521,
      "loss": 2.724719285964966,
      "step": 189
    },
    {
      "epoch": 0.09149461265274182,
      "grad_norm": 162.23480224609375,
      "learning_rate": 0.00019807472221414002,
      "loss": 3.716334342956543,
      "step": 190
    },
    {
      "epoch": 0.091976163245651,
      "grad_norm": 233.0549774169922,
      "learning_rate": 0.00019804414138351094,
      "loss": 3.276333808898926,
      "step": 191
    },
    {
      "epoch": 0.09245771383856016,
      "grad_norm": 93.35044860839844,
      "learning_rate": 0.00019801332199027467,
      "loss": 2.927642822265625,
      "step": 192
    },
    {
      "epoch": 0.09293926443146933,
      "grad_norm": 38.928810119628906,
      "learning_rate": 0.00019798226410942146,
      "loss": 2.8423051834106445,
      "step": 193
    },
    {
      "epoch": 0.0934208150243785,
      "grad_norm": 86.59122467041016,
      "learning_rate": 0.00019795096781652182,
      "loss": 2.7899434566497803,
      "step": 194
    },
    {
      "epoch": 0.09390236561728767,
      "grad_norm": 74.2641372680664,
      "learning_rate": 0.00019791943318772643,
      "loss": 3.0941240787506104,
      "step": 195
    },
    {
      "epoch": 0.09438391621019683,
      "grad_norm": 31.566852569580078,
      "learning_rate": 0.00019788766029976587,
      "loss": 2.926772117614746,
      "step": 196
    },
    {
      "epoch": 0.09486546680310601,
      "grad_norm": 22.040136337280273,
      "learning_rate": 0.0001978556492299504,
      "loss": 3.0792505741119385,
      "step": 197
    },
    {
      "epoch": 0.09534701739601517,
      "grad_norm": 40.565956115722656,
      "learning_rate": 0.00019782340005616996,
      "loss": 3.020026206970215,
      "step": 198
    },
    {
      "epoch": 0.09582856798892433,
      "grad_norm": 142.21755981445312,
      "learning_rate": 0.0001977909128568937,
      "loss": 2.9219775199890137,
      "step": 199
    },
    {
      "epoch": 0.0963101185818335,
      "grad_norm": 130.58206176757812,
      "learning_rate": 0.00019775818771117,
      "loss": 3.623426675796509,
      "step": 200
    },
    {
      "epoch": 0.09679166917474268,
      "grad_norm": 145.4685516357422,
      "learning_rate": 0.00019772522469862626,
      "loss": 2.5374817848205566,
      "step": 201
    },
    {
      "epoch": 0.09727321976765184,
      "grad_norm": 60.53111267089844,
      "learning_rate": 0.00019769202389946863,
      "loss": 3.070513963699341,
      "step": 202
    },
    {
      "epoch": 0.097754770360561,
      "grad_norm": 49.229087829589844,
      "learning_rate": 0.0001976585853944818,
      "loss": 3.7401344776153564,
      "step": 203
    },
    {
      "epoch": 0.09823632095347018,
      "grad_norm": 13.168242454528809,
      "learning_rate": 0.0001976249092650289,
      "loss": 2.871077537536621,
      "step": 204
    },
    {
      "epoch": 0.09871787154637934,
      "grad_norm": 20.444995880126953,
      "learning_rate": 0.00019759099559305124,
      "loss": 1.9388713836669922,
      "step": 205
    },
    {
      "epoch": 0.0991994221392885,
      "grad_norm": 27.093080520629883,
      "learning_rate": 0.00019755684446106812,
      "loss": 2.8161096572875977,
      "step": 206
    },
    {
      "epoch": 0.09968097273219767,
      "grad_norm": 16.779788970947266,
      "learning_rate": 0.00019752245595217662,
      "loss": 2.9468119144439697,
      "step": 207
    },
    {
      "epoch": 0.10016252332510685,
      "grad_norm": 91.08209991455078,
      "learning_rate": 0.00019748783015005144,
      "loss": 2.51340913772583,
      "step": 208
    },
    {
      "epoch": 0.10064407391801601,
      "grad_norm": 7.542270660400391,
      "learning_rate": 0.00019745296713894465,
      "loss": 3.5934925079345703,
      "step": 209
    },
    {
      "epoch": 0.10112562451092517,
      "grad_norm": 9.214580535888672,
      "learning_rate": 0.00019741786700368548,
      "loss": 2.3333933353424072,
      "step": 210
    },
    {
      "epoch": 0.10160717510383435,
      "grad_norm": 3.918077230453491,
      "learning_rate": 0.00019738252982968017,
      "loss": 2.94449782371521,
      "step": 211
    },
    {
      "epoch": 0.10208872569674352,
      "grad_norm": 8.527355194091797,
      "learning_rate": 0.00019734695570291168,
      "loss": 2.5250601768493652,
      "step": 212
    },
    {
      "epoch": 0.10257027628965268,
      "grad_norm": 10.562358856201172,
      "learning_rate": 0.00019731114470993962,
      "loss": 3.038823127746582,
      "step": 213
    },
    {
      "epoch": 0.10305182688256186,
      "grad_norm": 3.0567641258239746,
      "learning_rate": 0.0001972750969378998,
      "loss": 3.3528175354003906,
      "step": 214
    },
    {
      "epoch": 0.10353337747547102,
      "grad_norm": 4.74027681350708,
      "learning_rate": 0.00019723881247450434,
      "loss": 3.477555274963379,
      "step": 215
    },
    {
      "epoch": 0.10401492806838018,
      "grad_norm": 162.38894653320312,
      "learning_rate": 0.0001972022914080411,
      "loss": 4.073236465454102,
      "step": 216
    },
    {
      "epoch": 0.10449647866128935,
      "grad_norm": 19.321529388427734,
      "learning_rate": 0.00019716553382737379,
      "loss": 3.479513168334961,
      "step": 217
    },
    {
      "epoch": 0.10497802925419852,
      "grad_norm": 49.24386215209961,
      "learning_rate": 0.00019712853982194152,
      "loss": 3.6474721431732178,
      "step": 218
    },
    {
      "epoch": 0.10545957984710769,
      "grad_norm": 12.652568817138672,
      "learning_rate": 0.00019709130948175876,
      "loss": 3.355501174926758,
      "step": 219
    },
    {
      "epoch": 0.10594113044001685,
      "grad_norm": 4.129899024963379,
      "learning_rate": 0.0001970538428974149,
      "loss": 3.0170657634735107,
      "step": 220
    },
    {
      "epoch": 0.10642268103292603,
      "grad_norm": 79.1859130859375,
      "learning_rate": 0.00019701614016007436,
      "loss": 2.3874804973602295,
      "step": 221
    },
    {
      "epoch": 0.10690423162583519,
      "grad_norm": 26.86741065979004,
      "learning_rate": 0.00019697820136147597,
      "loss": 3.0221829414367676,
      "step": 222
    },
    {
      "epoch": 0.10738578221874436,
      "grad_norm": 18.000368118286133,
      "learning_rate": 0.00019694002659393305,
      "loss": 3.1974399089813232,
      "step": 223
    },
    {
      "epoch": 0.10786733281165352,
      "grad_norm": 43.30815124511719,
      "learning_rate": 0.0001969016159503331,
      "loss": 2.6321449279785156,
      "step": 224
    },
    {
      "epoch": 0.1083488834045627,
      "grad_norm": 6.368330955505371,
      "learning_rate": 0.00019686296952413747,
      "loss": 2.897644519805908,
      "step": 225
    },
    {
      "epoch": 0.10883043399747186,
      "grad_norm": 8.925248146057129,
      "learning_rate": 0.0001968240874093813,
      "loss": 2.635138988494873,
      "step": 226
    },
    {
      "epoch": 0.10931198459038102,
      "grad_norm": 23.18450164794922,
      "learning_rate": 0.00019678496970067325,
      "loss": 3.1921279430389404,
      "step": 227
    },
    {
      "epoch": 0.1097935351832902,
      "grad_norm": 3.1420140266418457,
      "learning_rate": 0.0001967456164931951,
      "loss": 2.8326311111450195,
      "step": 228
    },
    {
      "epoch": 0.11027508577619936,
      "grad_norm": 14.200254440307617,
      "learning_rate": 0.0001967060278827017,
      "loss": 2.7949109077453613,
      "step": 229
    },
    {
      "epoch": 0.11075663636910853,
      "grad_norm": 4.551650524139404,
      "learning_rate": 0.00019666620396552076,
      "loss": 2.591254949569702,
      "step": 230
    },
    {
      "epoch": 0.11123818696201769,
      "grad_norm": 3.1689698696136475,
      "learning_rate": 0.00019662614483855246,
      "loss": 2.6821606159210205,
      "step": 231
    },
    {
      "epoch": 0.11171973755492687,
      "grad_norm": 46.83055114746094,
      "learning_rate": 0.00019658585059926934,
      "loss": 3.060485363006592,
      "step": 232
    },
    {
      "epoch": 0.11220128814783603,
      "grad_norm": 4.280196666717529,
      "learning_rate": 0.00019654532134571594,
      "loss": 3.199352264404297,
      "step": 233
    },
    {
      "epoch": 0.1126828387407452,
      "grad_norm": 13.214926719665527,
      "learning_rate": 0.00019650455717650878,
      "loss": 2.765612840652466,
      "step": 234
    },
    {
      "epoch": 0.11316438933365437,
      "grad_norm": 3.716822385787964,
      "learning_rate": 0.00019646355819083589,
      "loss": 2.44455623626709,
      "step": 235
    },
    {
      "epoch": 0.11364593992656354,
      "grad_norm": 6.5937371253967285,
      "learning_rate": 0.0001964223244884566,
      "loss": 2.847715139389038,
      "step": 236
    },
    {
      "epoch": 0.1141274905194727,
      "grad_norm": 4.867831707000732,
      "learning_rate": 0.00019638085616970153,
      "loss": 2.624288320541382,
      "step": 237
    },
    {
      "epoch": 0.11460904111238188,
      "grad_norm": 5.746015548706055,
      "learning_rate": 0.00019633915333547202,
      "loss": 3.2930123805999756,
      "step": 238
    },
    {
      "epoch": 0.11509059170529104,
      "grad_norm": 16.419414520263672,
      "learning_rate": 0.00019629721608724004,
      "loss": 3.2415406703948975,
      "step": 239
    },
    {
      "epoch": 0.1155721422982002,
      "grad_norm": 6.976315021514893,
      "learning_rate": 0.0001962550445270481,
      "loss": 2.8769233226776123,
      "step": 240
    },
    {
      "epoch": 0.11605369289110937,
      "grad_norm": 3.027923583984375,
      "learning_rate": 0.00019621263875750864,
      "loss": 2.871272563934326,
      "step": 241
    },
    {
      "epoch": 0.11653524348401854,
      "grad_norm": 18.4444580078125,
      "learning_rate": 0.00019616999888180406,
      "loss": 3.3500335216522217,
      "step": 242
    },
    {
      "epoch": 0.11701679407692771,
      "grad_norm": 6.278454780578613,
      "learning_rate": 0.0001961271250036865,
      "loss": 2.8858418464660645,
      "step": 243
    },
    {
      "epoch": 0.11749834466983687,
      "grad_norm": 9.351664543151855,
      "learning_rate": 0.0001960840172274773,
      "loss": 2.4538235664367676,
      "step": 244
    },
    {
      "epoch": 0.11797989526274605,
      "grad_norm": 6.907669544219971,
      "learning_rate": 0.00019604067565806704,
      "loss": 2.3783886432647705,
      "step": 245
    },
    {
      "epoch": 0.11846144585565521,
      "grad_norm": 16.38750457763672,
      "learning_rate": 0.00019599710040091512,
      "loss": 2.9476239681243896,
      "step": 246
    },
    {
      "epoch": 0.11894299644856438,
      "grad_norm": 9.283736228942871,
      "learning_rate": 0.00019595329156204955,
      "loss": 3.8901989459991455,
      "step": 247
    },
    {
      "epoch": 0.11942454704147354,
      "grad_norm": 1.9510459899902344,
      "learning_rate": 0.00019590924924806676,
      "loss": 2.999789237976074,
      "step": 248
    },
    {
      "epoch": 0.11990609763438272,
      "grad_norm": 47.557395935058594,
      "learning_rate": 0.0001958649735661312,
      "loss": 2.9103145599365234,
      "step": 249
    },
    {
      "epoch": 0.12038764822729188,
      "grad_norm": 46.886192321777344,
      "learning_rate": 0.00019582046462397515,
      "loss": 3.3303701877593994,
      "step": 250
    },
    {
      "epoch": 0.12086919882020104,
      "grad_norm": 31.490894317626953,
      "learning_rate": 0.00019577572252989854,
      "loss": 2.652620315551758,
      "step": 251
    },
    {
      "epoch": 0.12135074941311022,
      "grad_norm": 7.914570331573486,
      "learning_rate": 0.00019573074739276858,
      "loss": 2.698190450668335,
      "step": 252
    },
    {
      "epoch": 0.12183230000601938,
      "grad_norm": 2.9368252754211426,
      "learning_rate": 0.00019568553932201947,
      "loss": 2.2536261081695557,
      "step": 253
    },
    {
      "epoch": 0.12231385059892855,
      "grad_norm": 4.056997776031494,
      "learning_rate": 0.00019564009842765225,
      "loss": 3.303607225418091,
      "step": 254
    },
    {
      "epoch": 0.12279540119183771,
      "grad_norm": 5.543824195861816,
      "learning_rate": 0.00019559442482023444,
      "loss": 2.9682204723358154,
      "step": 255
    },
    {
      "epoch": 0.12327695178474689,
      "grad_norm": 7.216851711273193,
      "learning_rate": 0.0001955485186108998,
      "loss": 2.562302589416504,
      "step": 256
    },
    {
      "epoch": 0.12375850237765605,
      "grad_norm": 4.248110771179199,
      "learning_rate": 0.00019550237991134805,
      "loss": 2.4880027770996094,
      "step": 257
    },
    {
      "epoch": 0.12424005297056522,
      "grad_norm": 13.040796279907227,
      "learning_rate": 0.00019545600883384467,
      "loss": 2.7459726333618164,
      "step": 258
    },
    {
      "epoch": 0.12472160356347439,
      "grad_norm": 4.381429195404053,
      "learning_rate": 0.0001954094054912205,
      "loss": 1.795504093170166,
      "step": 259
    },
    {
      "epoch": 0.12520315415638356,
      "grad_norm": 13.650172233581543,
      "learning_rate": 0.00019536256999687157,
      "loss": 2.9786317348480225,
      "step": 260
    },
    {
      "epoch": 0.12568470474929272,
      "grad_norm": 12.095928192138672,
      "learning_rate": 0.00019531550246475876,
      "loss": 2.799271583557129,
      "step": 261
    },
    {
      "epoch": 0.12616625534220188,
      "grad_norm": 5.742032527923584,
      "learning_rate": 0.00019526820300940756,
      "loss": 2.663119316101074,
      "step": 262
    },
    {
      "epoch": 0.12664780593511105,
      "grad_norm": 24.73152732849121,
      "learning_rate": 0.00019522067174590778,
      "loss": 2.4349358081817627,
      "step": 263
    },
    {
      "epoch": 0.12712935652802024,
      "grad_norm": 8.374258041381836,
      "learning_rate": 0.00019517290878991324,
      "loss": 2.927595615386963,
      "step": 264
    },
    {
      "epoch": 0.1276109071209294,
      "grad_norm": 6.259795665740967,
      "learning_rate": 0.0001951249142576416,
      "loss": 2.5986835956573486,
      "step": 265
    },
    {
      "epoch": 0.12809245771383856,
      "grad_norm": 3.8742265701293945,
      "learning_rate": 0.00019507668826587387,
      "loss": 2.512293815612793,
      "step": 266
    },
    {
      "epoch": 0.12857400830674773,
      "grad_norm": 3.0486607551574707,
      "learning_rate": 0.0001950282309319544,
      "loss": 3.0053353309631348,
      "step": 267
    },
    {
      "epoch": 0.1290555588996569,
      "grad_norm": 11.721296310424805,
      "learning_rate": 0.0001949795423737903,
      "loss": 3.079239845275879,
      "step": 268
    },
    {
      "epoch": 0.12953710949256605,
      "grad_norm": 3.3867506980895996,
      "learning_rate": 0.00019493062270985144,
      "loss": 2.4749596118927,
      "step": 269
    },
    {
      "epoch": 0.13001866008547522,
      "grad_norm": 3.4137227535247803,
      "learning_rate": 0.00019488147205916985,
      "loss": 3.2239797115325928,
      "step": 270
    },
    {
      "epoch": 0.1305002106783844,
      "grad_norm": 2.4201009273529053,
      "learning_rate": 0.00019483209054133976,
      "loss": 2.7603635787963867,
      "step": 271
    },
    {
      "epoch": 0.13098176127129357,
      "grad_norm": 9.493799209594727,
      "learning_rate": 0.00019478247827651708,
      "loss": 2.333451271057129,
      "step": 272
    },
    {
      "epoch": 0.13146331186420274,
      "grad_norm": 5.731814384460449,
      "learning_rate": 0.00019473263538541914,
      "loss": 2.739171028137207,
      "step": 273
    },
    {
      "epoch": 0.1319448624571119,
      "grad_norm": 6.986753940582275,
      "learning_rate": 0.00019468256198932455,
      "loss": 3.3378994464874268,
      "step": 274
    },
    {
      "epoch": 0.13242641305002106,
      "grad_norm": 2.473236560821533,
      "learning_rate": 0.00019463225821007268,
      "loss": 2.0879902839660645,
      "step": 275
    },
    {
      "epoch": 0.13290796364293023,
      "grad_norm": 2.870792865753174,
      "learning_rate": 0.00019458172417006347,
      "loss": 3.026413917541504,
      "step": 276
    },
    {
      "epoch": 0.1333895142358394,
      "grad_norm": 2.6082699298858643,
      "learning_rate": 0.00019453095999225726,
      "loss": 3.6383886337280273,
      "step": 277
    },
    {
      "epoch": 0.13387106482874858,
      "grad_norm": 16.141149520874023,
      "learning_rate": 0.0001944799658001742,
      "loss": 2.5672647953033447,
      "step": 278
    },
    {
      "epoch": 0.13435261542165775,
      "grad_norm": 6.090396404266357,
      "learning_rate": 0.00019442874171789418,
      "loss": 2.522592067718506,
      "step": 279
    },
    {
      "epoch": 0.1348341660145669,
      "grad_norm": 3.3704872131347656,
      "learning_rate": 0.00019437728787005657,
      "loss": 2.949490785598755,
      "step": 280
    },
    {
      "epoch": 0.13531571660747607,
      "grad_norm": 4.79742431640625,
      "learning_rate": 0.00019432560438185963,
      "loss": 3.4105935096740723,
      "step": 281
    },
    {
      "epoch": 0.13579726720038524,
      "grad_norm": 3.2389824390411377,
      "learning_rate": 0.00019427369137906046,
      "loss": 2.574420690536499,
      "step": 282
    },
    {
      "epoch": 0.1362788177932944,
      "grad_norm": 24.452220916748047,
      "learning_rate": 0.00019422154898797472,
      "loss": 2.651268243789673,
      "step": 283
    },
    {
      "epoch": 0.13676036838620356,
      "grad_norm": 4.435457229614258,
      "learning_rate": 0.00019416917733547603,
      "loss": 2.868741750717163,
      "step": 284
    },
    {
      "epoch": 0.13724191897911275,
      "grad_norm": 27.461868286132812,
      "learning_rate": 0.00019411657654899597,
      "loss": 3.2383131980895996,
      "step": 285
    },
    {
      "epoch": 0.13772346957202192,
      "grad_norm": 30.364715576171875,
      "learning_rate": 0.0001940637467565237,
      "loss": 3.1953234672546387,
      "step": 286
    },
    {
      "epoch": 0.13820502016493108,
      "grad_norm": 3.948760747909546,
      "learning_rate": 0.00019401068808660546,
      "loss": 2.670407295227051,
      "step": 287
    },
    {
      "epoch": 0.13868657075784024,
      "grad_norm": 6.765519618988037,
      "learning_rate": 0.0001939574006683445,
      "loss": 2.683110475540161,
      "step": 288
    },
    {
      "epoch": 0.1391681213507494,
      "grad_norm": 6.715476036071777,
      "learning_rate": 0.00019390388463140065,
      "loss": 2.843869209289551,
      "step": 289
    },
    {
      "epoch": 0.13964967194365857,
      "grad_norm": 3.894876718521118,
      "learning_rate": 0.00019385014010598998,
      "loss": 3.1736552715301514,
      "step": 290
    },
    {
      "epoch": 0.14013122253656773,
      "grad_norm": 13.497404098510742,
      "learning_rate": 0.00019379616722288456,
      "loss": 2.3255698680877686,
      "step": 291
    },
    {
      "epoch": 0.14061277312947693,
      "grad_norm": 11.700185775756836,
      "learning_rate": 0.0001937419661134121,
      "loss": 2.4291906356811523,
      "step": 292
    },
    {
      "epoch": 0.1410943237223861,
      "grad_norm": 13.204001426696777,
      "learning_rate": 0.0001936875369094556,
      "loss": 2.680598497390747,
      "step": 293
    },
    {
      "epoch": 0.14157587431529525,
      "grad_norm": 5.325042247772217,
      "learning_rate": 0.0001936328797434531,
      "loss": 2.162095308303833,
      "step": 294
    },
    {
      "epoch": 0.14205742490820442,
      "grad_norm": 7.290260314941406,
      "learning_rate": 0.00019357799474839735,
      "loss": 2.917503833770752,
      "step": 295
    },
    {
      "epoch": 0.14253897550111358,
      "grad_norm": 6.807450294494629,
      "learning_rate": 0.00019352288205783536,
      "loss": 2.5774335861206055,
      "step": 296
    },
    {
      "epoch": 0.14302052609402274,
      "grad_norm": 55.764862060546875,
      "learning_rate": 0.00019346754180586825,
      "loss": 2.1671903133392334,
      "step": 297
    },
    {
      "epoch": 0.14350207668693193,
      "grad_norm": 3.0785512924194336,
      "learning_rate": 0.00019341197412715082,
      "loss": 2.85532546043396,
      "step": 298
    },
    {
      "epoch": 0.1439836272798411,
      "grad_norm": 220.6442108154297,
      "learning_rate": 0.00019335617915689128,
      "loss": 2.67545223236084,
      "step": 299
    },
    {
      "epoch": 0.14446517787275026,
      "grad_norm": 6.0753350257873535,
      "learning_rate": 0.00019330015703085082,
      "loss": 2.8890280723571777,
      "step": 300
    },
    {
      "epoch": 0.14494672846565942,
      "grad_norm": 16.938892364501953,
      "learning_rate": 0.00019324390788534343,
      "loss": 2.950892686843872,
      "step": 301
    },
    {
      "epoch": 0.1454282790585686,
      "grad_norm": 4.887892246246338,
      "learning_rate": 0.00019318743185723546,
      "loss": 2.563281774520874,
      "step": 302
    },
    {
      "epoch": 0.14590982965147775,
      "grad_norm": 11.46098804473877,
      "learning_rate": 0.00019313072908394525,
      "loss": 2.922501564025879,
      "step": 303
    },
    {
      "epoch": 0.14639138024438691,
      "grad_norm": 4.212432861328125,
      "learning_rate": 0.00019307379970344294,
      "loss": 2.2657411098480225,
      "step": 304
    },
    {
      "epoch": 0.1468729308372961,
      "grad_norm": 3.8126840591430664,
      "learning_rate": 0.00019301664385425004,
      "loss": 2.4331977367401123,
      "step": 305
    },
    {
      "epoch": 0.14735448143020527,
      "grad_norm": 7.910400390625,
      "learning_rate": 0.0001929592616754391,
      "loss": 2.5270073413848877,
      "step": 306
    },
    {
      "epoch": 0.14783603202311443,
      "grad_norm": 2.302321672439575,
      "learning_rate": 0.00019290165330663336,
      "loss": 2.6447625160217285,
      "step": 307
    },
    {
      "epoch": 0.1483175826160236,
      "grad_norm": 3.6254374980926514,
      "learning_rate": 0.00019284381888800647,
      "loss": 2.1547369956970215,
      "step": 308
    },
    {
      "epoch": 0.14879913320893276,
      "grad_norm": 5.169315338134766,
      "learning_rate": 0.00019278575856028206,
      "loss": 1.9625461101531982,
      "step": 309
    },
    {
      "epoch": 0.14928068380184192,
      "grad_norm": 4.284989833831787,
      "learning_rate": 0.00019272747246473345,
      "loss": 2.291788101196289,
      "step": 310
    },
    {
      "epoch": 0.1497622343947511,
      "grad_norm": 7.371423244476318,
      "learning_rate": 0.00019266896074318334,
      "loss": 2.913665533065796,
      "step": 311
    },
    {
      "epoch": 0.15024378498766028,
      "grad_norm": 4.0233635902404785,
      "learning_rate": 0.00019261022353800344,
      "loss": 2.411832094192505,
      "step": 312
    },
    {
      "epoch": 0.15072533558056944,
      "grad_norm": 2.4259145259857178,
      "learning_rate": 0.00019255126099211402,
      "loss": 2.539597749710083,
      "step": 313
    },
    {
      "epoch": 0.1512068861734786,
      "grad_norm": 2.771742105484009,
      "learning_rate": 0.00019249207324898376,
      "loss": 2.866787910461426,
      "step": 314
    },
    {
      "epoch": 0.15168843676638777,
      "grad_norm": 4.182138919830322,
      "learning_rate": 0.0001924326604526292,
      "loss": 2.457122325897217,
      "step": 315
    },
    {
      "epoch": 0.15216998735929693,
      "grad_norm": 2.0575344562530518,
      "learning_rate": 0.00019237302274761458,
      "loss": 2.4884510040283203,
      "step": 316
    },
    {
      "epoch": 0.1526515379522061,
      "grad_norm": 4.57686185836792,
      "learning_rate": 0.0001923131602790513,
      "loss": 3.385136127471924,
      "step": 317
    },
    {
      "epoch": 0.15313308854511526,
      "grad_norm": 5.123331546783447,
      "learning_rate": 0.00019225307319259768,
      "loss": 2.4651260375976562,
      "step": 318
    },
    {
      "epoch": 0.15361463913802445,
      "grad_norm": 4.2519450187683105,
      "learning_rate": 0.00019219276163445862,
      "loss": 2.298020839691162,
      "step": 319
    },
    {
      "epoch": 0.1540961897309336,
      "grad_norm": 1.8849855661392212,
      "learning_rate": 0.00019213222575138522,
      "loss": 2.439756155014038,
      "step": 320
    },
    {
      "epoch": 0.15457774032384278,
      "grad_norm": 4.573194980621338,
      "learning_rate": 0.00019207146569067435,
      "loss": 2.433350086212158,
      "step": 321
    },
    {
      "epoch": 0.15505929091675194,
      "grad_norm": 4.597135543823242,
      "learning_rate": 0.00019201048160016838,
      "loss": 2.442746162414551,
      "step": 322
    },
    {
      "epoch": 0.1555408415096611,
      "grad_norm": 2.119748830795288,
      "learning_rate": 0.00019194927362825478,
      "loss": 2.1048712730407715,
      "step": 323
    },
    {
      "epoch": 0.15602239210257027,
      "grad_norm": 6.786062717437744,
      "learning_rate": 0.00019188784192386587,
      "loss": 2.125335931777954,
      "step": 324
    },
    {
      "epoch": 0.15650394269547943,
      "grad_norm": 3.449805974960327,
      "learning_rate": 0.00019182618663647817,
      "loss": 2.6910743713378906,
      "step": 325
    },
    {
      "epoch": 0.15698549328838862,
      "grad_norm": 9.522765159606934,
      "learning_rate": 0.0001917643079161124,
      "loss": 2.3984594345092773,
      "step": 326
    },
    {
      "epoch": 0.15746704388129779,
      "grad_norm": 4.611312389373779,
      "learning_rate": 0.00019170220591333283,
      "loss": 3.064091682434082,
      "step": 327
    },
    {
      "epoch": 0.15794859447420695,
      "grad_norm": 3.646578550338745,
      "learning_rate": 0.00019163988077924713,
      "loss": 2.4258337020874023,
      "step": 328
    },
    {
      "epoch": 0.1584301450671161,
      "grad_norm": 2.328768491744995,
      "learning_rate": 0.00019157733266550575,
      "loss": 2.39370059967041,
      "step": 329
    },
    {
      "epoch": 0.15891169566002528,
      "grad_norm": 10.541135787963867,
      "learning_rate": 0.00019151456172430183,
      "loss": 2.608804702758789,
      "step": 330
    },
    {
      "epoch": 0.15939324625293444,
      "grad_norm": 2.0800814628601074,
      "learning_rate": 0.0001914515681083707,
      "loss": 2.5314502716064453,
      "step": 331
    },
    {
      "epoch": 0.1598747968458436,
      "grad_norm": 2.676462173461914,
      "learning_rate": 0.00019138835197098937,
      "loss": 2.7351229190826416,
      "step": 332
    },
    {
      "epoch": 0.1603563474387528,
      "grad_norm": 10.314946174621582,
      "learning_rate": 0.00019132491346597643,
      "loss": 2.4312338829040527,
      "step": 333
    },
    {
      "epoch": 0.16083789803166196,
      "grad_norm": 14.583769798278809,
      "learning_rate": 0.00019126125274769145,
      "loss": 2.770493745803833,
      "step": 334
    },
    {
      "epoch": 0.16131944862457112,
      "grad_norm": 3.3250508308410645,
      "learning_rate": 0.00019119736997103476,
      "loss": 2.2206521034240723,
      "step": 335
    },
    {
      "epoch": 0.16180099921748028,
      "grad_norm": 4.549742221832275,
      "learning_rate": 0.000191133265291447,
      "loss": 1.8336033821105957,
      "step": 336
    },
    {
      "epoch": 0.16228254981038945,
      "grad_norm": 5.639762878417969,
      "learning_rate": 0.00019106893886490864,
      "loss": 2.9891672134399414,
      "step": 337
    },
    {
      "epoch": 0.1627641004032986,
      "grad_norm": 18.725431442260742,
      "learning_rate": 0.00019100439084793989,
      "loss": 2.00811505317688,
      "step": 338
    },
    {
      "epoch": 0.1632456509962078,
      "grad_norm": 5.564981460571289,
      "learning_rate": 0.00019093962139759998,
      "loss": 3.356139659881592,
      "step": 339
    },
    {
      "epoch": 0.16372720158911697,
      "grad_norm": 2.209601640701294,
      "learning_rate": 0.000190874630671487,
      "loss": 2.6676928997039795,
      "step": 340
    },
    {
      "epoch": 0.16420875218202613,
      "grad_norm": 3.3721823692321777,
      "learning_rate": 0.00019080941882773745,
      "loss": 2.5202577114105225,
      "step": 341
    },
    {
      "epoch": 0.1646903027749353,
      "grad_norm": 5.715636730194092,
      "learning_rate": 0.00019074398602502584,
      "loss": 3.0192487239837646,
      "step": 342
    },
    {
      "epoch": 0.16517185336784446,
      "grad_norm": 7.711333274841309,
      "learning_rate": 0.00019067833242256442,
      "loss": 2.9520139694213867,
      "step": 343
    },
    {
      "epoch": 0.16565340396075362,
      "grad_norm": 4.124719142913818,
      "learning_rate": 0.0001906124581801025,
      "loss": 2.1605372428894043,
      "step": 344
    },
    {
      "epoch": 0.16613495455366278,
      "grad_norm": 2.909250259399414,
      "learning_rate": 0.0001905463634579264,
      "loss": 2.0177078247070312,
      "step": 345
    },
    {
      "epoch": 0.16661650514657197,
      "grad_norm": 4.370840549468994,
      "learning_rate": 0.00019048004841685888,
      "loss": 2.249248504638672,
      "step": 346
    },
    {
      "epoch": 0.16709805573948114,
      "grad_norm": 4.269500732421875,
      "learning_rate": 0.00019041351321825883,
      "loss": 2.742091178894043,
      "step": 347
    },
    {
      "epoch": 0.1675796063323903,
      "grad_norm": 9.316685676574707,
      "learning_rate": 0.00019034675802402068,
      "loss": 2.349862575531006,
      "step": 348
    },
    {
      "epoch": 0.16806115692529947,
      "grad_norm": 8.709773063659668,
      "learning_rate": 0.00019027978299657436,
      "loss": 3.2503225803375244,
      "step": 349
    },
    {
      "epoch": 0.16854270751820863,
      "grad_norm": 9.557984352111816,
      "learning_rate": 0.00019021258829888456,
      "loss": 2.3776354789733887,
      "step": 350
    },
    {
      "epoch": 0.1690242581111178,
      "grad_norm": 12.156946182250977,
      "learning_rate": 0.00019014517409445052,
      "loss": 2.9352223873138428,
      "step": 351
    },
    {
      "epoch": 0.16950580870402696,
      "grad_norm": 3.726409912109375,
      "learning_rate": 0.00019007754054730554,
      "loss": 2.523411750793457,
      "step": 352
    },
    {
      "epoch": 0.16998735929693615,
      "grad_norm": 63.2518310546875,
      "learning_rate": 0.00019000968782201675,
      "loss": 2.3083438873291016,
      "step": 353
    },
    {
      "epoch": 0.1704689098898453,
      "grad_norm": 12.749381065368652,
      "learning_rate": 0.00018994161608368448,
      "loss": 2.552077054977417,
      "step": 354
    },
    {
      "epoch": 0.17095046048275447,
      "grad_norm": 6.490101337432861,
      "learning_rate": 0.00018987332549794196,
      "loss": 2.103773593902588,
      "step": 355
    },
    {
      "epoch": 0.17143201107566364,
      "grad_norm": 4.977334499359131,
      "learning_rate": 0.00018980481623095502,
      "loss": 2.4962143898010254,
      "step": 356
    },
    {
      "epoch": 0.1719135616685728,
      "grad_norm": 5.521564960479736,
      "learning_rate": 0.00018973608844942148,
      "loss": 2.1136443614959717,
      "step": 357
    },
    {
      "epoch": 0.17239511226148196,
      "grad_norm": 21.19316864013672,
      "learning_rate": 0.00018966714232057094,
      "loss": 2.514810800552368,
      "step": 358
    },
    {
      "epoch": 0.17287666285439113,
      "grad_norm": 5.176145076751709,
      "learning_rate": 0.00018959797801216418,
      "loss": 2.0549521446228027,
      "step": 359
    },
    {
      "epoch": 0.17335821344730032,
      "grad_norm": 21.51235008239746,
      "learning_rate": 0.000189528595692493,
      "loss": 2.580915689468384,
      "step": 360
    },
    {
      "epoch": 0.17383976404020948,
      "grad_norm": 40.23230743408203,
      "learning_rate": 0.00018945899553037956,
      "loss": 1.7977638244628906,
      "step": 361
    },
    {
      "epoch": 0.17432131463311865,
      "grad_norm": 7.59974479675293,
      "learning_rate": 0.00018938917769517613,
      "loss": 1.9560928344726562,
      "step": 362
    },
    {
      "epoch": 0.1748028652260278,
      "grad_norm": 15.393118858337402,
      "learning_rate": 0.00018931914235676458,
      "loss": 2.93941068649292,
      "step": 363
    },
    {
      "epoch": 0.17528441581893697,
      "grad_norm": 23.539203643798828,
      "learning_rate": 0.00018924888968555606,
      "loss": 2.5802347660064697,
      "step": 364
    },
    {
      "epoch": 0.17576596641184614,
      "grad_norm": 4.364682674407959,
      "learning_rate": 0.00018917841985249055,
      "loss": 2.6234960556030273,
      "step": 365
    },
    {
      "epoch": 0.1762475170047553,
      "grad_norm": 8.979281425476074,
      "learning_rate": 0.0001891077330290363,
      "loss": 3.4569814205169678,
      "step": 366
    },
    {
      "epoch": 0.1767290675976645,
      "grad_norm": 28.81070327758789,
      "learning_rate": 0.00018903682938718977,
      "loss": 2.709643840789795,
      "step": 367
    },
    {
      "epoch": 0.17721061819057365,
      "grad_norm": 7.868295669555664,
      "learning_rate": 0.00018896570909947475,
      "loss": 1.9880595207214355,
      "step": 368
    },
    {
      "epoch": 0.17769216878348282,
      "grad_norm": 4.864500522613525,
      "learning_rate": 0.00018889437233894234,
      "loss": 1.9182296991348267,
      "step": 369
    },
    {
      "epoch": 0.17817371937639198,
      "grad_norm": 13.386634826660156,
      "learning_rate": 0.0001888228192791703,
      "loss": 2.8505048751831055,
      "step": 370
    },
    {
      "epoch": 0.17865526996930114,
      "grad_norm": 3.9562807083129883,
      "learning_rate": 0.00018875105009426272,
      "loss": 2.8039145469665527,
      "step": 371
    },
    {
      "epoch": 0.1791368205622103,
      "grad_norm": 5.183873176574707,
      "learning_rate": 0.00018867906495884955,
      "loss": 2.2527785301208496,
      "step": 372
    },
    {
      "epoch": 0.17961837115511947,
      "grad_norm": 5.800826549530029,
      "learning_rate": 0.0001886068640480862,
      "loss": 3.047297477722168,
      "step": 373
    },
    {
      "epoch": 0.18009992174802866,
      "grad_norm": 10.969173431396484,
      "learning_rate": 0.00018853444753765306,
      "loss": 2.2293312549591064,
      "step": 374
    },
    {
      "epoch": 0.18058147234093783,
      "grad_norm": 17.965259552001953,
      "learning_rate": 0.00018846181560375525,
      "loss": 2.371647596359253,
      "step": 375
    },
    {
      "epoch": 0.181063022933847,
      "grad_norm": 11.411270141601562,
      "learning_rate": 0.0001883889684231219,
      "loss": 2.619664192199707,
      "step": 376
    },
    {
      "epoch": 0.18154457352675615,
      "grad_norm": 3.153895378112793,
      "learning_rate": 0.000188315906173006,
      "loss": 2.3961548805236816,
      "step": 377
    },
    {
      "epoch": 0.18202612411966532,
      "grad_norm": 6.123383522033691,
      "learning_rate": 0.0001882426290311838,
      "loss": 2.495826482772827,
      "step": 378
    },
    {
      "epoch": 0.18250767471257448,
      "grad_norm": 3.089853525161743,
      "learning_rate": 0.00018816913717595445,
      "loss": 2.325589418411255,
      "step": 379
    },
    {
      "epoch": 0.18298922530548364,
      "grad_norm": 13.272289276123047,
      "learning_rate": 0.00018809543078613953,
      "loss": 2.7516653537750244,
      "step": 380
    },
    {
      "epoch": 0.18347077589839283,
      "grad_norm": 18.73435401916504,
      "learning_rate": 0.00018802151004108263,
      "loss": 2.674243450164795,
      "step": 381
    },
    {
      "epoch": 0.183952326491302,
      "grad_norm": 2.825875997543335,
      "learning_rate": 0.0001879473751206489,
      "loss": 2.1949684619903564,
      "step": 382
    },
    {
      "epoch": 0.18443387708421116,
      "grad_norm": 4.46737813949585,
      "learning_rate": 0.00018787302620522467,
      "loss": 2.1944894790649414,
      "step": 383
    },
    {
      "epoch": 0.18491542767712033,
      "grad_norm": 22.417884826660156,
      "learning_rate": 0.00018779846347571693,
      "loss": 2.626586437225342,
      "step": 384
    },
    {
      "epoch": 0.1853969782700295,
      "grad_norm": 59.531646728515625,
      "learning_rate": 0.0001877236871135529,
      "loss": 2.341496706008911,
      "step": 385
    },
    {
      "epoch": 0.18587852886293865,
      "grad_norm": 7.0349650382995605,
      "learning_rate": 0.00018764869730067968,
      "loss": 2.2620255947113037,
      "step": 386
    },
    {
      "epoch": 0.18636007945584784,
      "grad_norm": 21.5872859954834,
      "learning_rate": 0.0001875734942195637,
      "loss": 2.4649035930633545,
      "step": 387
    },
    {
      "epoch": 0.186841630048757,
      "grad_norm": 67.29732513427734,
      "learning_rate": 0.0001874980780531903,
      "loss": 2.204517364501953,
      "step": 388
    },
    {
      "epoch": 0.18732318064166617,
      "grad_norm": 15.584892272949219,
      "learning_rate": 0.00018742244898506337,
      "loss": 3.06508469581604,
      "step": 389
    },
    {
      "epoch": 0.18780473123457533,
      "grad_norm": 17.616104125976562,
      "learning_rate": 0.00018734660719920475,
      "loss": 3.2268595695495605,
      "step": 390
    },
    {
      "epoch": 0.1882862818274845,
      "grad_norm": 11.898457527160645,
      "learning_rate": 0.00018727055288015397,
      "loss": 2.874145984649658,
      "step": 391
    },
    {
      "epoch": 0.18876783242039366,
      "grad_norm": 7.668893814086914,
      "learning_rate": 0.00018719428621296764,
      "loss": 1.7094461917877197,
      "step": 392
    },
    {
      "epoch": 0.18924938301330282,
      "grad_norm": 20.069002151489258,
      "learning_rate": 0.00018711780738321897,
      "loss": 2.360109329223633,
      "step": 393
    },
    {
      "epoch": 0.18973093360621202,
      "grad_norm": 15.45837688446045,
      "learning_rate": 0.00018704111657699758,
      "loss": 2.5932276248931885,
      "step": 394
    },
    {
      "epoch": 0.19021248419912118,
      "grad_norm": 19.858739852905273,
      "learning_rate": 0.0001869642139809088,
      "loss": 1.829392433166504,
      "step": 395
    },
    {
      "epoch": 0.19069403479203034,
      "grad_norm": 25.601045608520508,
      "learning_rate": 0.00018688709978207323,
      "loss": 2.500279664993286,
      "step": 396
    },
    {
      "epoch": 0.1911755853849395,
      "grad_norm": 125.8919906616211,
      "learning_rate": 0.00018680977416812644,
      "loss": 2.6939945220947266,
      "step": 397
    },
    {
      "epoch": 0.19165713597784867,
      "grad_norm": 12.126943588256836,
      "learning_rate": 0.00018673223732721837,
      "loss": 2.4771299362182617,
      "step": 398
    },
    {
      "epoch": 0.19213868657075783,
      "grad_norm": 43.97914505004883,
      "learning_rate": 0.0001866544894480129,
      "loss": 2.8298113346099854,
      "step": 399
    },
    {
      "epoch": 0.192620237163667,
      "grad_norm": 34.70051574707031,
      "learning_rate": 0.00018657653071968747,
      "loss": 2.167696475982666,
      "step": 400
    },
    {
      "epoch": 0.1931017877565762,
      "grad_norm": 37.10820007324219,
      "learning_rate": 0.00018649836133193253,
      "loss": 2.5001420974731445,
      "step": 401
    },
    {
      "epoch": 0.19358333834948535,
      "grad_norm": 92.15216827392578,
      "learning_rate": 0.00018641998147495112,
      "loss": 2.3628041744232178,
      "step": 402
    },
    {
      "epoch": 0.19406488894239451,
      "grad_norm": 18.841495513916016,
      "learning_rate": 0.00018634139133945837,
      "loss": 3.1646199226379395,
      "step": 403
    },
    {
      "epoch": 0.19454643953530368,
      "grad_norm": 10.359599113464355,
      "learning_rate": 0.00018626259111668105,
      "loss": 2.8942995071411133,
      "step": 404
    },
    {
      "epoch": 0.19502799012821284,
      "grad_norm": 19.581804275512695,
      "learning_rate": 0.00018618358099835723,
      "loss": 2.4930965900421143,
      "step": 405
    },
    {
      "epoch": 0.195509540721122,
      "grad_norm": 18.482358932495117,
      "learning_rate": 0.00018610436117673555,
      "loss": 1.7272776365280151,
      "step": 406
    },
    {
      "epoch": 0.19599109131403117,
      "grad_norm": 17.930448532104492,
      "learning_rate": 0.00018602493184457505,
      "loss": 1.9463963508605957,
      "step": 407
    },
    {
      "epoch": 0.19647264190694036,
      "grad_norm": 23.29218101501465,
      "learning_rate": 0.00018594529319514437,
      "loss": 2.4781172275543213,
      "step": 408
    },
    {
      "epoch": 0.19695419249984952,
      "grad_norm": 16.800519943237305,
      "learning_rate": 0.00018586544542222169,
      "loss": 1.9698387384414673,
      "step": 409
    },
    {
      "epoch": 0.1974357430927587,
      "grad_norm": 50.02911376953125,
      "learning_rate": 0.00018578538872009384,
      "loss": 3.2067863941192627,
      "step": 410
    },
    {
      "epoch": 0.19791729368566785,
      "grad_norm": 69.8697738647461,
      "learning_rate": 0.00018570512328355612,
      "loss": 2.638000965118408,
      "step": 411
    },
    {
      "epoch": 0.198398844278577,
      "grad_norm": 8.120281219482422,
      "learning_rate": 0.00018562464930791167,
      "loss": 2.573612928390503,
      "step": 412
    },
    {
      "epoch": 0.19888039487148618,
      "grad_norm": 22.212663650512695,
      "learning_rate": 0.00018554396698897116,
      "loss": 2.8517861366271973,
      "step": 413
    },
    {
      "epoch": 0.19936194546439534,
      "grad_norm": 5.5182342529296875,
      "learning_rate": 0.00018546307652305205,
      "loss": 2.0512664318084717,
      "step": 414
    },
    {
      "epoch": 0.19984349605730453,
      "grad_norm": 44.178955078125,
      "learning_rate": 0.00018538197810697842,
      "loss": 2.040891170501709,
      "step": 415
    },
    {
      "epoch": 0.2003250466502137,
      "grad_norm": 21.32073402404785,
      "learning_rate": 0.0001853006719380802,
      "loss": 2.519852638244629,
      "step": 416
    },
    {
      "epoch": 0.20080659724312286,
      "grad_norm": 7.96131706237793,
      "learning_rate": 0.00018521915821419284,
      "loss": 2.7485251426696777,
      "step": 417
    },
    {
      "epoch": 0.20128814783603202,
      "grad_norm": 43.572052001953125,
      "learning_rate": 0.00018513743713365698,
      "loss": 1.5083531141281128,
      "step": 418
    },
    {
      "epoch": 0.20176969842894119,
      "grad_norm": 28.865251541137695,
      "learning_rate": 0.00018505550889531765,
      "loss": 2.6472058296203613,
      "step": 419
    },
    {
      "epoch": 0.20225124902185035,
      "grad_norm": 6.294801235198975,
      "learning_rate": 0.00018497337369852395,
      "loss": 2.4348416328430176,
      "step": 420
    },
    {
      "epoch": 0.2027327996147595,
      "grad_norm": 12.776725769042969,
      "learning_rate": 0.0001848910317431286,
      "loss": 2.643418312072754,
      "step": 421
    },
    {
      "epoch": 0.2032143502076687,
      "grad_norm": 15.06001091003418,
      "learning_rate": 0.00018480848322948739,
      "loss": 2.379528760910034,
      "step": 422
    },
    {
      "epoch": 0.20369590080057787,
      "grad_norm": 11.404500961303711,
      "learning_rate": 0.00018472572835845873,
      "loss": 2.478358507156372,
      "step": 423
    },
    {
      "epoch": 0.20417745139348703,
      "grad_norm": 63.05402755737305,
      "learning_rate": 0.00018464276733140306,
      "loss": 2.638596773147583,
      "step": 424
    },
    {
      "epoch": 0.2046590019863962,
      "grad_norm": 30.686798095703125,
      "learning_rate": 0.0001845596003501826,
      "loss": 2.1478986740112305,
      "step": 425
    },
    {
      "epoch": 0.20514055257930536,
      "grad_norm": 25.790180206298828,
      "learning_rate": 0.00018447622761716057,
      "loss": 2.529923677444458,
      "step": 426
    },
    {
      "epoch": 0.20562210317221452,
      "grad_norm": 11.921496391296387,
      "learning_rate": 0.00018439264933520084,
      "loss": 2.4245188236236572,
      "step": 427
    },
    {
      "epoch": 0.2061036537651237,
      "grad_norm": 24.87253761291504,
      "learning_rate": 0.00018430886570766747,
      "loss": 2.8019957542419434,
      "step": 428
    },
    {
      "epoch": 0.20658520435803288,
      "grad_norm": 21.60533905029297,
      "learning_rate": 0.0001842248769384242,
      "loss": 1.7556593418121338,
      "step": 429
    },
    {
      "epoch": 0.20706675495094204,
      "grad_norm": 20.475053787231445,
      "learning_rate": 0.00018414068323183375,
      "loss": 2.2165701389312744,
      "step": 430
    },
    {
      "epoch": 0.2075483055438512,
      "grad_norm": 5.949760913848877,
      "learning_rate": 0.00018405628479275775,
      "loss": 2.1364946365356445,
      "step": 431
    },
    {
      "epoch": 0.20802985613676037,
      "grad_norm": 18.33285903930664,
      "learning_rate": 0.00018397168182655583,
      "loss": 3.267301082611084,
      "step": 432
    },
    {
      "epoch": 0.20851140672966953,
      "grad_norm": 7.811692714691162,
      "learning_rate": 0.00018388687453908527,
      "loss": 2.9456169605255127,
      "step": 433
    },
    {
      "epoch": 0.2089929573225787,
      "grad_norm": 30.292306900024414,
      "learning_rate": 0.00018380186313670058,
      "loss": 2.575460433959961,
      "step": 434
    },
    {
      "epoch": 0.20947450791548788,
      "grad_norm": 9.16192626953125,
      "learning_rate": 0.00018371664782625287,
      "loss": 1.9610531330108643,
      "step": 435
    },
    {
      "epoch": 0.20995605850839705,
      "grad_norm": 10.527608871459961,
      "learning_rate": 0.00018363122881508945,
      "loss": 1.7986665964126587,
      "step": 436
    },
    {
      "epoch": 0.2104376091013062,
      "grad_norm": 8.544146537780762,
      "learning_rate": 0.00018354560631105328,
      "loss": 2.9639439582824707,
      "step": 437
    },
    {
      "epoch": 0.21091915969421537,
      "grad_norm": 6.408632755279541,
      "learning_rate": 0.00018345978052248233,
      "loss": 2.4588840007781982,
      "step": 438
    },
    {
      "epoch": 0.21140071028712454,
      "grad_norm": 12.00464916229248,
      "learning_rate": 0.00018337375165820944,
      "loss": 1.77968168258667,
      "step": 439
    },
    {
      "epoch": 0.2118822608800337,
      "grad_norm": 23.762874603271484,
      "learning_rate": 0.00018328751992756137,
      "loss": 2.3309717178344727,
      "step": 440
    },
    {
      "epoch": 0.21236381147294286,
      "grad_norm": 11.160380363464355,
      "learning_rate": 0.0001832010855403586,
      "loss": 2.41969895362854,
      "step": 441
    },
    {
      "epoch": 0.21284536206585206,
      "grad_norm": 11.072091102600098,
      "learning_rate": 0.0001831144487069147,
      "loss": 2.800793409347534,
      "step": 442
    },
    {
      "epoch": 0.21332691265876122,
      "grad_norm": 20.44979476928711,
      "learning_rate": 0.0001830276096380358,
      "loss": 2.4336583614349365,
      "step": 443
    },
    {
      "epoch": 0.21380846325167038,
      "grad_norm": 27.052642822265625,
      "learning_rate": 0.0001829405685450202,
      "loss": 2.5551092624664307,
      "step": 444
    },
    {
      "epoch": 0.21429001384457955,
      "grad_norm": 11.80774974822998,
      "learning_rate": 0.00018285332563965765,
      "loss": 3.1216959953308105,
      "step": 445
    },
    {
      "epoch": 0.2147715644374887,
      "grad_norm": 464.5399169921875,
      "learning_rate": 0.00018276588113422905,
      "loss": 2.4834911823272705,
      "step": 446
    },
    {
      "epoch": 0.21525311503039787,
      "grad_norm": 145.11016845703125,
      "learning_rate": 0.00018267823524150575,
      "loss": 3.0340285301208496,
      "step": 447
    },
    {
      "epoch": 0.21573466562330704,
      "grad_norm": 10.452465057373047,
      "learning_rate": 0.00018259038817474923,
      "loss": 1.6671522855758667,
      "step": 448
    },
    {
      "epoch": 0.21621621621621623,
      "grad_norm": 106.97540283203125,
      "learning_rate": 0.0001825023401477104,
      "loss": 1.9741376638412476,
      "step": 449
    },
    {
      "epoch": 0.2166977668091254,
      "grad_norm": 1015.2591552734375,
      "learning_rate": 0.0001824140913746291,
      "loss": 2.4812142848968506,
      "step": 450
    },
    {
      "epoch": 0.21717931740203456,
      "grad_norm": 10.576818466186523,
      "learning_rate": 0.00018232564207023376,
      "loss": 2.422788619995117,
      "step": 451
    },
    {
      "epoch": 0.21766086799494372,
      "grad_norm": 12.816866874694824,
      "learning_rate": 0.00018223699244974064,
      "loss": 2.5695900917053223,
      "step": 452
    },
    {
      "epoch": 0.21814241858785288,
      "grad_norm": 30.523181915283203,
      "learning_rate": 0.00018214814272885343,
      "loss": 2.6562814712524414,
      "step": 453
    },
    {
      "epoch": 0.21862396918076205,
      "grad_norm": 32.75181579589844,
      "learning_rate": 0.00018205909312376276,
      "loss": 2.0012919902801514,
      "step": 454
    },
    {
      "epoch": 0.2191055197736712,
      "grad_norm": 289.2850341796875,
      "learning_rate": 0.00018196984385114554,
      "loss": 2.100579023361206,
      "step": 455
    },
    {
      "epoch": 0.2195870703665804,
      "grad_norm": 13.460122108459473,
      "learning_rate": 0.0001818803951281646,
      "loss": 2.7729341983795166,
      "step": 456
    },
    {
      "epoch": 0.22006862095948956,
      "grad_norm": 22.028789520263672,
      "learning_rate": 0.000181790747172468,
      "loss": 2.6116740703582764,
      "step": 457
    },
    {
      "epoch": 0.22055017155239873,
      "grad_norm": 82.16522979736328,
      "learning_rate": 0.00018170090020218864,
      "loss": 2.8228654861450195,
      "step": 458
    },
    {
      "epoch": 0.2210317221453079,
      "grad_norm": 30.354442596435547,
      "learning_rate": 0.00018161085443594365,
      "loss": 2.439753532409668,
      "step": 459
    },
    {
      "epoch": 0.22151327273821705,
      "grad_norm": 19.545732498168945,
      "learning_rate": 0.00018152061009283382,
      "loss": 3.067850351333618,
      "step": 460
    },
    {
      "epoch": 0.22199482333112622,
      "grad_norm": 27.8441219329834,
      "learning_rate": 0.00018143016739244314,
      "loss": 2.4648938179016113,
      "step": 461
    },
    {
      "epoch": 0.22247637392403538,
      "grad_norm": 27.314176559448242,
      "learning_rate": 0.0001813395265548383,
      "loss": 2.955810070037842,
      "step": 462
    },
    {
      "epoch": 0.22295792451694457,
      "grad_norm": 7.9851274490356445,
      "learning_rate": 0.00018124868780056814,
      "loss": 2.4873878955841064,
      "step": 463
    },
    {
      "epoch": 0.22343947510985374,
      "grad_norm": 44.27796173095703,
      "learning_rate": 0.0001811576513506629,
      "loss": 2.8058998584747314,
      "step": 464
    },
    {
      "epoch": 0.2239210257027629,
      "grad_norm": 34.31309509277344,
      "learning_rate": 0.00018106641742663397,
      "loss": 2.593855619430542,
      "step": 465
    },
    {
      "epoch": 0.22440257629567206,
      "grad_norm": 41.64469909667969,
      "learning_rate": 0.00018097498625047328,
      "loss": 2.3848559856414795,
      "step": 466
    },
    {
      "epoch": 0.22488412688858123,
      "grad_norm": 172.3015899658203,
      "learning_rate": 0.00018088335804465258,
      "loss": 2.729912757873535,
      "step": 467
    },
    {
      "epoch": 0.2253656774814904,
      "grad_norm": 72.72811126708984,
      "learning_rate": 0.00018079153303212318,
      "loss": 3.0460588932037354,
      "step": 468
    },
    {
      "epoch": 0.22584722807439955,
      "grad_norm": 7.656332492828369,
      "learning_rate": 0.0001806995114363152,
      "loss": 2.5046517848968506,
      "step": 469
    },
    {
      "epoch": 0.22632877866730874,
      "grad_norm": 86.75299835205078,
      "learning_rate": 0.00018060729348113707,
      "loss": 2.621473550796509,
      "step": 470
    },
    {
      "epoch": 0.2268103292602179,
      "grad_norm": 15.587586402893066,
      "learning_rate": 0.00018051487939097505,
      "loss": 2.0239064693450928,
      "step": 471
    },
    {
      "epoch": 0.22729187985312707,
      "grad_norm": 38.15410232543945,
      "learning_rate": 0.00018042226939069255,
      "loss": 2.1625752449035645,
      "step": 472
    },
    {
      "epoch": 0.22777343044603623,
      "grad_norm": 58.55221939086914,
      "learning_rate": 0.00018032946370562982,
      "loss": 2.9663166999816895,
      "step": 473
    },
    {
      "epoch": 0.2282549810389454,
      "grad_norm": 58.243927001953125,
      "learning_rate": 0.00018023646256160313,
      "loss": 2.2701830863952637,
      "step": 474
    },
    {
      "epoch": 0.22873653163185456,
      "grad_norm": 114.11939239501953,
      "learning_rate": 0.00018014326618490437,
      "loss": 2.250131845474243,
      "step": 475
    },
    {
      "epoch": 0.22921808222476375,
      "grad_norm": 19.021345138549805,
      "learning_rate": 0.0001800498748023005,
      "loss": 2.039703845977783,
      "step": 476
    },
    {
      "epoch": 0.22969963281767292,
      "grad_norm": 30.308340072631836,
      "learning_rate": 0.000179956288641033,
      "loss": 1.948765516281128,
      "step": 477
    },
    {
      "epoch": 0.23018118341058208,
      "grad_norm": 28.789335250854492,
      "learning_rate": 0.00017986250792881718,
      "loss": 2.888338327407837,
      "step": 478
    },
    {
      "epoch": 0.23066273400349124,
      "grad_norm": 35.87238311767578,
      "learning_rate": 0.00017976853289384184,
      "loss": 2.1339163780212402,
      "step": 479
    },
    {
      "epoch": 0.2311442845964004,
      "grad_norm": 21.675933837890625,
      "learning_rate": 0.00017967436376476855,
      "loss": 2.7641005516052246,
      "step": 480
    },
    {
      "epoch": 0.23162583518930957,
      "grad_norm": 14.848753929138184,
      "learning_rate": 0.0001795800007707312,
      "loss": 2.6576874256134033,
      "step": 481
    },
    {
      "epoch": 0.23210738578221873,
      "grad_norm": 6.806184768676758,
      "learning_rate": 0.00017948544414133534,
      "loss": 2.2545042037963867,
      "step": 482
    },
    {
      "epoch": 0.23258893637512792,
      "grad_norm": 18.57518196105957,
      "learning_rate": 0.00017939069410665773,
      "loss": 2.9779160022735596,
      "step": 483
    },
    {
      "epoch": 0.2330704869680371,
      "grad_norm": 11.426862716674805,
      "learning_rate": 0.0001792957508972457,
      "loss": 2.286898374557495,
      "step": 484
    },
    {
      "epoch": 0.23355203756094625,
      "grad_norm": 449.3490905761719,
      "learning_rate": 0.00017920061474411658,
      "loss": 3.4221880435943604,
      "step": 485
    },
    {
      "epoch": 0.23403358815385542,
      "grad_norm": 25.741392135620117,
      "learning_rate": 0.00017910528587875729,
      "loss": 2.628464937210083,
      "step": 486
    },
    {
      "epoch": 0.23451513874676458,
      "grad_norm": 34.56541442871094,
      "learning_rate": 0.00017900976453312352,
      "loss": 2.2378792762756348,
      "step": 487
    },
    {
      "epoch": 0.23499668933967374,
      "grad_norm": 55.960453033447266,
      "learning_rate": 0.00017891405093963938,
      "loss": 2.3625731468200684,
      "step": 488
    },
    {
      "epoch": 0.2354782399325829,
      "grad_norm": 62.46953582763672,
      "learning_rate": 0.00017881814533119675,
      "loss": 2.578016757965088,
      "step": 489
    },
    {
      "epoch": 0.2359597905254921,
      "grad_norm": 277.2417907714844,
      "learning_rate": 0.00017872204794115474,
      "loss": 2.07771635055542,
      "step": 490
    },
    {
      "epoch": 0.23644134111840126,
      "grad_norm": 49.079471588134766,
      "learning_rate": 0.0001786257590033391,
      "loss": 2.237464427947998,
      "step": 491
    },
    {
      "epoch": 0.23692289171131042,
      "grad_norm": 27.51399040222168,
      "learning_rate": 0.00017852927875204163,
      "loss": 2.8922066688537598,
      "step": 492
    },
    {
      "epoch": 0.2374044423042196,
      "grad_norm": 149.32382202148438,
      "learning_rate": 0.00017843260742201963,
      "loss": 2.72713041305542,
      "step": 493
    },
    {
      "epoch": 0.23788599289712875,
      "grad_norm": 154.91378784179688,
      "learning_rate": 0.00017833574524849535,
      "loss": 2.4105772972106934,
      "step": 494
    },
    {
      "epoch": 0.23836754349003791,
      "grad_norm": 34.86402893066406,
      "learning_rate": 0.00017823869246715553,
      "loss": 2.8086459636688232,
      "step": 495
    },
    {
      "epoch": 0.23884909408294708,
      "grad_norm": 47.89899444580078,
      "learning_rate": 0.00017814144931415043,
      "loss": 2.5183095932006836,
      "step": 496
    },
    {
      "epoch": 0.23933064467585627,
      "grad_norm": 52.987457275390625,
      "learning_rate": 0.0001780440160260938,
      "loss": 2.7695648670196533,
      "step": 497
    },
    {
      "epoch": 0.23981219526876543,
      "grad_norm": 26.18509292602539,
      "learning_rate": 0.00017794639284006184,
      "loss": 2.306504726409912,
      "step": 498
    },
    {
      "epoch": 0.2402937458616746,
      "grad_norm": 6.395096302032471,
      "learning_rate": 0.0001778485799935929,
      "loss": 2.467038154602051,
      "step": 499
    },
    {
      "epoch": 0.24077529645458376,
      "grad_norm": 12.862374305725098,
      "learning_rate": 0.00017775057772468679,
      "loss": 2.1746180057525635,
      "step": 500
    },
    {
      "epoch": 0.24125684704749292,
      "grad_norm": 17.83692741394043,
      "learning_rate": 0.00017765238627180424,
      "loss": 2.47994327545166,
      "step": 501
    },
    {
      "epoch": 0.2417383976404021,
      "grad_norm": 17.206254959106445,
      "learning_rate": 0.00017755400587386632,
      "loss": 2.160661458969116,
      "step": 502
    },
    {
      "epoch": 0.24221994823331125,
      "grad_norm": 12.832817077636719,
      "learning_rate": 0.00017745543677025378,
      "loss": 2.4732003211975098,
      "step": 503
    },
    {
      "epoch": 0.24270149882622044,
      "grad_norm": 53.17784118652344,
      "learning_rate": 0.00017735667920080661,
      "loss": 2.347672700881958,
      "step": 504
    },
    {
      "epoch": 0.2431830494191296,
      "grad_norm": 24.449148178100586,
      "learning_rate": 0.0001772577334058233,
      "loss": 2.395549774169922,
      "step": 505
    },
    {
      "epoch": 0.24366460001203877,
      "grad_norm": 235.66781616210938,
      "learning_rate": 0.00017715859962606043,
      "loss": 2.472766876220703,
      "step": 506
    },
    {
      "epoch": 0.24414615060494793,
      "grad_norm": 33.46131134033203,
      "learning_rate": 0.00017705927810273187,
      "loss": 2.684861660003662,
      "step": 507
    },
    {
      "epoch": 0.2446277011978571,
      "grad_norm": 16.207616806030273,
      "learning_rate": 0.00017695976907750844,
      "loss": 2.4997191429138184,
      "step": 508
    },
    {
      "epoch": 0.24510925179076626,
      "grad_norm": 9.04235553741455,
      "learning_rate": 0.00017686007279251706,
      "loss": 2.823452949523926,
      "step": 509
    },
    {
      "epoch": 0.24559080238367542,
      "grad_norm": 16.105796813964844,
      "learning_rate": 0.00017676018949034045,
      "loss": 2.274142026901245,
      "step": 510
    },
    {
      "epoch": 0.2460723529765846,
      "grad_norm": 10.982556343078613,
      "learning_rate": 0.0001766601194140162,
      "loss": 2.3691482543945312,
      "step": 511
    },
    {
      "epoch": 0.24655390356949378,
      "grad_norm": 54.55943298339844,
      "learning_rate": 0.0001765598628070365,
      "loss": 2.846904754638672,
      "step": 512
    },
    {
      "epoch": 0.24703545416240294,
      "grad_norm": 17.024335861206055,
      "learning_rate": 0.00017645941991334732,
      "loss": 2.9260151386260986,
      "step": 513
    },
    {
      "epoch": 0.2475170047553121,
      "grad_norm": 9.675862312316895,
      "learning_rate": 0.00017635879097734804,
      "loss": 2.5277135372161865,
      "step": 514
    },
    {
      "epoch": 0.24799855534822127,
      "grad_norm": 40.26479721069336,
      "learning_rate": 0.00017625797624389055,
      "loss": 2.398759603500366,
      "step": 515
    },
    {
      "epoch": 0.24848010594113043,
      "grad_norm": 59.35304641723633,
      "learning_rate": 0.00017615697595827897,
      "loss": 1.555891752243042,
      "step": 516
    },
    {
      "epoch": 0.2489616565340396,
      "grad_norm": 9.261337280273438,
      "learning_rate": 0.0001760557903662688,
      "loss": 2.568999767303467,
      "step": 517
    },
    {
      "epoch": 0.24944320712694878,
      "grad_norm": 14.720475196838379,
      "learning_rate": 0.00017595441971406648,
      "loss": 2.650329351425171,
      "step": 518
    },
    {
      "epoch": 0.24992475771985795,
      "grad_norm": 63.594017028808594,
      "learning_rate": 0.00017585286424832874,
      "loss": 3.071713447570801,
      "step": 519
    },
    {
      "epoch": 0.2504063083127671,
      "grad_norm": 86.94270324707031,
      "learning_rate": 0.00017575112421616202,
      "loss": 2.57895565032959,
      "step": 520
    },
    {
      "epoch": 0.2508878589056763,
      "grad_norm": 52.754966735839844,
      "learning_rate": 0.0001756491998651218,
      "loss": 2.1028871536254883,
      "step": 521
    },
    {
      "epoch": 0.25136940949858544,
      "grad_norm": 13.666325569152832,
      "learning_rate": 0.0001755470914432121,
      "loss": 1.934442162513733,
      "step": 522
    },
    {
      "epoch": 0.2518509600914946,
      "grad_norm": 26.120433807373047,
      "learning_rate": 0.0001754447991988848,
      "loss": 2.567622184753418,
      "step": 523
    },
    {
      "epoch": 0.25233251068440377,
      "grad_norm": 8.585360527038574,
      "learning_rate": 0.00017534232338103903,
      "loss": 2.4353749752044678,
      "step": 524
    },
    {
      "epoch": 0.25281406127731293,
      "grad_norm": 192.69387817382812,
      "learning_rate": 0.0001752396642390207,
      "loss": 2.6733694076538086,
      "step": 525
    },
    {
      "epoch": 0.2532956118702221,
      "grad_norm": 189.8889923095703,
      "learning_rate": 0.00017513682202262163,
      "loss": 2.719677686691284,
      "step": 526
    },
    {
      "epoch": 0.25377716246313126,
      "grad_norm": 88.1535873413086,
      "learning_rate": 0.00017503379698207918,
      "loss": 2.468797445297241,
      "step": 527
    },
    {
      "epoch": 0.2542587130560405,
      "grad_norm": 12.966519355773926,
      "learning_rate": 0.00017493058936807562,
      "loss": 2.1268651485443115,
      "step": 528
    },
    {
      "epoch": 0.25474026364894964,
      "grad_norm": 19.091760635375977,
      "learning_rate": 0.00017482719943173739,
      "loss": 2.5064504146575928,
      "step": 529
    },
    {
      "epoch": 0.2552218142418588,
      "grad_norm": 19.966796875,
      "learning_rate": 0.00017472362742463455,
      "loss": 2.6881327629089355,
      "step": 530
    },
    {
      "epoch": 0.25570336483476797,
      "grad_norm": 59.54732894897461,
      "learning_rate": 0.0001746198735987802,
      "loss": 2.624049663543701,
      "step": 531
    },
    {
      "epoch": 0.25618491542767713,
      "grad_norm": 29.037418365478516,
      "learning_rate": 0.00017451593820662988,
      "loss": 3.1915407180786133,
      "step": 532
    },
    {
      "epoch": 0.2566664660205863,
      "grad_norm": 15.46448802947998,
      "learning_rate": 0.00017441182150108086,
      "loss": 2.783754348754883,
      "step": 533
    },
    {
      "epoch": 0.25714801661349546,
      "grad_norm": 13.645626068115234,
      "learning_rate": 0.0001743075237354716,
      "loss": 3.077573537826538,
      "step": 534
    },
    {
      "epoch": 0.2576295672064046,
      "grad_norm": 28.609912872314453,
      "learning_rate": 0.00017420304516358113,
      "loss": 2.293807029724121,
      "step": 535
    },
    {
      "epoch": 0.2581111177993138,
      "grad_norm": 8.056380271911621,
      "learning_rate": 0.00017409838603962843,
      "loss": 2.546255350112915,
      "step": 536
    },
    {
      "epoch": 0.25859266839222295,
      "grad_norm": 10.664549827575684,
      "learning_rate": 0.00017399354661827178,
      "loss": 3.1626644134521484,
      "step": 537
    },
    {
      "epoch": 0.2590742189851321,
      "grad_norm": 7.4782819747924805,
      "learning_rate": 0.00017388852715460819,
      "loss": 2.656256675720215,
      "step": 538
    },
    {
      "epoch": 0.2595557695780413,
      "grad_norm": 17.77277946472168,
      "learning_rate": 0.00017378332790417273,
      "loss": 2.3249218463897705,
      "step": 539
    },
    {
      "epoch": 0.26003732017095044,
      "grad_norm": 14.926592826843262,
      "learning_rate": 0.00017367794912293794,
      "loss": 2.455500841140747,
      "step": 540
    },
    {
      "epoch": 0.26051887076385966,
      "grad_norm": 6.42595100402832,
      "learning_rate": 0.00017357239106731317,
      "loss": 2.6990835666656494,
      "step": 541
    },
    {
      "epoch": 0.2610004213567688,
      "grad_norm": 10.236385345458984,
      "learning_rate": 0.00017346665399414405,
      "loss": 2.9466168880462646,
      "step": 542
    },
    {
      "epoch": 0.261481971949678,
      "grad_norm": 38.052101135253906,
      "learning_rate": 0.00017336073816071168,
      "loss": 2.454197406768799,
      "step": 543
    },
    {
      "epoch": 0.26196352254258715,
      "grad_norm": 15.537681579589844,
      "learning_rate": 0.00017325464382473226,
      "loss": 2.6061935424804688,
      "step": 544
    },
    {
      "epoch": 0.2624450731354963,
      "grad_norm": 18.82579231262207,
      "learning_rate": 0.00017314837124435622,
      "loss": 2.728590488433838,
      "step": 545
    },
    {
      "epoch": 0.2629266237284055,
      "grad_norm": 9.638214111328125,
      "learning_rate": 0.00017304192067816782,
      "loss": 3.074964761734009,
      "step": 546
    },
    {
      "epoch": 0.26340817432131464,
      "grad_norm": 13.469003677368164,
      "learning_rate": 0.00017293529238518422,
      "loss": 2.637434244155884,
      "step": 547
    },
    {
      "epoch": 0.2638897249142238,
      "grad_norm": 46.80735397338867,
      "learning_rate": 0.0001728284866248552,
      "loss": 1.804068684577942,
      "step": 548
    },
    {
      "epoch": 0.26437127550713296,
      "grad_norm": 289.72039794921875,
      "learning_rate": 0.00017272150365706224,
      "loss": 2.302100658416748,
      "step": 549
    },
    {
      "epoch": 0.2648528261000421,
      "grad_norm": 8.992754936218262,
      "learning_rate": 0.00017261434374211802,
      "loss": 2.0606307983398438,
      "step": 550
    },
    {
      "epoch": 0.2653343766929513,
      "grad_norm": 13.549088478088379,
      "learning_rate": 0.00017250700714076586,
      "loss": 2.5550038814544678,
      "step": 551
    },
    {
      "epoch": 0.26581592728586045,
      "grad_norm": 34.892704010009766,
      "learning_rate": 0.00017239949411417888,
      "loss": 2.315070152282715,
      "step": 552
    },
    {
      "epoch": 0.2662974778787696,
      "grad_norm": 13.028685569763184,
      "learning_rate": 0.0001722918049239596,
      "loss": 2.589421510696411,
      "step": 553
    },
    {
      "epoch": 0.2667790284716788,
      "grad_norm": 24.265100479125977,
      "learning_rate": 0.00017218393983213902,
      "loss": 2.2328381538391113,
      "step": 554
    },
    {
      "epoch": 0.267260579064588,
      "grad_norm": 7.062382698059082,
      "learning_rate": 0.00017207589910117634,
      "loss": 1.8611013889312744,
      "step": 555
    },
    {
      "epoch": 0.26774212965749716,
      "grad_norm": 55.12726593017578,
      "learning_rate": 0.00017196768299395797,
      "loss": 2.52905011177063,
      "step": 556
    },
    {
      "epoch": 0.2682236802504063,
      "grad_norm": 47.61922836303711,
      "learning_rate": 0.00017185929177379714,
      "loss": 1.833607792854309,
      "step": 557
    },
    {
      "epoch": 0.2687052308433155,
      "grad_norm": 21.871103286743164,
      "learning_rate": 0.00017175072570443312,
      "loss": 3.340728282928467,
      "step": 558
    },
    {
      "epoch": 0.26918678143622465,
      "grad_norm": 14.290611267089844,
      "learning_rate": 0.00017164198505003066,
      "loss": 2.414210796356201,
      "step": 559
    },
    {
      "epoch": 0.2696683320291338,
      "grad_norm": 38.987510681152344,
      "learning_rate": 0.0001715330700751793,
      "loss": 1.6949281692504883,
      "step": 560
    },
    {
      "epoch": 0.270149882622043,
      "grad_norm": 32.56105041503906,
      "learning_rate": 0.00017142398104489273,
      "loss": 2.3986265659332275,
      "step": 561
    },
    {
      "epoch": 0.27063143321495214,
      "grad_norm": 8.45708179473877,
      "learning_rate": 0.00017131471822460814,
      "loss": 2.2397377490997314,
      "step": 562
    },
    {
      "epoch": 0.2711129838078613,
      "grad_norm": 30.394838333129883,
      "learning_rate": 0.00017120528188018565,
      "loss": 2.5779869556427,
      "step": 563
    },
    {
      "epoch": 0.27159453440077047,
      "grad_norm": 107.7567138671875,
      "learning_rate": 0.00017109567227790754,
      "loss": 2.3585710525512695,
      "step": 564
    },
    {
      "epoch": 0.27207608499367963,
      "grad_norm": 26.995412826538086,
      "learning_rate": 0.00017098588968447766,
      "loss": 2.232757329940796,
      "step": 565
    },
    {
      "epoch": 0.2725576355865888,
      "grad_norm": 40.55866241455078,
      "learning_rate": 0.00017087593436702084,
      "loss": 2.423438787460327,
      "step": 566
    },
    {
      "epoch": 0.27303918617949796,
      "grad_norm": 248.18673706054688,
      "learning_rate": 0.00017076580659308222,
      "loss": 2.8960745334625244,
      "step": 567
    },
    {
      "epoch": 0.2735207367724071,
      "grad_norm": 24.69264030456543,
      "learning_rate": 0.00017065550663062634,
      "loss": 1.9048912525177002,
      "step": 568
    },
    {
      "epoch": 0.27400228736531634,
      "grad_norm": 23.592605590820312,
      "learning_rate": 0.00017054503474803702,
      "loss": 2.209522008895874,
      "step": 569
    },
    {
      "epoch": 0.2744838379582255,
      "grad_norm": 19.044551849365234,
      "learning_rate": 0.00017043439121411618,
      "loss": 1.9869555234909058,
      "step": 570
    },
    {
      "epoch": 0.27496538855113467,
      "grad_norm": 14.33376693725586,
      "learning_rate": 0.0001703235762980835,
      "loss": 2.310392141342163,
      "step": 571
    },
    {
      "epoch": 0.27544693914404383,
      "grad_norm": 5.432505130767822,
      "learning_rate": 0.00017021259026957567,
      "loss": 2.254514694213867,
      "step": 572
    },
    {
      "epoch": 0.275928489736953,
      "grad_norm": 967.5568237304688,
      "learning_rate": 0.00017010143339864562,
      "loss": 2.232318878173828,
      "step": 573
    },
    {
      "epoch": 0.27641004032986216,
      "grad_norm": 20.084211349487305,
      "learning_rate": 0.0001699901059557621,
      "loss": 2.2254064083099365,
      "step": 574
    },
    {
      "epoch": 0.2768915909227713,
      "grad_norm": 63.53519821166992,
      "learning_rate": 0.00016987860821180895,
      "loss": 2.33610463142395,
      "step": 575
    },
    {
      "epoch": 0.2773731415156805,
      "grad_norm": 167.72093200683594,
      "learning_rate": 0.00016976694043808416,
      "loss": 2.92610502243042,
      "step": 576
    },
    {
      "epoch": 0.27785469210858965,
      "grad_norm": 58.92509078979492,
      "learning_rate": 0.00016965510290629972,
      "loss": 2.6089067459106445,
      "step": 577
    },
    {
      "epoch": 0.2783362427014988,
      "grad_norm": 10.799323081970215,
      "learning_rate": 0.00016954309588858044,
      "loss": 1.757743239402771,
      "step": 578
    },
    {
      "epoch": 0.278817793294408,
      "grad_norm": 11.51849365234375,
      "learning_rate": 0.00016943091965746366,
      "loss": 2.1270101070404053,
      "step": 579
    },
    {
      "epoch": 0.27929934388731714,
      "grad_norm": 5.338193416595459,
      "learning_rate": 0.00016931857448589845,
      "loss": 2.099097728729248,
      "step": 580
    },
    {
      "epoch": 0.2797808944802263,
      "grad_norm": 12.396195411682129,
      "learning_rate": 0.00016920606064724488,
      "loss": 2.6628668308258057,
      "step": 581
    },
    {
      "epoch": 0.28026244507313547,
      "grad_norm": 13.933058738708496,
      "learning_rate": 0.00016909337841527344,
      "loss": 2.8974523544311523,
      "step": 582
    },
    {
      "epoch": 0.2807439956660447,
      "grad_norm": 11.008697509765625,
      "learning_rate": 0.00016898052806416444,
      "loss": 2.3536436557769775,
      "step": 583
    },
    {
      "epoch": 0.28122554625895385,
      "grad_norm": 16.239116668701172,
      "learning_rate": 0.00016886750986850718,
      "loss": 2.319664478302002,
      "step": 584
    },
    {
      "epoch": 0.281707096851863,
      "grad_norm": 26.50018882751465,
      "learning_rate": 0.00016875432410329934,
      "loss": 2.7283129692077637,
      "step": 585
    },
    {
      "epoch": 0.2821886474447722,
      "grad_norm": 22.730674743652344,
      "learning_rate": 0.0001686409710439464,
      "loss": 2.2959179878234863,
      "step": 586
    },
    {
      "epoch": 0.28267019803768134,
      "grad_norm": 9.715805053710938,
      "learning_rate": 0.00016852745096626088,
      "loss": 2.2727138996124268,
      "step": 587
    },
    {
      "epoch": 0.2831517486305905,
      "grad_norm": 14.332803726196289,
      "learning_rate": 0.0001684137641464617,
      "loss": 1.8203465938568115,
      "step": 588
    },
    {
      "epoch": 0.28363329922349967,
      "grad_norm": 16.769222259521484,
      "learning_rate": 0.0001682999108611735,
      "loss": 2.321789264678955,
      "step": 589
    },
    {
      "epoch": 0.28411484981640883,
      "grad_norm": 15.815760612487793,
      "learning_rate": 0.00016818589138742587,
      "loss": 2.592203378677368,
      "step": 590
    },
    {
      "epoch": 0.284596400409318,
      "grad_norm": 28.57355499267578,
      "learning_rate": 0.00016807170600265296,
      "loss": 3.2235562801361084,
      "step": 591
    },
    {
      "epoch": 0.28507795100222716,
      "grad_norm": 42.78847122192383,
      "learning_rate": 0.00016795735498469246,
      "loss": 3.079529285430908,
      "step": 592
    },
    {
      "epoch": 0.2855595015951363,
      "grad_norm": 32.268470764160156,
      "learning_rate": 0.00016784283861178513,
      "loss": 2.7266907691955566,
      "step": 593
    },
    {
      "epoch": 0.2860410521880455,
      "grad_norm": 24.987525939941406,
      "learning_rate": 0.00016772815716257412,
      "loss": 2.5390384197235107,
      "step": 594
    },
    {
      "epoch": 0.28652260278095465,
      "grad_norm": 33.51164245605469,
      "learning_rate": 0.00016761331091610416,
      "loss": 2.590195894241333,
      "step": 595
    },
    {
      "epoch": 0.28700415337386387,
      "grad_norm": 333.5224304199219,
      "learning_rate": 0.00016749830015182107,
      "loss": 2.4049832820892334,
      "step": 596
    },
    {
      "epoch": 0.28748570396677303,
      "grad_norm": 18.038150787353516,
      "learning_rate": 0.00016738312514957086,
      "loss": 1.614684820175171,
      "step": 597
    },
    {
      "epoch": 0.2879672545596822,
      "grad_norm": 36.80668640136719,
      "learning_rate": 0.00016726778618959926,
      "loss": 2.692356586456299,
      "step": 598
    },
    {
      "epoch": 0.28844880515259136,
      "grad_norm": 24.034074783325195,
      "learning_rate": 0.00016715228355255093,
      "loss": 1.719312071800232,
      "step": 599
    },
    {
      "epoch": 0.2889303557455005,
      "grad_norm": 21.068912506103516,
      "learning_rate": 0.00016703661751946874,
      "loss": 2.5963356494903564,
      "step": 600
    },
    {
      "epoch": 0.2894119063384097,
      "grad_norm": 8.49599838256836,
      "learning_rate": 0.00016692078837179318,
      "loss": 2.1785662174224854,
      "step": 601
    },
    {
      "epoch": 0.28989345693131885,
      "grad_norm": 26.427785873413086,
      "learning_rate": 0.00016680479639136163,
      "loss": 2.4436824321746826,
      "step": 602
    },
    {
      "epoch": 0.290375007524228,
      "grad_norm": 32.854190826416016,
      "learning_rate": 0.0001666886418604077,
      "loss": 2.0304856300354004,
      "step": 603
    },
    {
      "epoch": 0.2908565581171372,
      "grad_norm": 21.69548988342285,
      "learning_rate": 0.0001665723250615604,
      "loss": 2.2865593433380127,
      "step": 604
    },
    {
      "epoch": 0.29133810871004634,
      "grad_norm": 44.677024841308594,
      "learning_rate": 0.00016645584627784381,
      "loss": 2.7337863445281982,
      "step": 605
    },
    {
      "epoch": 0.2918196593029555,
      "grad_norm": 17.892852783203125,
      "learning_rate": 0.0001663392057926759,
      "loss": 3.02669095993042,
      "step": 606
    },
    {
      "epoch": 0.29230120989586467,
      "grad_norm": 12.398837089538574,
      "learning_rate": 0.00016622240388986824,
      "loss": 2.6195502281188965,
      "step": 607
    },
    {
      "epoch": 0.29278276048877383,
      "grad_norm": 73.73602294921875,
      "learning_rate": 0.0001661054408536251,
      "loss": 2.8628742694854736,
      "step": 608
    },
    {
      "epoch": 0.293264311081683,
      "grad_norm": 25.772382736206055,
      "learning_rate": 0.00016598831696854288,
      "loss": 2.5869789123535156,
      "step": 609
    },
    {
      "epoch": 0.2937458616745922,
      "grad_norm": 13.439327239990234,
      "learning_rate": 0.00016587103251960937,
      "loss": 2.3485794067382812,
      "step": 610
    },
    {
      "epoch": 0.2942274122675014,
      "grad_norm": 78.83702850341797,
      "learning_rate": 0.00016575358779220294,
      "loss": 2.385709524154663,
      "step": 611
    },
    {
      "epoch": 0.29470896286041054,
      "grad_norm": 14.793705940246582,
      "learning_rate": 0.00016563598307209204,
      "loss": 2.0593929290771484,
      "step": 612
    },
    {
      "epoch": 0.2951905134533197,
      "grad_norm": 64.18264770507812,
      "learning_rate": 0.0001655182186454344,
      "loss": 1.8273186683654785,
      "step": 613
    },
    {
      "epoch": 0.29567206404622887,
      "grad_norm": 48.75865173339844,
      "learning_rate": 0.00016540029479877638,
      "loss": 1.9868485927581787,
      "step": 614
    },
    {
      "epoch": 0.29615361463913803,
      "grad_norm": 56.14519500732422,
      "learning_rate": 0.00016528221181905217,
      "loss": 2.613802671432495,
      "step": 615
    },
    {
      "epoch": 0.2966351652320472,
      "grad_norm": 14.928141593933105,
      "learning_rate": 0.00016516396999358322,
      "loss": 2.2463765144348145,
      "step": 616
    },
    {
      "epoch": 0.29711671582495636,
      "grad_norm": 22.276084899902344,
      "learning_rate": 0.00016504556961007748,
      "loss": 2.727055311203003,
      "step": 617
    },
    {
      "epoch": 0.2975982664178655,
      "grad_norm": 41.9312629699707,
      "learning_rate": 0.00016492701095662866,
      "loss": 2.3171141147613525,
      "step": 618
    },
    {
      "epoch": 0.2980798170107747,
      "grad_norm": 122.03096771240234,
      "learning_rate": 0.00016480829432171564,
      "loss": 2.0118510723114014,
      "step": 619
    },
    {
      "epoch": 0.29856136760368385,
      "grad_norm": 35.15409469604492,
      "learning_rate": 0.0001646894199942017,
      "loss": 2.2544853687286377,
      "step": 620
    },
    {
      "epoch": 0.299042918196593,
      "grad_norm": 7.291553497314453,
      "learning_rate": 0.0001645703882633338,
      "loss": 2.150805711746216,
      "step": 621
    },
    {
      "epoch": 0.2995244687895022,
      "grad_norm": 14.819894790649414,
      "learning_rate": 0.00016445119941874183,
      "loss": 2.109992027282715,
      "step": 622
    },
    {
      "epoch": 0.30000601938241134,
      "grad_norm": 16.467649459838867,
      "learning_rate": 0.00016433185375043809,
      "loss": 2.3878743648529053,
      "step": 623
    },
    {
      "epoch": 0.30048756997532056,
      "grad_norm": 16.413236618041992,
      "learning_rate": 0.00016421235154881638,
      "loss": 2.418107748031616,
      "step": 624
    },
    {
      "epoch": 0.3009691205682297,
      "grad_norm": 104.33487701416016,
      "learning_rate": 0.00016409269310465146,
      "loss": 2.095888137817383,
      "step": 625
    },
    {
      "epoch": 0.3014506711611389,
      "grad_norm": 93.90336608886719,
      "learning_rate": 0.00016397287870909813,
      "loss": 2.4379069805145264,
      "step": 626
    },
    {
      "epoch": 0.30193222175404805,
      "grad_norm": 6.064742088317871,
      "learning_rate": 0.00016385290865369079,
      "loss": 2.344343423843384,
      "step": 627
    },
    {
      "epoch": 0.3024137723469572,
      "grad_norm": 6.753592014312744,
      "learning_rate": 0.00016373278323034255,
      "loss": 3.0524675846099854,
      "step": 628
    },
    {
      "epoch": 0.3028953229398664,
      "grad_norm": 10.497713088989258,
      "learning_rate": 0.0001636125027313445,
      "loss": 2.62998628616333,
      "step": 629
    },
    {
      "epoch": 0.30337687353277554,
      "grad_norm": 295.0896301269531,
      "learning_rate": 0.00016349206744936518,
      "loss": 2.4529950618743896,
      "step": 630
    },
    {
      "epoch": 0.3038584241256847,
      "grad_norm": 6.026525497436523,
      "learning_rate": 0.00016337147767744967,
      "loss": 2.6777803897857666,
      "step": 631
    },
    {
      "epoch": 0.30433997471859386,
      "grad_norm": 17.004899978637695,
      "learning_rate": 0.0001632507337090189,
      "loss": 2.276102066040039,
      "step": 632
    },
    {
      "epoch": 0.304821525311503,
      "grad_norm": 13.892413139343262,
      "learning_rate": 0.0001631298358378692,
      "loss": 2.5310311317443848,
      "step": 633
    },
    {
      "epoch": 0.3053030759044122,
      "grad_norm": 43.393882751464844,
      "learning_rate": 0.00016300878435817113,
      "loss": 2.3083906173706055,
      "step": 634
    },
    {
      "epoch": 0.30578462649732135,
      "grad_norm": 18.534343719482422,
      "learning_rate": 0.00016288757956446918,
      "loss": 2.8893001079559326,
      "step": 635
    },
    {
      "epoch": 0.3062661770902305,
      "grad_norm": 61.918704986572266,
      "learning_rate": 0.00016276622175168083,
      "loss": 2.5668399333953857,
      "step": 636
    },
    {
      "epoch": 0.30674772768313974,
      "grad_norm": 11.04572868347168,
      "learning_rate": 0.0001626447112150959,
      "loss": 2.106292963027954,
      "step": 637
    },
    {
      "epoch": 0.3072292782760489,
      "grad_norm": 63.03369903564453,
      "learning_rate": 0.00016252304825037576,
      "loss": 2.066943407058716,
      "step": 638
    },
    {
      "epoch": 0.30771082886895806,
      "grad_norm": 11.14792251586914,
      "learning_rate": 0.0001624012331535528,
      "loss": 2.8888325691223145,
      "step": 639
    },
    {
      "epoch": 0.3081923794618672,
      "grad_norm": 5.257469654083252,
      "learning_rate": 0.00016227926622102947,
      "loss": 1.7506868839263916,
      "step": 640
    },
    {
      "epoch": 0.3086739300547764,
      "grad_norm": 14.95846176147461,
      "learning_rate": 0.00016215714774957772,
      "loss": 2.212770462036133,
      "step": 641
    },
    {
      "epoch": 0.30915548064768555,
      "grad_norm": 32.55873489379883,
      "learning_rate": 0.00016203487803633822,
      "loss": 2.5489087104797363,
      "step": 642
    },
    {
      "epoch": 0.3096370312405947,
      "grad_norm": 12.045295715332031,
      "learning_rate": 0.00016191245737881956,
      "loss": 2.1127991676330566,
      "step": 643
    },
    {
      "epoch": 0.3101185818335039,
      "grad_norm": 5.2542619705200195,
      "learning_rate": 0.00016178988607489777,
      "loss": 2.125332832336426,
      "step": 644
    },
    {
      "epoch": 0.31060013242641304,
      "grad_norm": 6.214310169219971,
      "learning_rate": 0.00016166716442281528,
      "loss": 2.1081697940826416,
      "step": 645
    },
    {
      "epoch": 0.3110816830193222,
      "grad_norm": 6.018392562866211,
      "learning_rate": 0.0001615442927211805,
      "loss": 2.5175414085388184,
      "step": 646
    },
    {
      "epoch": 0.31156323361223137,
      "grad_norm": 7.883767127990723,
      "learning_rate": 0.0001614212712689668,
      "loss": 2.4246666431427,
      "step": 647
    },
    {
      "epoch": 0.31204478420514054,
      "grad_norm": 182.22850036621094,
      "learning_rate": 0.00016129810036551198,
      "loss": 2.344752073287964,
      "step": 648
    },
    {
      "epoch": 0.3125263347980497,
      "grad_norm": 56.43985366821289,
      "learning_rate": 0.00016117478031051755,
      "loss": 2.286370277404785,
      "step": 649
    },
    {
      "epoch": 0.31300788539095886,
      "grad_norm": 22.010725021362305,
      "learning_rate": 0.00016105131140404787,
      "loss": 2.556460380554199,
      "step": 650
    },
    {
      "epoch": 0.3134894359838681,
      "grad_norm": 22.090927124023438,
      "learning_rate": 0.00016092769394652947,
      "loss": 2.856084108352661,
      "step": 651
    },
    {
      "epoch": 0.31397098657677724,
      "grad_norm": 122.61199951171875,
      "learning_rate": 0.0001608039282387504,
      "loss": 2.135643482208252,
      "step": 652
    },
    {
      "epoch": 0.3144525371696864,
      "grad_norm": 11.539798736572266,
      "learning_rate": 0.00016068001458185936,
      "loss": 2.402561664581299,
      "step": 653
    },
    {
      "epoch": 0.31493408776259557,
      "grad_norm": 19.182567596435547,
      "learning_rate": 0.0001605559532773651,
      "loss": 2.0156970024108887,
      "step": 654
    },
    {
      "epoch": 0.31541563835550473,
      "grad_norm": 32.8856086730957,
      "learning_rate": 0.00016043174462713566,
      "loss": 2.5581769943237305,
      "step": 655
    },
    {
      "epoch": 0.3158971889484139,
      "grad_norm": 6.733951091766357,
      "learning_rate": 0.00016030738893339753,
      "loss": 1.709293246269226,
      "step": 656
    },
    {
      "epoch": 0.31637873954132306,
      "grad_norm": 4.393002510070801,
      "learning_rate": 0.00016018288649873497,
      "loss": 2.59871506690979,
      "step": 657
    },
    {
      "epoch": 0.3168602901342322,
      "grad_norm": 5.657468318939209,
      "learning_rate": 0.0001600582376260894,
      "loss": 1.8606971502304077,
      "step": 658
    },
    {
      "epoch": 0.3173418407271414,
      "grad_norm": 4.717591762542725,
      "learning_rate": 0.00015993344261875847,
      "loss": 2.3747005462646484,
      "step": 659
    },
    {
      "epoch": 0.31782339132005055,
      "grad_norm": 17.555906295776367,
      "learning_rate": 0.00015980850178039547,
      "loss": 2.9213943481445312,
      "step": 660
    },
    {
      "epoch": 0.3183049419129597,
      "grad_norm": 20.832429885864258,
      "learning_rate": 0.00015968341541500842,
      "loss": 2.14072585105896,
      "step": 661
    },
    {
      "epoch": 0.3187864925058689,
      "grad_norm": 9.35422420501709,
      "learning_rate": 0.00015955818382695953,
      "loss": 2.0997297763824463,
      "step": 662
    },
    {
      "epoch": 0.31926804309877804,
      "grad_norm": 7.820318698883057,
      "learning_rate": 0.00015943280732096438,
      "loss": 2.7113537788391113,
      "step": 663
    },
    {
      "epoch": 0.3197495936916872,
      "grad_norm": 82.76976776123047,
      "learning_rate": 0.00015930728620209113,
      "loss": 2.383936882019043,
      "step": 664
    },
    {
      "epoch": 0.3202311442845964,
      "grad_norm": 16.245208740234375,
      "learning_rate": 0.00015918162077575976,
      "loss": 2.848801612854004,
      "step": 665
    },
    {
      "epoch": 0.3207126948775056,
      "grad_norm": 6.3594651222229,
      "learning_rate": 0.00015905581134774153,
      "loss": 2.2924814224243164,
      "step": 666
    },
    {
      "epoch": 0.32119424547041475,
      "grad_norm": 7.1573381423950195,
      "learning_rate": 0.0001589298582241579,
      "loss": 1.8828836679458618,
      "step": 667
    },
    {
      "epoch": 0.3216757960633239,
      "grad_norm": 57.59177780151367,
      "learning_rate": 0.00015880376171148014,
      "loss": 2.33325457572937,
      "step": 668
    },
    {
      "epoch": 0.3221573466562331,
      "grad_norm": 12.003190994262695,
      "learning_rate": 0.00015867752211652831,
      "loss": 2.0064451694488525,
      "step": 669
    },
    {
      "epoch": 0.32263889724914224,
      "grad_norm": 12.15641975402832,
      "learning_rate": 0.00015855113974647068,
      "loss": 2.530094861984253,
      "step": 670
    },
    {
      "epoch": 0.3231204478420514,
      "grad_norm": 16.05088996887207,
      "learning_rate": 0.0001584246149088229,
      "loss": 1.7892688512802124,
      "step": 671
    },
    {
      "epoch": 0.32360199843496057,
      "grad_norm": 5.292698860168457,
      "learning_rate": 0.0001582979479114472,
      "loss": 2.4936039447784424,
      "step": 672
    },
    {
      "epoch": 0.32408354902786973,
      "grad_norm": 18.902830123901367,
      "learning_rate": 0.0001581711390625519,
      "loss": 2.8403360843658447,
      "step": 673
    },
    {
      "epoch": 0.3245650996207789,
      "grad_norm": 31.117820739746094,
      "learning_rate": 0.0001580441886706903,
      "loss": 1.7805073261260986,
      "step": 674
    },
    {
      "epoch": 0.32504665021368806,
      "grad_norm": 12.562667846679688,
      "learning_rate": 0.00015791709704476015,
      "loss": 2.37481427192688,
      "step": 675
    },
    {
      "epoch": 0.3255282008065972,
      "grad_norm": 12.202144622802734,
      "learning_rate": 0.00015778986449400292,
      "loss": 1.7051687240600586,
      "step": 676
    },
    {
      "epoch": 0.3260097513995064,
      "grad_norm": 35.98042297363281,
      "learning_rate": 0.00015766249132800292,
      "loss": 2.4919230937957764,
      "step": 677
    },
    {
      "epoch": 0.3264913019924156,
      "grad_norm": 31.834667205810547,
      "learning_rate": 0.00015753497785668663,
      "loss": 1.7195661067962646,
      "step": 678
    },
    {
      "epoch": 0.32697285258532477,
      "grad_norm": 16.711477279663086,
      "learning_rate": 0.00015740732439032187,
      "loss": 2.2131874561309814,
      "step": 679
    },
    {
      "epoch": 0.32745440317823393,
      "grad_norm": 14.667591094970703,
      "learning_rate": 0.00015727953123951716,
      "loss": 3.0460386276245117,
      "step": 680
    },
    {
      "epoch": 0.3279359537711431,
      "grad_norm": 8.86426067352295,
      "learning_rate": 0.00015715159871522086,
      "loss": 1.7072958946228027,
      "step": 681
    },
    {
      "epoch": 0.32841750436405226,
      "grad_norm": 27.242816925048828,
      "learning_rate": 0.00015702352712872056,
      "loss": 2.2202649116516113,
      "step": 682
    },
    {
      "epoch": 0.3288990549569614,
      "grad_norm": 8.735687255859375,
      "learning_rate": 0.00015689531679164204,
      "loss": 2.658783197402954,
      "step": 683
    },
    {
      "epoch": 0.3293806055498706,
      "grad_norm": 56.72428894042969,
      "learning_rate": 0.00015676696801594886,
      "loss": 2.427795171737671,
      "step": 684
    },
    {
      "epoch": 0.32986215614277975,
      "grad_norm": 10.499786376953125,
      "learning_rate": 0.00015663848111394132,
      "loss": 2.6342363357543945,
      "step": 685
    },
    {
      "epoch": 0.3303437067356889,
      "grad_norm": 19.672365188598633,
      "learning_rate": 0.00015650985639825585,
      "loss": 2.5653975009918213,
      "step": 686
    },
    {
      "epoch": 0.3308252573285981,
      "grad_norm": 11.135467529296875,
      "learning_rate": 0.00015638109418186424,
      "loss": 2.592973470687866,
      "step": 687
    },
    {
      "epoch": 0.33130680792150724,
      "grad_norm": 9.212013244628906,
      "learning_rate": 0.00015625219477807277,
      "loss": 2.647275686264038,
      "step": 688
    },
    {
      "epoch": 0.3317883585144164,
      "grad_norm": 39.059932708740234,
      "learning_rate": 0.00015612315850052166,
      "loss": 2.411264181137085,
      "step": 689
    },
    {
      "epoch": 0.33226990910732557,
      "grad_norm": 41.978641510009766,
      "learning_rate": 0.00015599398566318396,
      "loss": 1.9640562534332275,
      "step": 690
    },
    {
      "epoch": 0.33275145970023473,
      "grad_norm": 21.49881935119629,
      "learning_rate": 0.00015586467658036524,
      "loss": 2.3599777221679688,
      "step": 691
    },
    {
      "epoch": 0.33323301029314395,
      "grad_norm": 19.055150985717773,
      "learning_rate": 0.00015573523156670244,
      "loss": 2.553987979888916,
      "step": 692
    },
    {
      "epoch": 0.3337145608860531,
      "grad_norm": 33.703575134277344,
      "learning_rate": 0.0001556056509371633,
      "loss": 2.4919118881225586,
      "step": 693
    },
    {
      "epoch": 0.3341961114789623,
      "grad_norm": 9.815055847167969,
      "learning_rate": 0.00015547593500704547,
      "loss": 1.6223469972610474,
      "step": 694
    },
    {
      "epoch": 0.33467766207187144,
      "grad_norm": 152.9043426513672,
      "learning_rate": 0.00015534608409197592,
      "loss": 1.458725094795227,
      "step": 695
    },
    {
      "epoch": 0.3351592126647806,
      "grad_norm": 11.295888900756836,
      "learning_rate": 0.00015521609850791004,
      "loss": 2.5387728214263916,
      "step": 696
    },
    {
      "epoch": 0.33564076325768977,
      "grad_norm": 14.85639762878418,
      "learning_rate": 0.0001550859785711308,
      "loss": 2.8755013942718506,
      "step": 697
    },
    {
      "epoch": 0.33612231385059893,
      "grad_norm": 8.181758880615234,
      "learning_rate": 0.0001549557245982482,
      "loss": 2.284562826156616,
      "step": 698
    },
    {
      "epoch": 0.3366038644435081,
      "grad_norm": 13.32393741607666,
      "learning_rate": 0.00015482533690619837,
      "loss": 2.453606128692627,
      "step": 699
    },
    {
      "epoch": 0.33708541503641726,
      "grad_norm": 7.509459495544434,
      "learning_rate": 0.00015469481581224272,
      "loss": 2.2195723056793213,
      "step": 700
    },
    {
      "epoch": 0.3375669656293264,
      "grad_norm": 21.973323822021484,
      "learning_rate": 0.0001545641616339673,
      "loss": 1.9575344324111938,
      "step": 701
    },
    {
      "epoch": 0.3380485162222356,
      "grad_norm": 12.047426223754883,
      "learning_rate": 0.00015443337468928206,
      "loss": 2.821323871612549,
      "step": 702
    },
    {
      "epoch": 0.33853006681514475,
      "grad_norm": 11.596617698669434,
      "learning_rate": 0.00015430245529641986,
      "loss": 2.5833516120910645,
      "step": 703
    },
    {
      "epoch": 0.3390116174080539,
      "grad_norm": 17.896316528320312,
      "learning_rate": 0.00015417140377393596,
      "loss": 2.65431547164917,
      "step": 704
    },
    {
      "epoch": 0.3394931680009631,
      "grad_norm": 279.6241760253906,
      "learning_rate": 0.00015404022044070704,
      "loss": 2.280409097671509,
      "step": 705
    },
    {
      "epoch": 0.3399747185938723,
      "grad_norm": 11.556683540344238,
      "learning_rate": 0.00015390890561593052,
      "loss": 2.7353715896606445,
      "step": 706
    },
    {
      "epoch": 0.34045626918678146,
      "grad_norm": 32.12382125854492,
      "learning_rate": 0.0001537774596191238,
      "loss": 2.1327977180480957,
      "step": 707
    },
    {
      "epoch": 0.3409378197796906,
      "grad_norm": 14.098381042480469,
      "learning_rate": 0.00015364588277012344,
      "loss": 2.3116259574890137,
      "step": 708
    },
    {
      "epoch": 0.3414193703725998,
      "grad_norm": 22.82956314086914,
      "learning_rate": 0.00015351417538908435,
      "loss": 2.150869846343994,
      "step": 709
    },
    {
      "epoch": 0.34190092096550895,
      "grad_norm": 19.959571838378906,
      "learning_rate": 0.0001533823377964791,
      "loss": 2.2382171154022217,
      "step": 710
    },
    {
      "epoch": 0.3423824715584181,
      "grad_norm": 29.652143478393555,
      "learning_rate": 0.00015325037031309704,
      "loss": 2.512044906616211,
      "step": 711
    },
    {
      "epoch": 0.3428640221513273,
      "grad_norm": 31.423152923583984,
      "learning_rate": 0.00015311827326004363,
      "loss": 2.375974178314209,
      "step": 712
    },
    {
      "epoch": 0.34334557274423644,
      "grad_norm": 10.05967903137207,
      "learning_rate": 0.0001529860469587396,
      "loss": 2.4383912086486816,
      "step": 713
    },
    {
      "epoch": 0.3438271233371456,
      "grad_norm": 53.269561767578125,
      "learning_rate": 0.00015285369173092015,
      "loss": 1.775095820426941,
      "step": 714
    },
    {
      "epoch": 0.34430867393005476,
      "grad_norm": 11.313742637634277,
      "learning_rate": 0.00015272120789863413,
      "loss": 2.16009783744812,
      "step": 715
    },
    {
      "epoch": 0.34479022452296393,
      "grad_norm": 86.95508575439453,
      "learning_rate": 0.00015258859578424342,
      "loss": 2.761847972869873,
      "step": 716
    },
    {
      "epoch": 0.3452717751158731,
      "grad_norm": 23.4042911529541,
      "learning_rate": 0.00015245585571042194,
      "loss": 2.228325843811035,
      "step": 717
    },
    {
      "epoch": 0.34575332570878226,
      "grad_norm": 43.9120979309082,
      "learning_rate": 0.00015232298800015506,
      "loss": 2.3220055103302,
      "step": 718
    },
    {
      "epoch": 0.3462348763016914,
      "grad_norm": 82.33618927001953,
      "learning_rate": 0.00015218999297673862,
      "loss": 2.7402896881103516,
      "step": 719
    },
    {
      "epoch": 0.34671642689460064,
      "grad_norm": 619.4547729492188,
      "learning_rate": 0.0001520568709637783,
      "loss": 2.3210721015930176,
      "step": 720
    },
    {
      "epoch": 0.3471979774875098,
      "grad_norm": 63.31781005859375,
      "learning_rate": 0.00015192362228518875,
      "loss": 2.2780189514160156,
      "step": 721
    },
    {
      "epoch": 0.34767952808041896,
      "grad_norm": 463.1142578125,
      "learning_rate": 0.00015179024726519284,
      "loss": 2.2689239978790283,
      "step": 722
    },
    {
      "epoch": 0.34816107867332813,
      "grad_norm": 9.101858139038086,
      "learning_rate": 0.00015165674622832085,
      "loss": 2.1227667331695557,
      "step": 723
    },
    {
      "epoch": 0.3486426292662373,
      "grad_norm": 17.459787368774414,
      "learning_rate": 0.0001515231194994097,
      "loss": 2.535543441772461,
      "step": 724
    },
    {
      "epoch": 0.34912417985914646,
      "grad_norm": 40.83462905883789,
      "learning_rate": 0.00015138936740360207,
      "loss": 2.485908031463623,
      "step": 725
    },
    {
      "epoch": 0.3496057304520556,
      "grad_norm": 23.384092330932617,
      "learning_rate": 0.00015125549026634585,
      "loss": 2.1880435943603516,
      "step": 726
    },
    {
      "epoch": 0.3500872810449648,
      "grad_norm": 19.147775650024414,
      "learning_rate": 0.00015112148841339295,
      "loss": 1.864107608795166,
      "step": 727
    },
    {
      "epoch": 0.35056883163787395,
      "grad_norm": 40.51984786987305,
      "learning_rate": 0.000150987362170799,
      "loss": 2.5536067485809326,
      "step": 728
    },
    {
      "epoch": 0.3510503822307831,
      "grad_norm": 36.96609115600586,
      "learning_rate": 0.00015085311186492206,
      "loss": 3.160794258117676,
      "step": 729
    },
    {
      "epoch": 0.3515319328236923,
      "grad_norm": 72.70391845703125,
      "learning_rate": 0.00015071873782242223,
      "loss": 2.856884479522705,
      "step": 730
    },
    {
      "epoch": 0.35201348341660144,
      "grad_norm": 17.10407257080078,
      "learning_rate": 0.0001505842403702606,
      "loss": 2.176137685775757,
      "step": 731
    },
    {
      "epoch": 0.3524950340095106,
      "grad_norm": 18.31989860534668,
      "learning_rate": 0.00015044961983569856,
      "loss": 2.362447738647461,
      "step": 732
    },
    {
      "epoch": 0.3529765846024198,
      "grad_norm": 18.42078399658203,
      "learning_rate": 0.00015031487654629702,
      "loss": 1.7001842260360718,
      "step": 733
    },
    {
      "epoch": 0.353458135195329,
      "grad_norm": 130.8725128173828,
      "learning_rate": 0.00015018001082991553,
      "loss": 2.425096273422241,
      "step": 734
    },
    {
      "epoch": 0.35393968578823815,
      "grad_norm": 35.70118713378906,
      "learning_rate": 0.0001500450230147116,
      "loss": 1.9881408214569092,
      "step": 735
    },
    {
      "epoch": 0.3544212363811473,
      "grad_norm": 70.96251678466797,
      "learning_rate": 0.00014990991342913974,
      "loss": 2.2264750003814697,
      "step": 736
    },
    {
      "epoch": 0.35490278697405647,
      "grad_norm": 17.593889236450195,
      "learning_rate": 0.00014977468240195084,
      "loss": 2.826266050338745,
      "step": 737
    },
    {
      "epoch": 0.35538433756696564,
      "grad_norm": 45.59511184692383,
      "learning_rate": 0.0001496393302621912,
      "loss": 2.1925604343414307,
      "step": 738
    },
    {
      "epoch": 0.3558658881598748,
      "grad_norm": 64.19400787353516,
      "learning_rate": 0.00014950385733920188,
      "loss": 2.2120072841644287,
      "step": 739
    },
    {
      "epoch": 0.35634743875278396,
      "grad_norm": 20.70261573791504,
      "learning_rate": 0.00014936826396261783,
      "loss": 2.150864362716675,
      "step": 740
    },
    {
      "epoch": 0.3568289893456931,
      "grad_norm": 12.643048286437988,
      "learning_rate": 0.00014923255046236705,
      "loss": 2.86883807182312,
      "step": 741
    },
    {
      "epoch": 0.3573105399386023,
      "grad_norm": 25.72430992126465,
      "learning_rate": 0.00014909671716866984,
      "loss": 2.5289440155029297,
      "step": 742
    },
    {
      "epoch": 0.35779209053151145,
      "grad_norm": 9.446917533874512,
      "learning_rate": 0.00014896076441203802,
      "loss": 3.018808364868164,
      "step": 743
    },
    {
      "epoch": 0.3582736411244206,
      "grad_norm": 9.55135440826416,
      "learning_rate": 0.000148824692523274,
      "loss": 2.399846076965332,
      "step": 744
    },
    {
      "epoch": 0.3587551917173298,
      "grad_norm": 19.984195709228516,
      "learning_rate": 0.0001486885018334702,
      "loss": 2.8135409355163574,
      "step": 745
    },
    {
      "epoch": 0.35923674231023894,
      "grad_norm": 11.93134593963623,
      "learning_rate": 0.00014855219267400797,
      "loss": 2.481745719909668,
      "step": 746
    },
    {
      "epoch": 0.35971829290314816,
      "grad_norm": 11.488639831542969,
      "learning_rate": 0.00014841576537655705,
      "loss": 2.4253947734832764,
      "step": 747
    },
    {
      "epoch": 0.3601998434960573,
      "grad_norm": 17.727985382080078,
      "learning_rate": 0.00014827922027307451,
      "loss": 2.4940805435180664,
      "step": 748
    },
    {
      "epoch": 0.3606813940889665,
      "grad_norm": 56.050777435302734,
      "learning_rate": 0.00014814255769580415,
      "loss": 2.178837776184082,
      "step": 749
    },
    {
      "epoch": 0.36116294468187565,
      "grad_norm": 5.134782314300537,
      "learning_rate": 0.00014800577797727558,
      "loss": 2.1701319217681885,
      "step": 750
    },
    {
      "epoch": 0.3616444952747848,
      "grad_norm": 15.442410469055176,
      "learning_rate": 0.00014786888145030343,
      "loss": 2.8307788372039795,
      "step": 751
    },
    {
      "epoch": 0.362126045867694,
      "grad_norm": 115.42985534667969,
      "learning_rate": 0.0001477318684479866,
      "loss": 2.305619239807129,
      "step": 752
    },
    {
      "epoch": 0.36260759646060314,
      "grad_norm": 8.349935531616211,
      "learning_rate": 0.00014759473930370736,
      "loss": 2.4903767108917236,
      "step": 753
    },
    {
      "epoch": 0.3630891470535123,
      "grad_norm": 7.437084197998047,
      "learning_rate": 0.0001474574943511306,
      "loss": 2.841675043106079,
      "step": 754
    },
    {
      "epoch": 0.36357069764642147,
      "grad_norm": 32.25315856933594,
      "learning_rate": 0.0001473201339242029,
      "loss": 2.8816728591918945,
      "step": 755
    },
    {
      "epoch": 0.36405224823933063,
      "grad_norm": 16.4697208404541,
      "learning_rate": 0.000147182658357152,
      "loss": 1.9870184659957886,
      "step": 756
    },
    {
      "epoch": 0.3645337988322398,
      "grad_norm": 14.342329978942871,
      "learning_rate": 0.00014704506798448566,
      "loss": 2.4430577754974365,
      "step": 757
    },
    {
      "epoch": 0.36501534942514896,
      "grad_norm": 16.125329971313477,
      "learning_rate": 0.00014690736314099101,
      "loss": 2.280200719833374,
      "step": 758
    },
    {
      "epoch": 0.3654969000180581,
      "grad_norm": 24.958593368530273,
      "learning_rate": 0.00014676954416173373,
      "loss": 2.145305633544922,
      "step": 759
    },
    {
      "epoch": 0.3659784506109673,
      "grad_norm": 15.995747566223145,
      "learning_rate": 0.00014663161138205724,
      "loss": 2.3670601844787598,
      "step": 760
    },
    {
      "epoch": 0.3664600012038765,
      "grad_norm": 16.23661231994629,
      "learning_rate": 0.00014649356513758176,
      "loss": 2.4573960304260254,
      "step": 761
    },
    {
      "epoch": 0.36694155179678567,
      "grad_norm": 66.82723999023438,
      "learning_rate": 0.00014635540576420374,
      "loss": 2.8562095165252686,
      "step": 762
    },
    {
      "epoch": 0.36742310238969483,
      "grad_norm": 47.47069549560547,
      "learning_rate": 0.0001462171335980948,
      "loss": 1.5873147249221802,
      "step": 763
    },
    {
      "epoch": 0.367904652982604,
      "grad_norm": 22.08259391784668,
      "learning_rate": 0.00014607874897570105,
      "loss": 2.616219997406006,
      "step": 764
    },
    {
      "epoch": 0.36838620357551316,
      "grad_norm": 28.182592391967773,
      "learning_rate": 0.0001459402522337422,
      "loss": 2.4972481727600098,
      "step": 765
    },
    {
      "epoch": 0.3688677541684223,
      "grad_norm": 16.64048957824707,
      "learning_rate": 0.00014580164370921078,
      "loss": 2.814821481704712,
      "step": 766
    },
    {
      "epoch": 0.3693493047613315,
      "grad_norm": 149.6190948486328,
      "learning_rate": 0.0001456629237393713,
      "loss": 2.00045108795166,
      "step": 767
    },
    {
      "epoch": 0.36983085535424065,
      "grad_norm": 20.149616241455078,
      "learning_rate": 0.00014552409266175952,
      "loss": 2.7852437496185303,
      "step": 768
    },
    {
      "epoch": 0.3703124059471498,
      "grad_norm": 14.692838668823242,
      "learning_rate": 0.00014538515081418142,
      "loss": 2.807375192642212,
      "step": 769
    },
    {
      "epoch": 0.370793956540059,
      "grad_norm": 39.71904754638672,
      "learning_rate": 0.00014524609853471264,
      "loss": 2.1538305282592773,
      "step": 770
    },
    {
      "epoch": 0.37127550713296814,
      "grad_norm": 34.077938079833984,
      "learning_rate": 0.00014510693616169741,
      "loss": 2.9155895709991455,
      "step": 771
    },
    {
      "epoch": 0.3717570577258773,
      "grad_norm": 15.514243125915527,
      "learning_rate": 0.0001449676640337479,
      "loss": 1.6634002923965454,
      "step": 772
    },
    {
      "epoch": 0.37223860831878647,
      "grad_norm": 10.669584274291992,
      "learning_rate": 0.00014482828248974335,
      "loss": 2.5936217308044434,
      "step": 773
    },
    {
      "epoch": 0.3727201589116957,
      "grad_norm": 21.374082565307617,
      "learning_rate": 0.00014468879186882916,
      "loss": 2.456894874572754,
      "step": 774
    },
    {
      "epoch": 0.37320170950460485,
      "grad_norm": 83.51744842529297,
      "learning_rate": 0.00014454919251041622,
      "loss": 3.1789278984069824,
      "step": 775
    },
    {
      "epoch": 0.373683260097514,
      "grad_norm": 33.52177810668945,
      "learning_rate": 0.00014440948475418,
      "loss": 2.933394193649292,
      "step": 776
    },
    {
      "epoch": 0.3741648106904232,
      "grad_norm": 8.093677520751953,
      "learning_rate": 0.00014426966894005966,
      "loss": 2.634166717529297,
      "step": 777
    },
    {
      "epoch": 0.37464636128333234,
      "grad_norm": 18.739912033081055,
      "learning_rate": 0.0001441297454082573,
      "loss": 2.587536334991455,
      "step": 778
    },
    {
      "epoch": 0.3751279118762415,
      "grad_norm": 30.26030731201172,
      "learning_rate": 0.00014398971449923722,
      "loss": 2.109557628631592,
      "step": 779
    },
    {
      "epoch": 0.37560946246915067,
      "grad_norm": 14.590058326721191,
      "learning_rate": 0.00014384957655372483,
      "loss": 2.113058567047119,
      "step": 780
    },
    {
      "epoch": 0.37609101306205983,
      "grad_norm": 28.56597900390625,
      "learning_rate": 0.00014370933191270617,
      "loss": 2.0218491554260254,
      "step": 781
    },
    {
      "epoch": 0.376572563654969,
      "grad_norm": 21.101346969604492,
      "learning_rate": 0.0001435689809174267,
      "loss": 1.8475074768066406,
      "step": 782
    },
    {
      "epoch": 0.37705411424787816,
      "grad_norm": 32.085758209228516,
      "learning_rate": 0.0001434285239093908,
      "loss": 2.2360904216766357,
      "step": 783
    },
    {
      "epoch": 0.3775356648407873,
      "grad_norm": 12.6206693649292,
      "learning_rate": 0.00014328796123036071,
      "loss": 2.223641872406006,
      "step": 784
    },
    {
      "epoch": 0.3780172154336965,
      "grad_norm": 22.7000675201416,
      "learning_rate": 0.0001431472932223559,
      "loss": 2.4043920040130615,
      "step": 785
    },
    {
      "epoch": 0.37849876602660565,
      "grad_norm": 11.420775413513184,
      "learning_rate": 0.00014300652022765207,
      "loss": 1.8558348417282104,
      "step": 786
    },
    {
      "epoch": 0.3789803166195148,
      "grad_norm": 20.850492477416992,
      "learning_rate": 0.00014286564258878033,
      "loss": 2.957514524459839,
      "step": 787
    },
    {
      "epoch": 0.37946186721242403,
      "grad_norm": 13.658242225646973,
      "learning_rate": 0.00014272466064852644,
      "loss": 2.7211670875549316,
      "step": 788
    },
    {
      "epoch": 0.3799434178053332,
      "grad_norm": 22.648622512817383,
      "learning_rate": 0.00014258357474993,
      "loss": 1.8853306770324707,
      "step": 789
    },
    {
      "epoch": 0.38042496839824236,
      "grad_norm": 42.14674758911133,
      "learning_rate": 0.0001424423852362835,
      "loss": 2.0882151126861572,
      "step": 790
    },
    {
      "epoch": 0.3809065189911515,
      "grad_norm": 23.28898048400879,
      "learning_rate": 0.00014230109245113158,
      "loss": 2.0655274391174316,
      "step": 791
    },
    {
      "epoch": 0.3813880695840607,
      "grad_norm": 14.7716703414917,
      "learning_rate": 0.00014215969673827018,
      "loss": 2.0541059970855713,
      "step": 792
    },
    {
      "epoch": 0.38186962017696985,
      "grad_norm": 23.05792999267578,
      "learning_rate": 0.00014201819844174564,
      "loss": 3.5902390480041504,
      "step": 793
    },
    {
      "epoch": 0.382351170769879,
      "grad_norm": 8.888059616088867,
      "learning_rate": 0.0001418765979058539,
      "loss": 2.7225797176361084,
      "step": 794
    },
    {
      "epoch": 0.3828327213627882,
      "grad_norm": 9.636628150939941,
      "learning_rate": 0.00014173489547513973,
      "loss": 1.963411808013916,
      "step": 795
    },
    {
      "epoch": 0.38331427195569734,
      "grad_norm": 21.469772338867188,
      "learning_rate": 0.00014159309149439582,
      "loss": 2.6633691787719727,
      "step": 796
    },
    {
      "epoch": 0.3837958225486065,
      "grad_norm": 6.895628929138184,
      "learning_rate": 0.00014145118630866187,
      "loss": 1.7178364992141724,
      "step": 797
    },
    {
      "epoch": 0.38427737314151567,
      "grad_norm": 83.82002258300781,
      "learning_rate": 0.000141309180263224,
      "loss": 2.3097710609436035,
      "step": 798
    },
    {
      "epoch": 0.38475892373442483,
      "grad_norm": 14.463266372680664,
      "learning_rate": 0.0001411670737036135,
      "loss": 2.166623115539551,
      "step": 799
    },
    {
      "epoch": 0.385240474327334,
      "grad_norm": 15.33006477355957,
      "learning_rate": 0.0001410248669756065,
      "loss": 3.09580135345459,
      "step": 800
    },
    {
      "epoch": 0.38572202492024316,
      "grad_norm": 17.808835983276367,
      "learning_rate": 0.00014088256042522264,
      "loss": 2.4256997108459473,
      "step": 801
    },
    {
      "epoch": 0.3862035755131524,
      "grad_norm": 21.961111068725586,
      "learning_rate": 0.00014074015439872458,
      "loss": 2.629826307296753,
      "step": 802
    },
    {
      "epoch": 0.38668512610606154,
      "grad_norm": 7.332415580749512,
      "learning_rate": 0.00014059764924261703,
      "loss": 1.9779536724090576,
      "step": 803
    },
    {
      "epoch": 0.3871666766989707,
      "grad_norm": 23.305709838867188,
      "learning_rate": 0.00014045504530364584,
      "loss": 3.416907787322998,
      "step": 804
    },
    {
      "epoch": 0.38764822729187987,
      "grad_norm": 27.28586769104004,
      "learning_rate": 0.00014031234292879725,
      "loss": 2.4080734252929688,
      "step": 805
    },
    {
      "epoch": 0.38812977788478903,
      "grad_norm": 586.978515625,
      "learning_rate": 0.00014016954246529696,
      "loss": 1.4648150205612183,
      "step": 806
    },
    {
      "epoch": 0.3886113284776982,
      "grad_norm": 20.42868423461914,
      "learning_rate": 0.00014002664426060942,
      "loss": 2.634538173675537,
      "step": 807
    },
    {
      "epoch": 0.38909287907060736,
      "grad_norm": 14.130228042602539,
      "learning_rate": 0.00013988364866243693,
      "loss": 2.2775707244873047,
      "step": 808
    },
    {
      "epoch": 0.3895744296635165,
      "grad_norm": 18.00023078918457,
      "learning_rate": 0.00013974055601871868,
      "loss": 2.241536855697632,
      "step": 809
    },
    {
      "epoch": 0.3900559802564257,
      "grad_norm": 25.267431259155273,
      "learning_rate": 0.00013959736667762998,
      "loss": 2.404820680618286,
      "step": 810
    },
    {
      "epoch": 0.39053753084933485,
      "grad_norm": 22.867725372314453,
      "learning_rate": 0.00013945408098758156,
      "loss": 1.9086689949035645,
      "step": 811
    },
    {
      "epoch": 0.391019081442244,
      "grad_norm": 26.95574951171875,
      "learning_rate": 0.0001393106992972184,
      "loss": 2.402761697769165,
      "step": 812
    },
    {
      "epoch": 0.3915006320351532,
      "grad_norm": 11.837051391601562,
      "learning_rate": 0.00013916722195541926,
      "loss": 2.540243625640869,
      "step": 813
    },
    {
      "epoch": 0.39198218262806234,
      "grad_norm": 9.721311569213867,
      "learning_rate": 0.00013902364931129557,
      "loss": 2.317450523376465,
      "step": 814
    },
    {
      "epoch": 0.39246373322097156,
      "grad_norm": 10.867044448852539,
      "learning_rate": 0.00013887998171419058,
      "loss": 1.985897183418274,
      "step": 815
    },
    {
      "epoch": 0.3929452838138807,
      "grad_norm": 21.08979034423828,
      "learning_rate": 0.00013873621951367862,
      "loss": 2.6040198802948,
      "step": 816
    },
    {
      "epoch": 0.3934268344067899,
      "grad_norm": 18.665307998657227,
      "learning_rate": 0.00013859236305956425,
      "loss": 1.7664538621902466,
      "step": 817
    },
    {
      "epoch": 0.39390838499969905,
      "grad_norm": 24.571481704711914,
      "learning_rate": 0.00013844841270188132,
      "loss": 2.99163556098938,
      "step": 818
    },
    {
      "epoch": 0.3943899355926082,
      "grad_norm": 11.726415634155273,
      "learning_rate": 0.00013830436879089228,
      "loss": 2.352592945098877,
      "step": 819
    },
    {
      "epoch": 0.3948714861855174,
      "grad_norm": 11.107429504394531,
      "learning_rate": 0.00013816023167708704,
      "loss": 2.0371978282928467,
      "step": 820
    },
    {
      "epoch": 0.39535303677842654,
      "grad_norm": 49.09531784057617,
      "learning_rate": 0.00013801600171118244,
      "loss": 2.0960886478424072,
      "step": 821
    },
    {
      "epoch": 0.3958345873713357,
      "grad_norm": 32.24345779418945,
      "learning_rate": 0.00013787167924412112,
      "loss": 2.4086012840270996,
      "step": 822
    },
    {
      "epoch": 0.39631613796424486,
      "grad_norm": 88.44732666015625,
      "learning_rate": 0.0001377272646270709,
      "loss": 2.7128047943115234,
      "step": 823
    },
    {
      "epoch": 0.396797688557154,
      "grad_norm": 16.648704528808594,
      "learning_rate": 0.00013758275821142382,
      "loss": 2.2062923908233643,
      "step": 824
    },
    {
      "epoch": 0.3972792391500632,
      "grad_norm": 19.85439682006836,
      "learning_rate": 0.00013743816034879523,
      "loss": 2.9356064796447754,
      "step": 825
    },
    {
      "epoch": 0.39776078974297235,
      "grad_norm": 193.0967559814453,
      "learning_rate": 0.000137293471391023,
      "loss": 1.6957634687423706,
      "step": 826
    },
    {
      "epoch": 0.3982423403358815,
      "grad_norm": 10.219818115234375,
      "learning_rate": 0.00013714869169016667,
      "loss": 2.03159236907959,
      "step": 827
    },
    {
      "epoch": 0.3987238909287907,
      "grad_norm": 15.834129333496094,
      "learning_rate": 0.00013700382159850656,
      "loss": 2.2972817420959473,
      "step": 828
    },
    {
      "epoch": 0.3992054415216999,
      "grad_norm": 30.749324798583984,
      "learning_rate": 0.00013685886146854297,
      "loss": 2.428724765777588,
      "step": 829
    },
    {
      "epoch": 0.39968699211460906,
      "grad_norm": 41.065269470214844,
      "learning_rate": 0.00013671381165299525,
      "loss": 2.3912770748138428,
      "step": 830
    },
    {
      "epoch": 0.4001685427075182,
      "grad_norm": 12.740118980407715,
      "learning_rate": 0.00013656867250480098,
      "loss": 2.997006893157959,
      "step": 831
    },
    {
      "epoch": 0.4006500933004274,
      "grad_norm": 31.502334594726562,
      "learning_rate": 0.00013642344437711512,
      "loss": 1.6357946395874023,
      "step": 832
    },
    {
      "epoch": 0.40113164389333655,
      "grad_norm": 22.659038543701172,
      "learning_rate": 0.00013627812762330912,
      "loss": 2.1570558547973633,
      "step": 833
    },
    {
      "epoch": 0.4016131944862457,
      "grad_norm": 15.71182632446289,
      "learning_rate": 0.00013613272259697007,
      "loss": 2.264995574951172,
      "step": 834
    },
    {
      "epoch": 0.4020947450791549,
      "grad_norm": 308.48846435546875,
      "learning_rate": 0.00013598722965189986,
      "loss": 2.365528106689453,
      "step": 835
    },
    {
      "epoch": 0.40257629567206404,
      "grad_norm": 29.754657745361328,
      "learning_rate": 0.0001358416491421143,
      "loss": 2.1980721950531006,
      "step": 836
    },
    {
      "epoch": 0.4030578462649732,
      "grad_norm": 11.028679847717285,
      "learning_rate": 0.00013569598142184225,
      "loss": 2.4445250034332275,
      "step": 837
    },
    {
      "epoch": 0.40353939685788237,
      "grad_norm": 4.830398082733154,
      "learning_rate": 0.00013555022684552483,
      "loss": 2.35137939453125,
      "step": 838
    },
    {
      "epoch": 0.40402094745079153,
      "grad_norm": 8.913195610046387,
      "learning_rate": 0.00013540438576781441,
      "loss": 2.6344826221466064,
      "step": 839
    },
    {
      "epoch": 0.4045024980437007,
      "grad_norm": 31.099693298339844,
      "learning_rate": 0.0001352584585435739,
      "loss": 2.8653483390808105,
      "step": 840
    },
    {
      "epoch": 0.40498404863660986,
      "grad_norm": 16.464357376098633,
      "learning_rate": 0.00013511244552787583,
      "loss": 2.0258588790893555,
      "step": 841
    },
    {
      "epoch": 0.405465599229519,
      "grad_norm": 29.67659568786621,
      "learning_rate": 0.00013496634707600147,
      "loss": 2.5830748081207275,
      "step": 842
    },
    {
      "epoch": 0.40594714982242824,
      "grad_norm": 1151.9818115234375,
      "learning_rate": 0.0001348201635434399,
      "loss": 2.2813544273376465,
      "step": 843
    },
    {
      "epoch": 0.4064287004153374,
      "grad_norm": 17.772857666015625,
      "learning_rate": 0.0001346738952858873,
      "loss": 2.522679567337036,
      "step": 844
    },
    {
      "epoch": 0.40691025100824657,
      "grad_norm": 9.578217506408691,
      "learning_rate": 0.000134527542659246,
      "loss": 2.860045909881592,
      "step": 845
    },
    {
      "epoch": 0.40739180160115573,
      "grad_norm": 15.167280197143555,
      "learning_rate": 0.00013438110601962362,
      "loss": 2.572575330734253,
      "step": 846
    },
    {
      "epoch": 0.4078733521940649,
      "grad_norm": 4.969293117523193,
      "learning_rate": 0.00013423458572333214,
      "loss": 1.7423328161239624,
      "step": 847
    },
    {
      "epoch": 0.40835490278697406,
      "grad_norm": 5.667901515960693,
      "learning_rate": 0.0001340879821268872,
      "loss": 2.4790377616882324,
      "step": 848
    },
    {
      "epoch": 0.4088364533798832,
      "grad_norm": 8.130163192749023,
      "learning_rate": 0.000133941295587007,
      "loss": 2.8290212154388428,
      "step": 849
    },
    {
      "epoch": 0.4093180039727924,
      "grad_norm": 15.14918041229248,
      "learning_rate": 0.00013379452646061164,
      "loss": 2.109996795654297,
      "step": 850
    },
    {
      "epoch": 0.40979955456570155,
      "grad_norm": 9.55135440826416,
      "learning_rate": 0.0001336476751048222,
      "loss": 1.988523006439209,
      "step": 851
    },
    {
      "epoch": 0.4102811051586107,
      "grad_norm": 31.71970558166504,
      "learning_rate": 0.00013350074187695979,
      "loss": 2.061418056488037,
      "step": 852
    },
    {
      "epoch": 0.4107626557515199,
      "grad_norm": 19.40265464782715,
      "learning_rate": 0.00013335372713454467,
      "loss": 1.8229784965515137,
      "step": 853
    },
    {
      "epoch": 0.41124420634442904,
      "grad_norm": 41.00231170654297,
      "learning_rate": 0.0001332066312352956,
      "loss": 1.9506301879882812,
      "step": 854
    },
    {
      "epoch": 0.4117257569373382,
      "grad_norm": 13.342862129211426,
      "learning_rate": 0.00013305945453712868,
      "loss": 2.250372886657715,
      "step": 855
    },
    {
      "epoch": 0.4122073075302474,
      "grad_norm": 14.666425704956055,
      "learning_rate": 0.0001329121973981567,
      "loss": 2.092913866043091,
      "step": 856
    },
    {
      "epoch": 0.4126888581231566,
      "grad_norm": 13.335161209106445,
      "learning_rate": 0.00013276486017668807,
      "loss": 2.378786563873291,
      "step": 857
    },
    {
      "epoch": 0.41317040871606575,
      "grad_norm": 4.384708404541016,
      "learning_rate": 0.0001326174432312262,
      "loss": 2.399697780609131,
      "step": 858
    },
    {
      "epoch": 0.4136519593089749,
      "grad_norm": 12.765717506408691,
      "learning_rate": 0.00013246994692046836,
      "loss": 2.048110246658325,
      "step": 859
    },
    {
      "epoch": 0.4141335099018841,
      "grad_norm": 12.102499961853027,
      "learning_rate": 0.000132322371603305,
      "loss": 2.6647229194641113,
      "step": 860
    },
    {
      "epoch": 0.41461506049479324,
      "grad_norm": 14.99820327758789,
      "learning_rate": 0.0001321747176388188,
      "loss": 2.2723073959350586,
      "step": 861
    },
    {
      "epoch": 0.4150966110877024,
      "grad_norm": 10.078459739685059,
      "learning_rate": 0.00013202698538628376,
      "loss": 2.254176616668701,
      "step": 862
    },
    {
      "epoch": 0.41557816168061157,
      "grad_norm": 14.237257957458496,
      "learning_rate": 0.00013187917520516448,
      "loss": 3.346674919128418,
      "step": 863
    },
    {
      "epoch": 0.41605971227352073,
      "grad_norm": 85.713623046875,
      "learning_rate": 0.00013173128745511508,
      "loss": 2.8501322269439697,
      "step": 864
    },
    {
      "epoch": 0.4165412628664299,
      "grad_norm": 100.66522979736328,
      "learning_rate": 0.0001315833224959784,
      "loss": 1.996762990951538,
      "step": 865
    },
    {
      "epoch": 0.41702281345933906,
      "grad_norm": 42.97366714477539,
      "learning_rate": 0.00013143528068778525,
      "loss": 2.5755796432495117,
      "step": 866
    },
    {
      "epoch": 0.4175043640522482,
      "grad_norm": 15.18427848815918,
      "learning_rate": 0.00013128716239075338,
      "loss": 2.6287522315979004,
      "step": 867
    },
    {
      "epoch": 0.4179859146451574,
      "grad_norm": 12.193682670593262,
      "learning_rate": 0.00013113896796528664,
      "loss": 2.8639004230499268,
      "step": 868
    },
    {
      "epoch": 0.41846746523806655,
      "grad_norm": 5.41328239440918,
      "learning_rate": 0.00013099069777197412,
      "loss": 2.111783504486084,
      "step": 869
    },
    {
      "epoch": 0.41894901583097577,
      "grad_norm": 13.271903991699219,
      "learning_rate": 0.0001308423521715893,
      "loss": 2.8832991123199463,
      "step": 870
    },
    {
      "epoch": 0.41943056642388493,
      "grad_norm": 6.5283894538879395,
      "learning_rate": 0.00013069393152508906,
      "loss": 2.6641082763671875,
      "step": 871
    },
    {
      "epoch": 0.4199121170167941,
      "grad_norm": 9.133336067199707,
      "learning_rate": 0.00013054543619361303,
      "loss": 2.508265495300293,
      "step": 872
    },
    {
      "epoch": 0.42039366760970326,
      "grad_norm": 5.490320205688477,
      "learning_rate": 0.0001303968665384824,
      "loss": 2.2276337146759033,
      "step": 873
    },
    {
      "epoch": 0.4208752182026124,
      "grad_norm": 12.389824867248535,
      "learning_rate": 0.00013024822292119934,
      "loss": 2.3230903148651123,
      "step": 874
    },
    {
      "epoch": 0.4213567687955216,
      "grad_norm": 8.775754928588867,
      "learning_rate": 0.0001300995057034459,
      "loss": 2.082286834716797,
      "step": 875
    },
    {
      "epoch": 0.42183831938843075,
      "grad_norm": 11.6084623336792,
      "learning_rate": 0.00012995071524708325,
      "loss": 2.3789405822753906,
      "step": 876
    },
    {
      "epoch": 0.4223198699813399,
      "grad_norm": 21.294368743896484,
      "learning_rate": 0.00012980185191415074,
      "loss": 3.0745584964752197,
      "step": 877
    },
    {
      "epoch": 0.4228014205742491,
      "grad_norm": 438.5638427734375,
      "learning_rate": 0.0001296529160668651,
      "loss": 2.4127461910247803,
      "step": 878
    },
    {
      "epoch": 0.42328297116715824,
      "grad_norm": 13.651358604431152,
      "learning_rate": 0.00012950390806761944,
      "loss": 2.5501036643981934,
      "step": 879
    },
    {
      "epoch": 0.4237645217600674,
      "grad_norm": 11.225810050964355,
      "learning_rate": 0.0001293548282789825,
      "loss": 2.051784038543701,
      "step": 880
    },
    {
      "epoch": 0.42424607235297657,
      "grad_norm": 5.239315509796143,
      "learning_rate": 0.00012920567706369758,
      "loss": 2.0852131843566895,
      "step": 881
    },
    {
      "epoch": 0.42472762294588573,
      "grad_norm": 18.22057342529297,
      "learning_rate": 0.00012905645478468192,
      "loss": 2.672658920288086,
      "step": 882
    },
    {
      "epoch": 0.4252091735387949,
      "grad_norm": 5.371629238128662,
      "learning_rate": 0.00012890716180502564,
      "loss": 2.5662004947662354,
      "step": 883
    },
    {
      "epoch": 0.4256907241317041,
      "grad_norm": 6.654219627380371,
      "learning_rate": 0.00012875779848799078,
      "loss": 2.8605830669403076,
      "step": 884
    },
    {
      "epoch": 0.4261722747246133,
      "grad_norm": 8.148382186889648,
      "learning_rate": 0.00012860836519701063,
      "loss": 2.4015510082244873,
      "step": 885
    },
    {
      "epoch": 0.42665382531752244,
      "grad_norm": 14.335244178771973,
      "learning_rate": 0.00012845886229568873,
      "loss": 2.0592806339263916,
      "step": 886
    },
    {
      "epoch": 0.4271353759104316,
      "grad_norm": 34.31779479980469,
      "learning_rate": 0.00012830929014779797,
      "loss": 2.384749174118042,
      "step": 887
    },
    {
      "epoch": 0.42761692650334077,
      "grad_norm": 16.25575828552246,
      "learning_rate": 0.0001281596491172797,
      "loss": 2.1389267444610596,
      "step": 888
    },
    {
      "epoch": 0.42809847709624993,
      "grad_norm": 7.066389560699463,
      "learning_rate": 0.00012800993956824303,
      "loss": 2.5966358184814453,
      "step": 889
    },
    {
      "epoch": 0.4285800276891591,
      "grad_norm": 13.552995681762695,
      "learning_rate": 0.00012786016186496358,
      "loss": 2.123291254043579,
      "step": 890
    },
    {
      "epoch": 0.42906157828206826,
      "grad_norm": 19.40772819519043,
      "learning_rate": 0.000127710316371883,
      "loss": 1.6868525743484497,
      "step": 891
    },
    {
      "epoch": 0.4295431288749774,
      "grad_norm": 3.586195707321167,
      "learning_rate": 0.0001275604034536077,
      "loss": 1.600808024406433,
      "step": 892
    },
    {
      "epoch": 0.4300246794678866,
      "grad_norm": 65.8642807006836,
      "learning_rate": 0.0001274104234749083,
      "loss": 2.675398588180542,
      "step": 893
    },
    {
      "epoch": 0.43050623006079575,
      "grad_norm": 13.059523582458496,
      "learning_rate": 0.00012726037680071853,
      "loss": 2.656344413757324,
      "step": 894
    },
    {
      "epoch": 0.4309877806537049,
      "grad_norm": 14.47943115234375,
      "learning_rate": 0.00012711026379613434,
      "loss": 2.341548204421997,
      "step": 895
    },
    {
      "epoch": 0.4314693312466141,
      "grad_norm": 13.900928497314453,
      "learning_rate": 0.00012696008482641325,
      "loss": 1.7322189807891846,
      "step": 896
    },
    {
      "epoch": 0.43195088183952324,
      "grad_norm": 27.031618118286133,
      "learning_rate": 0.00012680984025697313,
      "loss": 2.6577975749969482,
      "step": 897
    },
    {
      "epoch": 0.43243243243243246,
      "grad_norm": 30.76296043395996,
      "learning_rate": 0.00012665953045339152,
      "loss": 3.0694146156311035,
      "step": 898
    },
    {
      "epoch": 0.4329139830253416,
      "grad_norm": 22.99774742126465,
      "learning_rate": 0.0001265091557814047,
      "loss": 2.7756118774414062,
      "step": 899
    },
    {
      "epoch": 0.4333955336182508,
      "grad_norm": 88.32677459716797,
      "learning_rate": 0.00012635871660690676,
      "loss": 2.4822280406951904,
      "step": 900
    },
    {
      "epoch": 0.43387708421115995,
      "grad_norm": 29.441482543945312,
      "learning_rate": 0.0001262082132959488,
      "loss": 2.13531231880188,
      "step": 901
    },
    {
      "epoch": 0.4343586348040691,
      "grad_norm": 58.75855255126953,
      "learning_rate": 0.00012605764621473792,
      "loss": 1.9812266826629639,
      "step": 902
    },
    {
      "epoch": 0.4348401853969783,
      "grad_norm": 9.035697937011719,
      "learning_rate": 0.00012590701572963642,
      "loss": 2.1329078674316406,
      "step": 903
    },
    {
      "epoch": 0.43532173598988744,
      "grad_norm": 12.136872291564941,
      "learning_rate": 0.00012575632220716078,
      "loss": 2.112855911254883,
      "step": 904
    },
    {
      "epoch": 0.4358032865827966,
      "grad_norm": 38.1376953125,
      "learning_rate": 0.000125605566013981,
      "loss": 2.0239322185516357,
      "step": 905
    },
    {
      "epoch": 0.43628483717570576,
      "grad_norm": 20.527809143066406,
      "learning_rate": 0.00012545474751691953,
      "loss": 2.439727306365967,
      "step": 906
    },
    {
      "epoch": 0.4367663877686149,
      "grad_norm": 15.7946138381958,
      "learning_rate": 0.00012530386708295036,
      "loss": 2.478649616241455,
      "step": 907
    },
    {
      "epoch": 0.4372479383615241,
      "grad_norm": 27.246219635009766,
      "learning_rate": 0.00012515292507919829,
      "loss": 2.0023107528686523,
      "step": 908
    },
    {
      "epoch": 0.43772948895443325,
      "grad_norm": 112.68732452392578,
      "learning_rate": 0.0001250019218729378,
      "loss": 1.441702127456665,
      "step": 909
    },
    {
      "epoch": 0.4382110395473424,
      "grad_norm": 31.121158599853516,
      "learning_rate": 0.00012485085783159238,
      "loss": 1.7150264978408813,
      "step": 910
    },
    {
      "epoch": 0.43869259014025164,
      "grad_norm": 15.044858932495117,
      "learning_rate": 0.00012469973332273354,
      "loss": 2.258483648300171,
      "step": 911
    },
    {
      "epoch": 0.4391741407331608,
      "grad_norm": 17.50322151184082,
      "learning_rate": 0.00012454854871407994,
      "loss": 2.0657384395599365,
      "step": 912
    },
    {
      "epoch": 0.43965569132606996,
      "grad_norm": 17.459611892700195,
      "learning_rate": 0.00012439730437349635,
      "loss": 2.3094117641448975,
      "step": 913
    },
    {
      "epoch": 0.4401372419189791,
      "grad_norm": 62.13080978393555,
      "learning_rate": 0.00012424600066899302,
      "loss": 2.4538960456848145,
      "step": 914
    },
    {
      "epoch": 0.4406187925118883,
      "grad_norm": 30.682697296142578,
      "learning_rate": 0.00012409463796872464,
      "loss": 2.938070774078369,
      "step": 915
    },
    {
      "epoch": 0.44110034310479745,
      "grad_norm": 9.966513633728027,
      "learning_rate": 0.0001239432166409893,
      "loss": 2.743199348449707,
      "step": 916
    },
    {
      "epoch": 0.4415818936977066,
      "grad_norm": 18.706790924072266,
      "learning_rate": 0.00012379173705422795,
      "loss": 2.24320650100708,
      "step": 917
    },
    {
      "epoch": 0.4420634442906158,
      "grad_norm": 27.022436141967773,
      "learning_rate": 0.00012364019957702315,
      "loss": 1.7564663887023926,
      "step": 918
    },
    {
      "epoch": 0.44254499488352494,
      "grad_norm": 19.064210891723633,
      "learning_rate": 0.00012348860457809838,
      "loss": 2.70522403717041,
      "step": 919
    },
    {
      "epoch": 0.4430265454764341,
      "grad_norm": 6.989287376403809,
      "learning_rate": 0.00012333695242631705,
      "loss": 2.180572271347046,
      "step": 920
    },
    {
      "epoch": 0.44350809606934327,
      "grad_norm": 23.664546966552734,
      "learning_rate": 0.0001231852434906817,
      "loss": 2.6627821922302246,
      "step": 921
    },
    {
      "epoch": 0.44398964666225244,
      "grad_norm": 12.833931922912598,
      "learning_rate": 0.00012303347814033292,
      "loss": 2.548715591430664,
      "step": 922
    },
    {
      "epoch": 0.4444711972551616,
      "grad_norm": 38.04713821411133,
      "learning_rate": 0.0001228816567445487,
      "loss": 2.3360087871551514,
      "step": 923
    },
    {
      "epoch": 0.44495274784807076,
      "grad_norm": 14.112283706665039,
      "learning_rate": 0.0001227297796727433,
      "loss": 2.655651092529297,
      "step": 924
    },
    {
      "epoch": 0.44543429844098,
      "grad_norm": 14.717061996459961,
      "learning_rate": 0.00012257784729446656,
      "loss": 2.6265709400177,
      "step": 925
    },
    {
      "epoch": 0.44591584903388914,
      "grad_norm": 10.041119575500488,
      "learning_rate": 0.00012242585997940275,
      "loss": 1.9705994129180908,
      "step": 926
    },
    {
      "epoch": 0.4463973996267983,
      "grad_norm": 21.85250473022461,
      "learning_rate": 0.0001222738180973699,
      "loss": 2.6686758995056152,
      "step": 927
    },
    {
      "epoch": 0.44687895021970747,
      "grad_norm": 14.835731506347656,
      "learning_rate": 0.00012212172201831885,
      "loss": 2.396178960800171,
      "step": 928
    },
    {
      "epoch": 0.44736050081261663,
      "grad_norm": 118.12613677978516,
      "learning_rate": 0.00012196957211233222,
      "loss": 2.7011897563934326,
      "step": 929
    },
    {
      "epoch": 0.4478420514055258,
      "grad_norm": 40.325199127197266,
      "learning_rate": 0.00012181736874962371,
      "loss": 2.9641497135162354,
      "step": 930
    },
    {
      "epoch": 0.44832360199843496,
      "grad_norm": 7.144264221191406,
      "learning_rate": 0.00012166511230053696,
      "loss": 2.3032116889953613,
      "step": 931
    },
    {
      "epoch": 0.4488051525913441,
      "grad_norm": 92.3143539428711,
      "learning_rate": 0.00012151280313554486,
      "loss": 3.064533233642578,
      "step": 932
    },
    {
      "epoch": 0.4492867031842533,
      "grad_norm": 6.987591743469238,
      "learning_rate": 0.00012136044162524858,
      "loss": 2.5664448738098145,
      "step": 933
    },
    {
      "epoch": 0.44976825377716245,
      "grad_norm": 24.731054306030273,
      "learning_rate": 0.00012120802814037663,
      "loss": 3.236140251159668,
      "step": 934
    },
    {
      "epoch": 0.4502498043700716,
      "grad_norm": 11.102771759033203,
      "learning_rate": 0.00012105556305178399,
      "loss": 2.7384438514709473,
      "step": 935
    },
    {
      "epoch": 0.4507313549629808,
      "grad_norm": 14.167506217956543,
      "learning_rate": 0.00012090304673045123,
      "loss": 2.658315658569336,
      "step": 936
    },
    {
      "epoch": 0.45121290555588994,
      "grad_norm": 307.52490234375,
      "learning_rate": 0.00012075047954748353,
      "loss": 1.7295807600021362,
      "step": 937
    },
    {
      "epoch": 0.4516944561487991,
      "grad_norm": 27.391090393066406,
      "learning_rate": 0.00012059786187410984,
      "loss": 2.1294455528259277,
      "step": 938
    },
    {
      "epoch": 0.4521760067417083,
      "grad_norm": 5.7963361740112305,
      "learning_rate": 0.000120445194081682,
      "loss": 2.3103768825531006,
      "step": 939
    },
    {
      "epoch": 0.4526575573346175,
      "grad_norm": 28.18410873413086,
      "learning_rate": 0.00012029247654167379,
      "loss": 2.244744300842285,
      "step": 940
    },
    {
      "epoch": 0.45313910792752665,
      "grad_norm": 18.265235900878906,
      "learning_rate": 0.00012013970962568002,
      "loss": 2.048414468765259,
      "step": 941
    },
    {
      "epoch": 0.4536206585204358,
      "grad_norm": 19.082666397094727,
      "learning_rate": 0.00011998689370541562,
      "loss": 2.2295174598693848,
      "step": 942
    },
    {
      "epoch": 0.454102209113345,
      "grad_norm": 17.45819091796875,
      "learning_rate": 0.00011983402915271478,
      "loss": 1.5896775722503662,
      "step": 943
    },
    {
      "epoch": 0.45458375970625414,
      "grad_norm": 12.403421401977539,
      "learning_rate": 0.00011968111633953007,
      "loss": 2.1518304347991943,
      "step": 944
    },
    {
      "epoch": 0.4550653102991633,
      "grad_norm": 8.7493896484375,
      "learning_rate": 0.0001195281556379314,
      "loss": 2.2050747871398926,
      "step": 945
    },
    {
      "epoch": 0.45554686089207247,
      "grad_norm": 20.03658676147461,
      "learning_rate": 0.0001193751474201053,
      "loss": 2.042210340499878,
      "step": 946
    },
    {
      "epoch": 0.45602841148498163,
      "grad_norm": 70.87582397460938,
      "learning_rate": 0.00011922209205835382,
      "loss": 3.0684869289398193,
      "step": 947
    },
    {
      "epoch": 0.4565099620778908,
      "grad_norm": 28.550127029418945,
      "learning_rate": 0.0001190689899250938,
      "loss": 2.484306812286377,
      "step": 948
    },
    {
      "epoch": 0.45699151267079996,
      "grad_norm": 34.30080795288086,
      "learning_rate": 0.00011891584139285582,
      "loss": 2.4512619972229004,
      "step": 949
    },
    {
      "epoch": 0.4574730632637091,
      "grad_norm": 64.99056243896484,
      "learning_rate": 0.00011876264683428344,
      "loss": 2.263272285461426,
      "step": 950
    },
    {
      "epoch": 0.4579546138566183,
      "grad_norm": 62.16273880004883,
      "learning_rate": 0.00011860940662213211,
      "loss": 1.8559774160385132,
      "step": 951
    },
    {
      "epoch": 0.4584361644495275,
      "grad_norm": 6.9525227546691895,
      "learning_rate": 0.00011845612112926843,
      "loss": 2.4261887073516846,
      "step": 952
    },
    {
      "epoch": 0.45891771504243667,
      "grad_norm": 22.440385818481445,
      "learning_rate": 0.00011830279072866921,
      "loss": 2.091871976852417,
      "step": 953
    },
    {
      "epoch": 0.45939926563534583,
      "grad_norm": 66.49950408935547,
      "learning_rate": 0.00011814941579342044,
      "loss": 1.9058862924575806,
      "step": 954
    },
    {
      "epoch": 0.459880816228255,
      "grad_norm": 11.461359024047852,
      "learning_rate": 0.00011799599669671654,
      "loss": 2.2700350284576416,
      "step": 955
    },
    {
      "epoch": 0.46036236682116416,
      "grad_norm": 9.563632011413574,
      "learning_rate": 0.00011784253381185937,
      "loss": 1.712271809577942,
      "step": 956
    },
    {
      "epoch": 0.4608439174140733,
      "grad_norm": 34.22868728637695,
      "learning_rate": 0.0001176890275122573,
      "loss": 1.7839194536209106,
      "step": 957
    },
    {
      "epoch": 0.4613254680069825,
      "grad_norm": 8.948943138122559,
      "learning_rate": 0.0001175354781714244,
      "loss": 2.019385814666748,
      "step": 958
    },
    {
      "epoch": 0.46180701859989165,
      "grad_norm": 305.67852783203125,
      "learning_rate": 0.0001173818861629794,
      "loss": 2.1160688400268555,
      "step": 959
    },
    {
      "epoch": 0.4622885691928008,
      "grad_norm": 23.723508834838867,
      "learning_rate": 0.00011722825186064494,
      "loss": 2.2922606468200684,
      "step": 960
    },
    {
      "epoch": 0.46277011978571,
      "grad_norm": 57.50545883178711,
      "learning_rate": 0.00011707457563824646,
      "loss": 2.5574254989624023,
      "step": 961
    },
    {
      "epoch": 0.46325167037861914,
      "grad_norm": 20.3013858795166,
      "learning_rate": 0.00011692085786971149,
      "loss": 2.1685080528259277,
      "step": 962
    },
    {
      "epoch": 0.4637332209715283,
      "grad_norm": 260.6160583496094,
      "learning_rate": 0.00011676709892906858,
      "loss": 2.5958447456359863,
      "step": 963
    },
    {
      "epoch": 0.46421477156443747,
      "grad_norm": 9.925374984741211,
      "learning_rate": 0.00011661329919044656,
      "loss": 2.3834147453308105,
      "step": 964
    },
    {
      "epoch": 0.46469632215734663,
      "grad_norm": 49.93943405151367,
      "learning_rate": 0.00011645945902807341,
      "loss": 2.1057956218719482,
      "step": 965
    },
    {
      "epoch": 0.46517787275025585,
      "grad_norm": 11.001801490783691,
      "learning_rate": 0.00011630557881627553,
      "loss": 2.2969653606414795,
      "step": 966
    },
    {
      "epoch": 0.465659423343165,
      "grad_norm": 43.308013916015625,
      "learning_rate": 0.0001161516589294768,
      "loss": 2.217454433441162,
      "step": 967
    },
    {
      "epoch": 0.4661409739360742,
      "grad_norm": 18.335887908935547,
      "learning_rate": 0.00011599769974219757,
      "loss": 2.01397705078125,
      "step": 968
    },
    {
      "epoch": 0.46662252452898334,
      "grad_norm": 13.926067352294922,
      "learning_rate": 0.0001158437016290539,
      "loss": 1.9933595657348633,
      "step": 969
    },
    {
      "epoch": 0.4671040751218925,
      "grad_norm": 239.49549865722656,
      "learning_rate": 0.00011568966496475649,
      "loss": 2.079679489135742,
      "step": 970
    },
    {
      "epoch": 0.46758562571480167,
      "grad_norm": 47.9632568359375,
      "learning_rate": 0.00011553559012410984,
      "loss": 2.8512816429138184,
      "step": 971
    },
    {
      "epoch": 0.46806717630771083,
      "grad_norm": 37.513328552246094,
      "learning_rate": 0.00011538147748201138,
      "loss": 2.730964422225952,
      "step": 972
    },
    {
      "epoch": 0.46854872690062,
      "grad_norm": 24.78883934020996,
      "learning_rate": 0.00011522732741345053,
      "loss": 1.613823413848877,
      "step": 973
    },
    {
      "epoch": 0.46903027749352916,
      "grad_norm": 19.514179229736328,
      "learning_rate": 0.00011507314029350776,
      "loss": 2.799879789352417,
      "step": 974
    },
    {
      "epoch": 0.4695118280864383,
      "grad_norm": 19.122217178344727,
      "learning_rate": 0.00011491891649735366,
      "loss": 2.1543126106262207,
      "step": 975
    },
    {
      "epoch": 0.4699933786793475,
      "grad_norm": 15.700153350830078,
      "learning_rate": 0.00011476465640024814,
      "loss": 2.402622699737549,
      "step": 976
    },
    {
      "epoch": 0.47047492927225665,
      "grad_norm": 588.669921875,
      "learning_rate": 0.00011461036037753934,
      "loss": 1.793561339378357,
      "step": 977
    },
    {
      "epoch": 0.4709564798651658,
      "grad_norm": 74.76912689208984,
      "learning_rate": 0.00011445602880466288,
      "loss": 1.7009739875793457,
      "step": 978
    },
    {
      "epoch": 0.471438030458075,
      "grad_norm": 72.33135223388672,
      "learning_rate": 0.00011430166205714088,
      "loss": 2.2983901500701904,
      "step": 979
    },
    {
      "epoch": 0.4719195810509842,
      "grad_norm": 4.52475118637085,
      "learning_rate": 0.00011414726051058102,
      "loss": 1.5396496057510376,
      "step": 980
    },
    {
      "epoch": 0.47240113164389336,
      "grad_norm": 70.50374603271484,
      "learning_rate": 0.0001139928245406757,
      "loss": 2.1088833808898926,
      "step": 981
    },
    {
      "epoch": 0.4728826822368025,
      "grad_norm": 28.2486629486084,
      "learning_rate": 0.00011383835452320097,
      "loss": 2.2650277614593506,
      "step": 982
    },
    {
      "epoch": 0.4733642328297117,
      "grad_norm": 12.153045654296875,
      "learning_rate": 0.00011368385083401585,
      "loss": 2.6818063259124756,
      "step": 983
    },
    {
      "epoch": 0.47384578342262085,
      "grad_norm": 7.406557559967041,
      "learning_rate": 0.00011352931384906125,
      "loss": 2.0958433151245117,
      "step": 984
    },
    {
      "epoch": 0.47432733401553,
      "grad_norm": 220.32781982421875,
      "learning_rate": 0.00011337474394435908,
      "loss": 2.6583778858184814,
      "step": 985
    },
    {
      "epoch": 0.4748088846084392,
      "grad_norm": 34.111080169677734,
      "learning_rate": 0.00011322014149601136,
      "loss": 2.329210042953491,
      "step": 986
    },
    {
      "epoch": 0.47529043520134834,
      "grad_norm": 14.60641098022461,
      "learning_rate": 0.00011306550688019926,
      "loss": 2.480710983276367,
      "step": 987
    },
    {
      "epoch": 0.4757719857942575,
      "grad_norm": 55.09360885620117,
      "learning_rate": 0.0001129108404731823,
      "loss": 2.8529584407806396,
      "step": 988
    },
    {
      "epoch": 0.47625353638716666,
      "grad_norm": 123.96564483642578,
      "learning_rate": 0.0001127561426512973,
      "loss": 2.336360216140747,
      "step": 989
    },
    {
      "epoch": 0.47673508698007583,
      "grad_norm": 12.115889549255371,
      "learning_rate": 0.0001126014137909575,
      "loss": 1.9523794651031494,
      "step": 990
    },
    {
      "epoch": 0.477216637572985,
      "grad_norm": 56.36611557006836,
      "learning_rate": 0.00011244665426865174,
      "loss": 2.7924561500549316,
      "step": 991
    },
    {
      "epoch": 0.47769818816589416,
      "grad_norm": 6.836651802062988,
      "learning_rate": 0.00011229186446094338,
      "loss": 1.8846291303634644,
      "step": 992
    },
    {
      "epoch": 0.4781797387588034,
      "grad_norm": 36.05104446411133,
      "learning_rate": 0.00011213704474446951,
      "loss": 2.4663965702056885,
      "step": 993
    },
    {
      "epoch": 0.47866128935171254,
      "grad_norm": 12.921553611755371,
      "learning_rate": 0.00011198219549594,
      "loss": 2.0567374229431152,
      "step": 994
    },
    {
      "epoch": 0.4791428399446217,
      "grad_norm": 7.4218668937683105,
      "learning_rate": 0.00011182731709213659,
      "loss": 2.3970751762390137,
      "step": 995
    },
    {
      "epoch": 0.47962439053753086,
      "grad_norm": 85.6276626586914,
      "learning_rate": 0.00011167240990991192,
      "loss": 2.3039000034332275,
      "step": 996
    },
    {
      "epoch": 0.48010594113044003,
      "grad_norm": 44.052364349365234,
      "learning_rate": 0.00011151747432618871,
      "loss": 1.9650468826293945,
      "step": 997
    },
    {
      "epoch": 0.4805874917233492,
      "grad_norm": 21.3747615814209,
      "learning_rate": 0.00011136251071795871,
      "loss": 2.4129350185394287,
      "step": 998
    },
    {
      "epoch": 0.48106904231625836,
      "grad_norm": 17.667707443237305,
      "learning_rate": 0.00011120751946228197,
      "loss": 2.1272449493408203,
      "step": 999
    },
    {
      "epoch": 0.4815505929091675,
      "grad_norm": 126.1726303100586,
      "learning_rate": 0.00011105250093628565,
      "loss": 2.43666410446167,
      "step": 1000
    },
    {
      "epoch": 0.4820321435020767,
      "grad_norm": 26.33510971069336,
      "learning_rate": 0.00011089745551716344,
      "loss": 2.119425058364868,
      "step": 1001
    },
    {
      "epoch": 0.48251369409498585,
      "grad_norm": 96.45536041259766,
      "learning_rate": 0.00011074238358217437,
      "loss": 2.7094638347625732,
      "step": 1002
    },
    {
      "epoch": 0.482995244687895,
      "grad_norm": 14.70875358581543,
      "learning_rate": 0.00011058728550864197,
      "loss": 1.7299500703811646,
      "step": 1003
    },
    {
      "epoch": 0.4834767952808042,
      "grad_norm": 18.052946090698242,
      "learning_rate": 0.00011043216167395344,
      "loss": 1.8247722387313843,
      "step": 1004
    },
    {
      "epoch": 0.48395834587371334,
      "grad_norm": 11.908700942993164,
      "learning_rate": 0.00011027701245555865,
      "loss": 2.276327610015869,
      "step": 1005
    },
    {
      "epoch": 0.4844398964666225,
      "grad_norm": 15.220044136047363,
      "learning_rate": 0.00011012183823096917,
      "loss": 2.194338321685791,
      "step": 1006
    },
    {
      "epoch": 0.4849214470595317,
      "grad_norm": 9.63715648651123,
      "learning_rate": 0.00010996663937775751,
      "loss": 2.72568416595459,
      "step": 1007
    },
    {
      "epoch": 0.4854029976524409,
      "grad_norm": 21.990446090698242,
      "learning_rate": 0.000109811416273556,
      "loss": 2.666494607925415,
      "step": 1008
    },
    {
      "epoch": 0.48588454824535005,
      "grad_norm": 23.3886661529541,
      "learning_rate": 0.00010965616929605609,
      "loss": 2.148258924484253,
      "step": 1009
    },
    {
      "epoch": 0.4863660988382592,
      "grad_norm": 26.94173812866211,
      "learning_rate": 0.0001095008988230072,
      "loss": 2.8899612426757812,
      "step": 1010
    },
    {
      "epoch": 0.48684764943116837,
      "grad_norm": 53.144126892089844,
      "learning_rate": 0.00010934560523221602,
      "loss": 2.8107616901397705,
      "step": 1011
    },
    {
      "epoch": 0.48732920002407754,
      "grad_norm": 23.97219467163086,
      "learning_rate": 0.00010919028890154543,
      "loss": 2.254169225692749,
      "step": 1012
    },
    {
      "epoch": 0.4878107506169867,
      "grad_norm": 26.7773494720459,
      "learning_rate": 0.00010903495020891375,
      "loss": 2.380323648452759,
      "step": 1013
    },
    {
      "epoch": 0.48829230120989586,
      "grad_norm": 45.04411315917969,
      "learning_rate": 0.00010887958953229349,
      "loss": 2.098320484161377,
      "step": 1014
    },
    {
      "epoch": 0.488773851802805,
      "grad_norm": 14.55712890625,
      "learning_rate": 0.00010872420724971088,
      "loss": 1.9085264205932617,
      "step": 1015
    },
    {
      "epoch": 0.4892554023957142,
      "grad_norm": 10.270770072937012,
      "learning_rate": 0.0001085688037392446,
      "loss": 2.722369909286499,
      "step": 1016
    },
    {
      "epoch": 0.48973695298862335,
      "grad_norm": 16.86016845703125,
      "learning_rate": 0.000108413379379025,
      "loss": 2.6092162132263184,
      "step": 1017
    },
    {
      "epoch": 0.4902185035815325,
      "grad_norm": 22.261260986328125,
      "learning_rate": 0.00010825793454723325,
      "loss": 2.252892255783081,
      "step": 1018
    },
    {
      "epoch": 0.4907000541744417,
      "grad_norm": 11.824828147888184,
      "learning_rate": 0.00010810246962210018,
      "loss": 2.0573577880859375,
      "step": 1019
    },
    {
      "epoch": 0.49118160476735084,
      "grad_norm": 4.87546968460083,
      "learning_rate": 0.00010794698498190557,
      "loss": 2.0614213943481445,
      "step": 1020
    },
    {
      "epoch": 0.49166315536026006,
      "grad_norm": 10.471446990966797,
      "learning_rate": 0.00010779148100497722,
      "loss": 2.5306975841522217,
      "step": 1021
    },
    {
      "epoch": 0.4921447059531692,
      "grad_norm": 153.94094848632812,
      "learning_rate": 0.00010763595806968996,
      "loss": 2.5121841430664062,
      "step": 1022
    },
    {
      "epoch": 0.4926262565460784,
      "grad_norm": 16.671825408935547,
      "learning_rate": 0.00010748041655446473,
      "loss": 2.9045732021331787,
      "step": 1023
    },
    {
      "epoch": 0.49310780713898755,
      "grad_norm": 19.37563133239746,
      "learning_rate": 0.00010732485683776768,
      "loss": 2.5483720302581787,
      "step": 1024
    },
    {
      "epoch": 0.4935893577318967,
      "grad_norm": 9.075236320495605,
      "learning_rate": 0.00010716927929810925,
      "loss": 2.929577350616455,
      "step": 1025
    },
    {
      "epoch": 0.4940709083248059,
      "grad_norm": 299.20208740234375,
      "learning_rate": 0.00010701368431404326,
      "loss": 2.8736696243286133,
      "step": 1026
    },
    {
      "epoch": 0.49455245891771504,
      "grad_norm": 19.924327850341797,
      "learning_rate": 0.00010685807226416598,
      "loss": 2.3752894401550293,
      "step": 1027
    },
    {
      "epoch": 0.4950340095106242,
      "grad_norm": 28.035192489624023,
      "learning_rate": 0.00010670244352711518,
      "loss": 2.8525238037109375,
      "step": 1028
    },
    {
      "epoch": 0.49551556010353337,
      "grad_norm": 10.351886749267578,
      "learning_rate": 0.00010654679848156925,
      "loss": 2.157212972640991,
      "step": 1029
    },
    {
      "epoch": 0.49599711069644253,
      "grad_norm": 15.769464492797852,
      "learning_rate": 0.00010639113750624625,
      "loss": 2.7796597480773926,
      "step": 1030
    },
    {
      "epoch": 0.4964786612893517,
      "grad_norm": 9.208419799804688,
      "learning_rate": 0.00010623546097990303,
      "loss": 2.3438267707824707,
      "step": 1031
    },
    {
      "epoch": 0.49696021188226086,
      "grad_norm": 74.24439239501953,
      "learning_rate": 0.00010607976928133423,
      "loss": 1.6457449197769165,
      "step": 1032
    },
    {
      "epoch": 0.49744176247517,
      "grad_norm": 9.602882385253906,
      "learning_rate": 0.00010592406278937144,
      "loss": 2.065162420272827,
      "step": 1033
    },
    {
      "epoch": 0.4979233130680792,
      "grad_norm": 25.36962127685547,
      "learning_rate": 0.00010576834188288226,
      "loss": 2.0549800395965576,
      "step": 1034
    },
    {
      "epoch": 0.4984048636609884,
      "grad_norm": 9.105463981628418,
      "learning_rate": 0.00010561260694076935,
      "loss": 2.628776788711548,
      "step": 1035
    },
    {
      "epoch": 0.49888641425389757,
      "grad_norm": 7.268791198730469,
      "learning_rate": 0.00010545685834196948,
      "loss": 2.1897194385528564,
      "step": 1036
    },
    {
      "epoch": 0.49936796484680673,
      "grad_norm": 27.030635833740234,
      "learning_rate": 0.00010530109646545272,
      "loss": 1.9632889032363892,
      "step": 1037
    },
    {
      "epoch": 0.4998495154397159,
      "grad_norm": 24.189130783081055,
      "learning_rate": 0.0001051453216902214,
      "loss": 2.6585946083068848,
      "step": 1038
    },
    {
      "epoch": 0.500331066032625,
      "grad_norm": 73.73881530761719,
      "learning_rate": 0.00010498953439530925,
      "loss": 2.2996621131896973,
      "step": 1039
    },
    {
      "epoch": 0.5008126166255342,
      "grad_norm": 13.143025398254395,
      "learning_rate": 0.00010483373495978046,
      "loss": 1.3206506967544556,
      "step": 1040
    },
    {
      "epoch": 0.5012941672184434,
      "grad_norm": 14.0916109085083,
      "learning_rate": 0.00010467792376272877,
      "loss": 2.0801286697387695,
      "step": 1041
    },
    {
      "epoch": 0.5017757178113526,
      "grad_norm": 67.77770233154297,
      "learning_rate": 0.00010452210118327652,
      "loss": 2.437566041946411,
      "step": 1042
    },
    {
      "epoch": 0.5022572684042618,
      "grad_norm": 14.923233032226562,
      "learning_rate": 0.00010436626760057378,
      "loss": 2.824202060699463,
      "step": 1043
    },
    {
      "epoch": 0.5027388189971709,
      "grad_norm": 42.17008972167969,
      "learning_rate": 0.00010421042339379732,
      "loss": 2.692499876022339,
      "step": 1044
    },
    {
      "epoch": 0.5032203695900801,
      "grad_norm": 11.313698768615723,
      "learning_rate": 0.00010405456894214987,
      "loss": 2.0081417560577393,
      "step": 1045
    },
    {
      "epoch": 0.5037019201829892,
      "grad_norm": 43.75627899169922,
      "learning_rate": 0.00010389870462485902,
      "loss": 2.93949556350708,
      "step": 1046
    },
    {
      "epoch": 0.5041834707758984,
      "grad_norm": 635.4976806640625,
      "learning_rate": 0.00010374283082117635,
      "loss": 2.0673253536224365,
      "step": 1047
    },
    {
      "epoch": 0.5046650213688075,
      "grad_norm": 38.93379592895508,
      "learning_rate": 0.00010358694791037653,
      "loss": 3.0953595638275146,
      "step": 1048
    },
    {
      "epoch": 0.5051465719617168,
      "grad_norm": 55.024993896484375,
      "learning_rate": 0.00010343105627175644,
      "loss": 2.4686226844787598,
      "step": 1049
    },
    {
      "epoch": 0.5056281225546259,
      "grad_norm": 57.147247314453125,
      "learning_rate": 0.00010327515628463415,
      "loss": 2.0493650436401367,
      "step": 1050
    },
    {
      "epoch": 0.5061096731475351,
      "grad_norm": 76.14037322998047,
      "learning_rate": 0.00010311924832834808,
      "loss": 2.1856329441070557,
      "step": 1051
    },
    {
      "epoch": 0.5065912237404442,
      "grad_norm": 12.523856163024902,
      "learning_rate": 0.00010296333278225599,
      "loss": 1.5865777730941772,
      "step": 1052
    },
    {
      "epoch": 0.5070727743333534,
      "grad_norm": 11.923735618591309,
      "learning_rate": 0.00010280741002573413,
      "loss": 1.8631356954574585,
      "step": 1053
    },
    {
      "epoch": 0.5075543249262625,
      "grad_norm": 3.8487656116485596,
      "learning_rate": 0.00010265148043817632,
      "loss": 2.393988847732544,
      "step": 1054
    },
    {
      "epoch": 0.5080358755191717,
      "grad_norm": 30.52573585510254,
      "learning_rate": 0.00010249554439899298,
      "loss": 2.166527032852173,
      "step": 1055
    },
    {
      "epoch": 0.508517426112081,
      "grad_norm": 215.092529296875,
      "learning_rate": 0.00010233960228761022,
      "loss": 2.506967306137085,
      "step": 1056
    },
    {
      "epoch": 0.5089989767049901,
      "grad_norm": 6.124822616577148,
      "learning_rate": 0.00010218365448346893,
      "loss": 2.3026435375213623,
      "step": 1057
    },
    {
      "epoch": 0.5094805272978993,
      "grad_norm": 10.063334465026855,
      "learning_rate": 0.00010202770136602388,
      "loss": 3.072728157043457,
      "step": 1058
    },
    {
      "epoch": 0.5099620778908084,
      "grad_norm": 17.063600540161133,
      "learning_rate": 0.00010187174331474271,
      "loss": 2.6237363815307617,
      "step": 1059
    },
    {
      "epoch": 0.5104436284837176,
      "grad_norm": 13.365060806274414,
      "learning_rate": 0.00010171578070910512,
      "loss": 2.419801712036133,
      "step": 1060
    },
    {
      "epoch": 0.5109251790766267,
      "grad_norm": 7.213529109954834,
      "learning_rate": 0.00010155981392860185,
      "loss": 2.3244476318359375,
      "step": 1061
    },
    {
      "epoch": 0.5114067296695359,
      "grad_norm": 9.195516586303711,
      "learning_rate": 0.00010140384335273386,
      "loss": 2.021839141845703,
      "step": 1062
    },
    {
      "epoch": 0.511888280262445,
      "grad_norm": 95.70538330078125,
      "learning_rate": 0.00010124786936101127,
      "loss": 2.0325417518615723,
      "step": 1063
    },
    {
      "epoch": 0.5123698308553543,
      "grad_norm": 143.5034942626953,
      "learning_rate": 0.00010109189233295255,
      "loss": 3.029468536376953,
      "step": 1064
    },
    {
      "epoch": 0.5128513814482634,
      "grad_norm": 32.80686950683594,
      "learning_rate": 0.00010093591264808358,
      "loss": 2.3047916889190674,
      "step": 1065
    },
    {
      "epoch": 0.5133329320411726,
      "grad_norm": 39.095333099365234,
      "learning_rate": 0.00010077993068593663,
      "loss": 3.141740560531616,
      "step": 1066
    },
    {
      "epoch": 0.5138144826340817,
      "grad_norm": 21.843677520751953,
      "learning_rate": 0.00010062394682604963,
      "loss": 1.9767414331436157,
      "step": 1067
    },
    {
      "epoch": 0.5142960332269909,
      "grad_norm": 14.495436668395996,
      "learning_rate": 0.00010046796144796497,
      "loss": 2.7554240226745605,
      "step": 1068
    },
    {
      "epoch": 0.5147775838199001,
      "grad_norm": 50.05131149291992,
      "learning_rate": 0.0001003119749312289,
      "loss": 2.4667556285858154,
      "step": 1069
    },
    {
      "epoch": 0.5152591344128092,
      "grad_norm": 17.913272857666016,
      "learning_rate": 0.00010015598765539031,
      "loss": 2.9713399410247803,
      "step": 1070
    },
    {
      "epoch": 0.5157406850057185,
      "grad_norm": 16.660213470458984,
      "learning_rate": 0.0001,
      "loss": 2.682767868041992,
      "step": 1071
    },
    {
      "epoch": 0.5162222355986276,
      "grad_norm": 26.415983200073242,
      "learning_rate": 9.984401234460971e-05,
      "loss": 2.1377644538879395,
      "step": 1072
    },
    {
      "epoch": 0.5167037861915368,
      "grad_norm": 7.351304054260254,
      "learning_rate": 9.968802506877111e-05,
      "loss": 2.595102071762085,
      "step": 1073
    },
    {
      "epoch": 0.5171853367844459,
      "grad_norm": 53.990108489990234,
      "learning_rate": 9.953203855203504e-05,
      "loss": 2.812002182006836,
      "step": 1074
    },
    {
      "epoch": 0.5176668873773551,
      "grad_norm": 7.367271423339844,
      "learning_rate": 9.93760531739504e-05,
      "loss": 2.2550337314605713,
      "step": 1075
    },
    {
      "epoch": 0.5181484379702642,
      "grad_norm": 4.550349235534668,
      "learning_rate": 9.922006931406338e-05,
      "loss": 2.805056571960449,
      "step": 1076
    },
    {
      "epoch": 0.5186299885631734,
      "grad_norm": 25.182273864746094,
      "learning_rate": 9.906408735191643e-05,
      "loss": 2.605013847351074,
      "step": 1077
    },
    {
      "epoch": 0.5191115391560825,
      "grad_norm": 275.34112548828125,
      "learning_rate": 9.890810766704745e-05,
      "loss": 3.015632390975952,
      "step": 1078
    },
    {
      "epoch": 0.5195930897489918,
      "grad_norm": 16.621885299682617,
      "learning_rate": 9.875213063898875e-05,
      "loss": 2.248640537261963,
      "step": 1079
    },
    {
      "epoch": 0.5200746403419009,
      "grad_norm": 140.3821563720703,
      "learning_rate": 9.859615664726615e-05,
      "loss": 2.376784563064575,
      "step": 1080
    },
    {
      "epoch": 0.5205561909348101,
      "grad_norm": 19.14274024963379,
      "learning_rate": 9.844018607139818e-05,
      "loss": 2.1443281173706055,
      "step": 1081
    },
    {
      "epoch": 0.5210377415277193,
      "grad_norm": 21.490278244018555,
      "learning_rate": 9.828421929089493e-05,
      "loss": 1.8448046445846558,
      "step": 1082
    },
    {
      "epoch": 0.5215192921206284,
      "grad_norm": 11.920368194580078,
      "learning_rate": 9.812825668525733e-05,
      "loss": 2.0224578380584717,
      "step": 1083
    },
    {
      "epoch": 0.5220008427135376,
      "grad_norm": 102.5870132446289,
      "learning_rate": 9.797229863397615e-05,
      "loss": 2.3114168643951416,
      "step": 1084
    },
    {
      "epoch": 0.5224823933064467,
      "grad_norm": 14.616949081420898,
      "learning_rate": 9.781634551653108e-05,
      "loss": 2.142085552215576,
      "step": 1085
    },
    {
      "epoch": 0.522963943899356,
      "grad_norm": 249.16409301757812,
      "learning_rate": 9.766039771238982e-05,
      "loss": 2.8898019790649414,
      "step": 1086
    },
    {
      "epoch": 0.5234454944922651,
      "grad_norm": 9.112092018127441,
      "learning_rate": 9.750445560100706e-05,
      "loss": 2.402754306793213,
      "step": 1087
    },
    {
      "epoch": 0.5239270450851743,
      "grad_norm": 12.344494819641113,
      "learning_rate": 9.73485195618237e-05,
      "loss": 2.041764259338379,
      "step": 1088
    },
    {
      "epoch": 0.5244085956780834,
      "grad_norm": 13.685338020324707,
      "learning_rate": 9.719258997426588e-05,
      "loss": 2.0068838596343994,
      "step": 1089
    },
    {
      "epoch": 0.5248901462709926,
      "grad_norm": 12.23843002319336,
      "learning_rate": 9.703666721774402e-05,
      "loss": 2.8421456813812256,
      "step": 1090
    },
    {
      "epoch": 0.5253716968639017,
      "grad_norm": 81.18212890625,
      "learning_rate": 9.688075167165194e-05,
      "loss": 2.072566032409668,
      "step": 1091
    },
    {
      "epoch": 0.525853247456811,
      "grad_norm": 59.159061431884766,
      "learning_rate": 9.672484371536586e-05,
      "loss": 2.743121862411499,
      "step": 1092
    },
    {
      "epoch": 0.52633479804972,
      "grad_norm": 17.69476890563965,
      "learning_rate": 9.656894372824358e-05,
      "loss": 1.9757683277130127,
      "step": 1093
    },
    {
      "epoch": 0.5268163486426293,
      "grad_norm": 11.421501159667969,
      "learning_rate": 9.64130520896235e-05,
      "loss": 2.378312110900879,
      "step": 1094
    },
    {
      "epoch": 0.5272978992355384,
      "grad_norm": 8.225211143493652,
      "learning_rate": 9.625716917882367e-05,
      "loss": 1.6604636907577515,
      "step": 1095
    },
    {
      "epoch": 0.5277794498284476,
      "grad_norm": 5.353133201599121,
      "learning_rate": 9.6101295375141e-05,
      "loss": 1.9079499244689941,
      "step": 1096
    },
    {
      "epoch": 0.5282610004213568,
      "grad_norm": 67.28994750976562,
      "learning_rate": 9.594543105785013e-05,
      "loss": 1.875327706336975,
      "step": 1097
    },
    {
      "epoch": 0.5287425510142659,
      "grad_norm": 37.35131072998047,
      "learning_rate": 9.578957660620267e-05,
      "loss": 2.832167863845825,
      "step": 1098
    },
    {
      "epoch": 0.5292241016071751,
      "grad_norm": 21.422325134277344,
      "learning_rate": 9.563373239942623e-05,
      "loss": 2.799801826477051,
      "step": 1099
    },
    {
      "epoch": 0.5297056522000843,
      "grad_norm": 83.00056457519531,
      "learning_rate": 9.547789881672348e-05,
      "loss": 1.931170105934143,
      "step": 1100
    },
    {
      "epoch": 0.5301872027929935,
      "grad_norm": 184.89781188964844,
      "learning_rate": 9.532207623727126e-05,
      "loss": 2.8245348930358887,
      "step": 1101
    },
    {
      "epoch": 0.5306687533859026,
      "grad_norm": 11.138921737670898,
      "learning_rate": 9.516626504021957e-05,
      "loss": 2.20916485786438,
      "step": 1102
    },
    {
      "epoch": 0.5311503039788118,
      "grad_norm": 111.35883331298828,
      "learning_rate": 9.501046560469079e-05,
      "loss": 2.6226935386657715,
      "step": 1103
    },
    {
      "epoch": 0.5316318545717209,
      "grad_norm": 12.155516624450684,
      "learning_rate": 9.485467830977864e-05,
      "loss": 2.528757333755493,
      "step": 1104
    },
    {
      "epoch": 0.5321134051646301,
      "grad_norm": 18.68534278869629,
      "learning_rate": 9.469890353454732e-05,
      "loss": 2.338724136352539,
      "step": 1105
    },
    {
      "epoch": 0.5325949557575392,
      "grad_norm": 82.87825775146484,
      "learning_rate": 9.454314165803054e-05,
      "loss": 2.3277747631073,
      "step": 1106
    },
    {
      "epoch": 0.5330765063504485,
      "grad_norm": 9.790543556213379,
      "learning_rate": 9.438739305923067e-05,
      "loss": 1.976706862449646,
      "step": 1107
    },
    {
      "epoch": 0.5335580569433576,
      "grad_norm": 6.1930460929870605,
      "learning_rate": 9.423165811711777e-05,
      "loss": 2.2161691188812256,
      "step": 1108
    },
    {
      "epoch": 0.5340396075362668,
      "grad_norm": 9.372110366821289,
      "learning_rate": 9.407593721062859e-05,
      "loss": 1.8378986120224,
      "step": 1109
    },
    {
      "epoch": 0.534521158129176,
      "grad_norm": 9.762017250061035,
      "learning_rate": 9.39202307186658e-05,
      "loss": 2.2324860095977783,
      "step": 1110
    },
    {
      "epoch": 0.5350027087220851,
      "grad_norm": 19.02521514892578,
      "learning_rate": 9.3764539020097e-05,
      "loss": 2.7592709064483643,
      "step": 1111
    },
    {
      "epoch": 0.5354842593149943,
      "grad_norm": 83.5306396484375,
      "learning_rate": 9.360886249375376e-05,
      "loss": 2.804290771484375,
      "step": 1112
    },
    {
      "epoch": 0.5359658099079034,
      "grad_norm": 47.548683166503906,
      "learning_rate": 9.345320151843078e-05,
      "loss": 2.207322120666504,
      "step": 1113
    },
    {
      "epoch": 0.5364473605008127,
      "grad_norm": 36.81177520751953,
      "learning_rate": 9.329755647288485e-05,
      "loss": 2.427548885345459,
      "step": 1114
    },
    {
      "epoch": 0.5369289110937218,
      "grad_norm": 15.521409034729004,
      "learning_rate": 9.314192773583403e-05,
      "loss": 3.2903988361358643,
      "step": 1115
    },
    {
      "epoch": 0.537410461686631,
      "grad_norm": 315.6177062988281,
      "learning_rate": 9.298631568595674e-05,
      "loss": 2.166135311126709,
      "step": 1116
    },
    {
      "epoch": 0.5378920122795401,
      "grad_norm": 10.520930290222168,
      "learning_rate": 9.283072070189075e-05,
      "loss": 1.6131629943847656,
      "step": 1117
    },
    {
      "epoch": 0.5383735628724493,
      "grad_norm": 256.9414978027344,
      "learning_rate": 9.267514316223234e-05,
      "loss": 2.0874032974243164,
      "step": 1118
    },
    {
      "epoch": 0.5388551134653584,
      "grad_norm": 13.544004440307617,
      "learning_rate": 9.251958344553528e-05,
      "loss": 1.860609769821167,
      "step": 1119
    },
    {
      "epoch": 0.5393366640582676,
      "grad_norm": 37.13266372680664,
      "learning_rate": 9.23640419303101e-05,
      "loss": 1.7161972522735596,
      "step": 1120
    },
    {
      "epoch": 0.5398182146511767,
      "grad_norm": 10.960495948791504,
      "learning_rate": 9.220851899502283e-05,
      "loss": 2.4221668243408203,
      "step": 1121
    },
    {
      "epoch": 0.540299765244086,
      "grad_norm": 91.92877197265625,
      "learning_rate": 9.205301501809448e-05,
      "loss": 2.094351291656494,
      "step": 1122
    },
    {
      "epoch": 0.5407813158369951,
      "grad_norm": 10.657490730285645,
      "learning_rate": 9.189753037789987e-05,
      "loss": 2.1763787269592285,
      "step": 1123
    },
    {
      "epoch": 0.5412628664299043,
      "grad_norm": 14.443318367004395,
      "learning_rate": 9.174206545276677e-05,
      "loss": 2.2091174125671387,
      "step": 1124
    },
    {
      "epoch": 0.5417444170228135,
      "grad_norm": 5.839054107666016,
      "learning_rate": 9.158662062097501e-05,
      "loss": 1.8781640529632568,
      "step": 1125
    },
    {
      "epoch": 0.5422259676157226,
      "grad_norm": 9.624420166015625,
      "learning_rate": 9.143119626075542e-05,
      "loss": 1.8980016708374023,
      "step": 1126
    },
    {
      "epoch": 0.5427075182086318,
      "grad_norm": 4.277372360229492,
      "learning_rate": 9.127579275028914e-05,
      "loss": 2.38729190826416,
      "step": 1127
    },
    {
      "epoch": 0.5431890688015409,
      "grad_norm": 9.327408790588379,
      "learning_rate": 9.112041046770653e-05,
      "loss": 1.9484585523605347,
      "step": 1128
    },
    {
      "epoch": 0.5436706193944502,
      "grad_norm": 29.489200592041016,
      "learning_rate": 9.096504979108629e-05,
      "loss": 2.467000961303711,
      "step": 1129
    },
    {
      "epoch": 0.5441521699873593,
      "grad_norm": 17.470821380615234,
      "learning_rate": 9.080971109845458e-05,
      "loss": 2.3266167640686035,
      "step": 1130
    },
    {
      "epoch": 0.5446337205802685,
      "grad_norm": 13.80547046661377,
      "learning_rate": 9.0654394767784e-05,
      "loss": 2.1969544887542725,
      "step": 1131
    },
    {
      "epoch": 0.5451152711731776,
      "grad_norm": 15.331083297729492,
      "learning_rate": 9.049910117699281e-05,
      "loss": 2.2655296325683594,
      "step": 1132
    },
    {
      "epoch": 0.5455968217660868,
      "grad_norm": 9.45826530456543,
      "learning_rate": 9.034383070394393e-05,
      "loss": 2.561169147491455,
      "step": 1133
    },
    {
      "epoch": 0.5460783723589959,
      "grad_norm": 142.43203735351562,
      "learning_rate": 9.0188583726444e-05,
      "loss": 2.171473503112793,
      "step": 1134
    },
    {
      "epoch": 0.5465599229519051,
      "grad_norm": 26.193525314331055,
      "learning_rate": 9.00333606222425e-05,
      "loss": 2.1737046241760254,
      "step": 1135
    },
    {
      "epoch": 0.5470414735448142,
      "grad_norm": 14.36005973815918,
      "learning_rate": 8.987816176903082e-05,
      "loss": 3.196882486343384,
      "step": 1136
    },
    {
      "epoch": 0.5475230241377235,
      "grad_norm": 46.5828971862793,
      "learning_rate": 8.972298754444136e-05,
      "loss": 1.9005335569381714,
      "step": 1137
    },
    {
      "epoch": 0.5480045747306327,
      "grad_norm": 8.25678825378418,
      "learning_rate": 8.956783832604654e-05,
      "loss": 1.8309731483459473,
      "step": 1138
    },
    {
      "epoch": 0.5484861253235418,
      "grad_norm": 24.233068466186523,
      "learning_rate": 8.941271449135806e-05,
      "loss": 2.3136003017425537,
      "step": 1139
    },
    {
      "epoch": 0.548967675916451,
      "grad_norm": 38.34886169433594,
      "learning_rate": 8.925761641782567e-05,
      "loss": 3.1326913833618164,
      "step": 1140
    },
    {
      "epoch": 0.5494492265093601,
      "grad_norm": 12.302376747131348,
      "learning_rate": 8.910254448283659e-05,
      "loss": 2.333984613418579,
      "step": 1141
    },
    {
      "epoch": 0.5499307771022693,
      "grad_norm": 8.601741790771484,
      "learning_rate": 8.894749906371439e-05,
      "loss": 2.7756214141845703,
      "step": 1142
    },
    {
      "epoch": 0.5504123276951784,
      "grad_norm": 16.34593391418457,
      "learning_rate": 8.87924805377181e-05,
      "loss": 2.1010594367980957,
      "step": 1143
    },
    {
      "epoch": 0.5508938782880877,
      "grad_norm": 33.453853607177734,
      "learning_rate": 8.863748928204131e-05,
      "loss": 2.0740132331848145,
      "step": 1144
    },
    {
      "epoch": 0.5513754288809968,
      "grad_norm": 16.957557678222656,
      "learning_rate": 8.848252567381131e-05,
      "loss": 2.153198719024658,
      "step": 1145
    },
    {
      "epoch": 0.551856979473906,
      "grad_norm": 15.135165214538574,
      "learning_rate": 8.83275900900881e-05,
      "loss": 2.2636942863464355,
      "step": 1146
    },
    {
      "epoch": 0.5523385300668151,
      "grad_norm": 7.14637565612793,
      "learning_rate": 8.817268290786343e-05,
      "loss": 2.5485894680023193,
      "step": 1147
    },
    {
      "epoch": 0.5528200806597243,
      "grad_norm": 12.22465991973877,
      "learning_rate": 8.801780450406002e-05,
      "loss": 2.3168957233428955,
      "step": 1148
    },
    {
      "epoch": 0.5533016312526334,
      "grad_norm": 6.009751796722412,
      "learning_rate": 8.786295525553053e-05,
      "loss": 2.1653571128845215,
      "step": 1149
    },
    {
      "epoch": 0.5537831818455426,
      "grad_norm": 16.84650230407715,
      "learning_rate": 8.770813553905664e-05,
      "loss": 2.2020506858825684,
      "step": 1150
    },
    {
      "epoch": 0.5542647324384519,
      "grad_norm": 7.872344017028809,
      "learning_rate": 8.755334573134829e-05,
      "loss": 1.6681760549545288,
      "step": 1151
    },
    {
      "epoch": 0.554746283031361,
      "grad_norm": 10.908254623413086,
      "learning_rate": 8.739858620904251e-05,
      "loss": 2.308558940887451,
      "step": 1152
    },
    {
      "epoch": 0.5552278336242702,
      "grad_norm": 24.181278228759766,
      "learning_rate": 8.724385734870271e-05,
      "loss": 2.236781597137451,
      "step": 1153
    },
    {
      "epoch": 0.5557093842171793,
      "grad_norm": 24.284093856811523,
      "learning_rate": 8.708915952681769e-05,
      "loss": 2.577540397644043,
      "step": 1154
    },
    {
      "epoch": 0.5561909348100885,
      "grad_norm": 21.613914489746094,
      "learning_rate": 8.693449311980074e-05,
      "loss": 2.1722052097320557,
      "step": 1155
    },
    {
      "epoch": 0.5566724854029976,
      "grad_norm": 17.819786071777344,
      "learning_rate": 8.677985850398866e-05,
      "loss": 2.567808151245117,
      "step": 1156
    },
    {
      "epoch": 0.5571540359959068,
      "grad_norm": 19.339096069335938,
      "learning_rate": 8.662525605564093e-05,
      "loss": 2.529391288757324,
      "step": 1157
    },
    {
      "epoch": 0.557635586588816,
      "grad_norm": 45.50886917114258,
      "learning_rate": 8.647068615093875e-05,
      "loss": 2.6208953857421875,
      "step": 1158
    },
    {
      "epoch": 0.5581171371817252,
      "grad_norm": 25.579593658447266,
      "learning_rate": 8.631614916598419e-05,
      "loss": 2.090017557144165,
      "step": 1159
    },
    {
      "epoch": 0.5585986877746343,
      "grad_norm": 56.835105895996094,
      "learning_rate": 8.616164547679906e-05,
      "loss": 2.8545312881469727,
      "step": 1160
    },
    {
      "epoch": 0.5590802383675435,
      "grad_norm": 23.898399353027344,
      "learning_rate": 8.600717545932435e-05,
      "loss": 2.417771577835083,
      "step": 1161
    },
    {
      "epoch": 0.5595617889604526,
      "grad_norm": 31.614198684692383,
      "learning_rate": 8.5852739489419e-05,
      "loss": 2.7640137672424316,
      "step": 1162
    },
    {
      "epoch": 0.5600433395533618,
      "grad_norm": 13.714906692504883,
      "learning_rate": 8.569833794285915e-05,
      "loss": 2.402658224105835,
      "step": 1163
    },
    {
      "epoch": 0.5605248901462709,
      "grad_norm": 27.37522315979004,
      "learning_rate": 8.554397119533714e-05,
      "loss": 2.2738587856292725,
      "step": 1164
    },
    {
      "epoch": 0.5610064407391802,
      "grad_norm": 12.01988697052002,
      "learning_rate": 8.538963962246069e-05,
      "loss": 2.182497262954712,
      "step": 1165
    },
    {
      "epoch": 0.5614879913320894,
      "grad_norm": 19.965551376342773,
      "learning_rate": 8.523534359975189e-05,
      "loss": 2.6488656997680664,
      "step": 1166
    },
    {
      "epoch": 0.5619695419249985,
      "grad_norm": 51.41749572753906,
      "learning_rate": 8.508108350264635e-05,
      "loss": 2.0909788608551025,
      "step": 1167
    },
    {
      "epoch": 0.5624510925179077,
      "grad_norm": 13.603612899780273,
      "learning_rate": 8.492685970649228e-05,
      "loss": 2.569422483444214,
      "step": 1168
    },
    {
      "epoch": 0.5629326431108168,
      "grad_norm": 185.65110778808594,
      "learning_rate": 8.477267258654949e-05,
      "loss": 2.494154214859009,
      "step": 1169
    },
    {
      "epoch": 0.563414193703726,
      "grad_norm": 7.727108955383301,
      "learning_rate": 8.461852251798866e-05,
      "loss": 2.8486297130584717,
      "step": 1170
    },
    {
      "epoch": 0.5638957442966351,
      "grad_norm": 12.928702354431152,
      "learning_rate": 8.44644098758902e-05,
      "loss": 2.13787841796875,
      "step": 1171
    },
    {
      "epoch": 0.5643772948895444,
      "grad_norm": 11.850486755371094,
      "learning_rate": 8.431033503524354e-05,
      "loss": 2.4282002449035645,
      "step": 1172
    },
    {
      "epoch": 0.5648588454824535,
      "grad_norm": 35.56986618041992,
      "learning_rate": 8.415629837094611e-05,
      "loss": 2.6985409259796143,
      "step": 1173
    },
    {
      "epoch": 0.5653403960753627,
      "grad_norm": 57.927207946777344,
      "learning_rate": 8.400230025780243e-05,
      "loss": 2.546330451965332,
      "step": 1174
    },
    {
      "epoch": 0.5658219466682718,
      "grad_norm": 65.2827377319336,
      "learning_rate": 8.384834107052321e-05,
      "loss": 3.10028076171875,
      "step": 1175
    },
    {
      "epoch": 0.566303497261181,
      "grad_norm": 21.35835838317871,
      "learning_rate": 8.369442118372447e-05,
      "loss": 2.818431854248047,
      "step": 1176
    },
    {
      "epoch": 0.5667850478540901,
      "grad_norm": 62.85612869262695,
      "learning_rate": 8.35405409719266e-05,
      "loss": 1.923595905303955,
      "step": 1177
    },
    {
      "epoch": 0.5672665984469993,
      "grad_norm": 17.9860897064209,
      "learning_rate": 8.338670080955349e-05,
      "loss": 1.8890471458435059,
      "step": 1178
    },
    {
      "epoch": 0.5677481490399086,
      "grad_norm": 21.004173278808594,
      "learning_rate": 8.323290107093143e-05,
      "loss": 2.0957984924316406,
      "step": 1179
    },
    {
      "epoch": 0.5682296996328177,
      "grad_norm": 17.85140609741211,
      "learning_rate": 8.307914213028856e-05,
      "loss": 1.8637080192565918,
      "step": 1180
    },
    {
      "epoch": 0.5687112502257269,
      "grad_norm": 12.667609214782715,
      "learning_rate": 8.292542436175356e-05,
      "loss": 2.3848049640655518,
      "step": 1181
    },
    {
      "epoch": 0.569192800818636,
      "grad_norm": 8.292073249816895,
      "learning_rate": 8.277174813935508e-05,
      "loss": 2.461505651473999,
      "step": 1182
    },
    {
      "epoch": 0.5696743514115452,
      "grad_norm": 11.332146644592285,
      "learning_rate": 8.261811383702061e-05,
      "loss": 1.6659163236618042,
      "step": 1183
    },
    {
      "epoch": 0.5701559020044543,
      "grad_norm": 31.406423568725586,
      "learning_rate": 8.246452182857562e-05,
      "loss": 2.4935760498046875,
      "step": 1184
    },
    {
      "epoch": 0.5706374525973635,
      "grad_norm": 7.594135761260986,
      "learning_rate": 8.231097248774274e-05,
      "loss": 2.3467965126037598,
      "step": 1185
    },
    {
      "epoch": 0.5711190031902726,
      "grad_norm": 73.69046783447266,
      "learning_rate": 8.215746618814067e-05,
      "loss": 2.4363956451416016,
      "step": 1186
    },
    {
      "epoch": 0.5716005537831819,
      "grad_norm": 44.238258361816406,
      "learning_rate": 8.200400330328348e-05,
      "loss": 2.927931070327759,
      "step": 1187
    },
    {
      "epoch": 0.572082104376091,
      "grad_norm": 7.729905128479004,
      "learning_rate": 8.185058420657957e-05,
      "loss": 2.2292041778564453,
      "step": 1188
    },
    {
      "epoch": 0.5725636549690002,
      "grad_norm": 8.371865272521973,
      "learning_rate": 8.16972092713308e-05,
      "loss": 2.128082036972046,
      "step": 1189
    },
    {
      "epoch": 0.5730452055619093,
      "grad_norm": 7.899158954620361,
      "learning_rate": 8.154387887073158e-05,
      "loss": 2.339620590209961,
      "step": 1190
    },
    {
      "epoch": 0.5735267561548185,
      "grad_norm": 22.610212326049805,
      "learning_rate": 8.139059337786792e-05,
      "loss": 1.4548091888427734,
      "step": 1191
    },
    {
      "epoch": 0.5740083067477277,
      "grad_norm": 17.726255416870117,
      "learning_rate": 8.12373531657166e-05,
      "loss": 2.3227202892303467,
      "step": 1192
    },
    {
      "epoch": 0.5744898573406368,
      "grad_norm": 33.47049331665039,
      "learning_rate": 8.108415860714418e-05,
      "loss": 2.4262101650238037,
      "step": 1193
    },
    {
      "epoch": 0.5749714079335461,
      "grad_norm": 16.705562591552734,
      "learning_rate": 8.093101007490622e-05,
      "loss": 2.299403190612793,
      "step": 1194
    },
    {
      "epoch": 0.5754529585264552,
      "grad_norm": 24.114139556884766,
      "learning_rate": 8.077790794164619e-05,
      "loss": 2.352832555770874,
      "step": 1195
    },
    {
      "epoch": 0.5759345091193644,
      "grad_norm": 38.8360481262207,
      "learning_rate": 8.062485257989471e-05,
      "loss": 2.5595381259918213,
      "step": 1196
    },
    {
      "epoch": 0.5764160597122735,
      "grad_norm": 8.444920539855957,
      "learning_rate": 8.047184436206864e-05,
      "loss": 2.0756032466888428,
      "step": 1197
    },
    {
      "epoch": 0.5768976103051827,
      "grad_norm": 151.50033569335938,
      "learning_rate": 8.031888366046998e-05,
      "loss": 2.380545139312744,
      "step": 1198
    },
    {
      "epoch": 0.5773791608980918,
      "grad_norm": 8.535476684570312,
      "learning_rate": 8.016597084728526e-05,
      "loss": 1.7740936279296875,
      "step": 1199
    },
    {
      "epoch": 0.577860711491001,
      "grad_norm": 24.515899658203125,
      "learning_rate": 8.001310629458443e-05,
      "loss": 2.376180648803711,
      "step": 1200
    },
    {
      "epoch": 0.5783422620839102,
      "grad_norm": 33.11871337890625,
      "learning_rate": 7.986029037432002e-05,
      "loss": 2.862229347229004,
      "step": 1201
    },
    {
      "epoch": 0.5788238126768194,
      "grad_norm": 69.29415893554688,
      "learning_rate": 7.970752345832623e-05,
      "loss": 2.6108827590942383,
      "step": 1202
    },
    {
      "epoch": 0.5793053632697285,
      "grad_norm": 119.43251037597656,
      "learning_rate": 7.9554805918318e-05,
      "loss": 2.069953203201294,
      "step": 1203
    },
    {
      "epoch": 0.5797869138626377,
      "grad_norm": 45.29293441772461,
      "learning_rate": 7.940213812589018e-05,
      "loss": 2.4665067195892334,
      "step": 1204
    },
    {
      "epoch": 0.5802684644555468,
      "grad_norm": 22.256900787353516,
      "learning_rate": 7.92495204525165e-05,
      "loss": 1.1562796831130981,
      "step": 1205
    },
    {
      "epoch": 0.580750015048456,
      "grad_norm": 11.430119514465332,
      "learning_rate": 7.909695326954878e-05,
      "loss": 2.3145864009857178,
      "step": 1206
    },
    {
      "epoch": 0.5812315656413652,
      "grad_norm": 37.21303176879883,
      "learning_rate": 7.894443694821602e-05,
      "loss": 2.3835883140563965,
      "step": 1207
    },
    {
      "epoch": 0.5817131162342744,
      "grad_norm": 30.77141761779785,
      "learning_rate": 7.879197185962339e-05,
      "loss": 2.1142942905426025,
      "step": 1208
    },
    {
      "epoch": 0.5821946668271836,
      "grad_norm": 16.251977920532227,
      "learning_rate": 7.863955837475144e-05,
      "loss": 2.4438607692718506,
      "step": 1209
    },
    {
      "epoch": 0.5826762174200927,
      "grad_norm": 10.522878646850586,
      "learning_rate": 7.848719686445515e-05,
      "loss": 2.1993446350097656,
      "step": 1210
    },
    {
      "epoch": 0.5831577680130019,
      "grad_norm": 39.63413619995117,
      "learning_rate": 7.833488769946306e-05,
      "loss": 2.6261885166168213,
      "step": 1211
    },
    {
      "epoch": 0.583639318605911,
      "grad_norm": 51.13256072998047,
      "learning_rate": 7.818263125037633e-05,
      "loss": 2.2882232666015625,
      "step": 1212
    },
    {
      "epoch": 0.5841208691988202,
      "grad_norm": 30.826757431030273,
      "learning_rate": 7.803042788766777e-05,
      "loss": 2.432309627532959,
      "step": 1213
    },
    {
      "epoch": 0.5846024197917293,
      "grad_norm": 13.847315788269043,
      "learning_rate": 7.787827798168115e-05,
      "loss": 1.5906122922897339,
      "step": 1214
    },
    {
      "epoch": 0.5850839703846386,
      "grad_norm": 56.76832580566406,
      "learning_rate": 7.772618190263009e-05,
      "loss": 1.939192771911621,
      "step": 1215
    },
    {
      "epoch": 0.5855655209775477,
      "grad_norm": 30.038509368896484,
      "learning_rate": 7.757414002059726e-05,
      "loss": 1.917020320892334,
      "step": 1216
    },
    {
      "epoch": 0.5860470715704569,
      "grad_norm": 26.527870178222656,
      "learning_rate": 7.742215270553349e-05,
      "loss": 2.213056802749634,
      "step": 1217
    },
    {
      "epoch": 0.586528622163366,
      "grad_norm": 10.537469863891602,
      "learning_rate": 7.727022032725672e-05,
      "loss": 1.8030946254730225,
      "step": 1218
    },
    {
      "epoch": 0.5870101727562752,
      "grad_norm": 7.510496616363525,
      "learning_rate": 7.711834325545135e-05,
      "loss": 1.413394570350647,
      "step": 1219
    },
    {
      "epoch": 0.5874917233491844,
      "grad_norm": 11.800708770751953,
      "learning_rate": 7.696652185966711e-05,
      "loss": 2.2051899433135986,
      "step": 1220
    },
    {
      "epoch": 0.5879732739420935,
      "grad_norm": 26.870344161987305,
      "learning_rate": 7.681475650931834e-05,
      "loss": 2.4985151290893555,
      "step": 1221
    },
    {
      "epoch": 0.5884548245350028,
      "grad_norm": 41.480552673339844,
      "learning_rate": 7.666304757368297e-05,
      "loss": 2.2709155082702637,
      "step": 1222
    },
    {
      "epoch": 0.5889363751279119,
      "grad_norm": 75.65475463867188,
      "learning_rate": 7.651139542190164e-05,
      "loss": 2.097583293914795,
      "step": 1223
    },
    {
      "epoch": 0.5894179257208211,
      "grad_norm": 24.68280792236328,
      "learning_rate": 7.635980042297687e-05,
      "loss": 1.9695138931274414,
      "step": 1224
    },
    {
      "epoch": 0.5898994763137302,
      "grad_norm": 28.980314254760742,
      "learning_rate": 7.620826294577208e-05,
      "loss": 2.0611696243286133,
      "step": 1225
    },
    {
      "epoch": 0.5903810269066394,
      "grad_norm": 27.662321090698242,
      "learning_rate": 7.605678335901071e-05,
      "loss": 2.449669599533081,
      "step": 1226
    },
    {
      "epoch": 0.5908625774995485,
      "grad_norm": 17.8123836517334,
      "learning_rate": 7.59053620312754e-05,
      "loss": 2.8606488704681396,
      "step": 1227
    },
    {
      "epoch": 0.5913441280924577,
      "grad_norm": 29.2480411529541,
      "learning_rate": 7.575399933100697e-05,
      "loss": 2.2489373683929443,
      "step": 1228
    },
    {
      "epoch": 0.5918256786853668,
      "grad_norm": 23.714282989501953,
      "learning_rate": 7.560269562650368e-05,
      "loss": 2.373307228088379,
      "step": 1229
    },
    {
      "epoch": 0.5923072292782761,
      "grad_norm": 35.201534271240234,
      "learning_rate": 7.54514512859201e-05,
      "loss": 2.663574457168579,
      "step": 1230
    },
    {
      "epoch": 0.5927887798711852,
      "grad_norm": 30.902999877929688,
      "learning_rate": 7.530026667726645e-05,
      "loss": 2.6064364910125732,
      "step": 1231
    },
    {
      "epoch": 0.5932703304640944,
      "grad_norm": 42.93708801269531,
      "learning_rate": 7.51491421684076e-05,
      "loss": 2.551081657409668,
      "step": 1232
    },
    {
      "epoch": 0.5937518810570036,
      "grad_norm": 16.499406814575195,
      "learning_rate": 7.49980781270622e-05,
      "loss": 2.407074213027954,
      "step": 1233
    },
    {
      "epoch": 0.5942334316499127,
      "grad_norm": 16.399744033813477,
      "learning_rate": 7.484707492080172e-05,
      "loss": 2.571624517440796,
      "step": 1234
    },
    {
      "epoch": 0.5947149822428219,
      "grad_norm": 51.23805236816406,
      "learning_rate": 7.469613291704962e-05,
      "loss": 2.678925037384033,
      "step": 1235
    },
    {
      "epoch": 0.595196532835731,
      "grad_norm": 11.47387409210205,
      "learning_rate": 7.45452524830805e-05,
      "loss": 2.832627534866333,
      "step": 1236
    },
    {
      "epoch": 0.5956780834286403,
      "grad_norm": 65.69676208496094,
      "learning_rate": 7.439443398601903e-05,
      "loss": 2.1494476795196533,
      "step": 1237
    },
    {
      "epoch": 0.5961596340215494,
      "grad_norm": 18.338808059692383,
      "learning_rate": 7.424367779283926e-05,
      "loss": 2.7471704483032227,
      "step": 1238
    },
    {
      "epoch": 0.5966411846144586,
      "grad_norm": 9.042954444885254,
      "learning_rate": 7.409298427036364e-05,
      "loss": 1.5273665189743042,
      "step": 1239
    },
    {
      "epoch": 0.5971227352073677,
      "grad_norm": 10.174403190612793,
      "learning_rate": 7.39423537852621e-05,
      "loss": 2.149813413619995,
      "step": 1240
    },
    {
      "epoch": 0.5976042858002769,
      "grad_norm": 20.98665428161621,
      "learning_rate": 7.379178670405123e-05,
      "loss": 2.1023685932159424,
      "step": 1241
    },
    {
      "epoch": 0.598085836393186,
      "grad_norm": 45.264923095703125,
      "learning_rate": 7.364128339309326e-05,
      "loss": 1.913124918937683,
      "step": 1242
    },
    {
      "epoch": 0.5985673869860952,
      "grad_norm": 55.62042999267578,
      "learning_rate": 7.349084421859533e-05,
      "loss": 2.7282886505126953,
      "step": 1243
    },
    {
      "epoch": 0.5990489375790043,
      "grad_norm": 11.065959930419922,
      "learning_rate": 7.334046954660852e-05,
      "loss": 2.0767393112182617,
      "step": 1244
    },
    {
      "epoch": 0.5995304881719136,
      "grad_norm": 11.288713455200195,
      "learning_rate": 7.31901597430269e-05,
      "loss": 2.115659236907959,
      "step": 1245
    },
    {
      "epoch": 0.6000120387648227,
      "grad_norm": 15.306757926940918,
      "learning_rate": 7.303991517358678e-05,
      "loss": 3.0556938648223877,
      "step": 1246
    },
    {
      "epoch": 0.6004935893577319,
      "grad_norm": 29.190322875976562,
      "learning_rate": 7.288973620386568e-05,
      "loss": 3.1266684532165527,
      "step": 1247
    },
    {
      "epoch": 0.6009751399506411,
      "grad_norm": 34.27463150024414,
      "learning_rate": 7.273962319928151e-05,
      "loss": 1.8381602764129639,
      "step": 1248
    },
    {
      "epoch": 0.6014566905435502,
      "grad_norm": 18.342647552490234,
      "learning_rate": 7.258957652509171e-05,
      "loss": 1.8825194835662842,
      "step": 1249
    },
    {
      "epoch": 0.6019382411364594,
      "grad_norm": 37.74098205566406,
      "learning_rate": 7.24395965463923e-05,
      "loss": 1.8026567697525024,
      "step": 1250
    },
    {
      "epoch": 0.6024197917293685,
      "grad_norm": 34.58066177368164,
      "learning_rate": 7.228968362811702e-05,
      "loss": 2.6359450817108154,
      "step": 1251
    },
    {
      "epoch": 0.6029013423222778,
      "grad_norm": 14.020258903503418,
      "learning_rate": 7.21398381350364e-05,
      "loss": 2.2163562774658203,
      "step": 1252
    },
    {
      "epoch": 0.6033828929151869,
      "grad_norm": 30.611221313476562,
      "learning_rate": 7.199006043175698e-05,
      "loss": 2.2025816440582275,
      "step": 1253
    },
    {
      "epoch": 0.6038644435080961,
      "grad_norm": 10.048371315002441,
      "learning_rate": 7.184035088272028e-05,
      "loss": 2.4751670360565186,
      "step": 1254
    },
    {
      "epoch": 0.6043459941010052,
      "grad_norm": 33.655296325683594,
      "learning_rate": 7.169070985220208e-05,
      "loss": 2.0740110874176025,
      "step": 1255
    },
    {
      "epoch": 0.6048275446939144,
      "grad_norm": 16.93131446838379,
      "learning_rate": 7.154113770431132e-05,
      "loss": 2.0388967990875244,
      "step": 1256
    },
    {
      "epoch": 0.6053090952868235,
      "grad_norm": 6.185673236846924,
      "learning_rate": 7.13916348029894e-05,
      "loss": 2.0776736736297607,
      "step": 1257
    },
    {
      "epoch": 0.6057906458797327,
      "grad_norm": 25.684268951416016,
      "learning_rate": 7.124220151200926e-05,
      "loss": 2.186455726623535,
      "step": 1258
    },
    {
      "epoch": 0.6062721964726419,
      "grad_norm": 14.669422149658203,
      "learning_rate": 7.10928381949744e-05,
      "loss": 2.203671455383301,
      "step": 1259
    },
    {
      "epoch": 0.6067537470655511,
      "grad_norm": 21.866186141967773,
      "learning_rate": 7.094354521531807e-05,
      "loss": 1.9504424333572388,
      "step": 1260
    },
    {
      "epoch": 0.6072352976584603,
      "grad_norm": 155.9264678955078,
      "learning_rate": 7.079432293630244e-05,
      "loss": 2.3743724822998047,
      "step": 1261
    },
    {
      "epoch": 0.6077168482513694,
      "grad_norm": 29.378334045410156,
      "learning_rate": 7.064517172101753e-05,
      "loss": 2.993642568588257,
      "step": 1262
    },
    {
      "epoch": 0.6081983988442786,
      "grad_norm": 40.488956451416016,
      "learning_rate": 7.04960919323806e-05,
      "loss": 1.9555840492248535,
      "step": 1263
    },
    {
      "epoch": 0.6086799494371877,
      "grad_norm": 6.96743631362915,
      "learning_rate": 7.034708393313493e-05,
      "loss": 2.7717864513397217,
      "step": 1264
    },
    {
      "epoch": 0.609161500030097,
      "grad_norm": 13.187249183654785,
      "learning_rate": 7.019814808584928e-05,
      "loss": 2.696075916290283,
      "step": 1265
    },
    {
      "epoch": 0.609643050623006,
      "grad_norm": 10.988441467285156,
      "learning_rate": 7.004928475291678e-05,
      "loss": 2.284101963043213,
      "step": 1266
    },
    {
      "epoch": 0.6101246012159153,
      "grad_norm": 7.85512113571167,
      "learning_rate": 6.990049429655412e-05,
      "loss": 2.287336587905884,
      "step": 1267
    },
    {
      "epoch": 0.6106061518088244,
      "grad_norm": 8.732757568359375,
      "learning_rate": 6.97517770788007e-05,
      "loss": 2.100341558456421,
      "step": 1268
    },
    {
      "epoch": 0.6110877024017336,
      "grad_norm": 13.856315612792969,
      "learning_rate": 6.960313346151761e-05,
      "loss": 2.922771453857422,
      "step": 1269
    },
    {
      "epoch": 0.6115692529946427,
      "grad_norm": 9.407246589660645,
      "learning_rate": 6.9454563806387e-05,
      "loss": 2.6743123531341553,
      "step": 1270
    },
    {
      "epoch": 0.6120508035875519,
      "grad_norm": 20.809160232543945,
      "learning_rate": 6.930606847491094e-05,
      "loss": 1.988687515258789,
      "step": 1271
    },
    {
      "epoch": 0.612532354180461,
      "grad_norm": 74.72474670410156,
      "learning_rate": 6.915764782841072e-05,
      "loss": 2.846127510070801,
      "step": 1272
    },
    {
      "epoch": 0.6130139047733703,
      "grad_norm": 27.60916519165039,
      "learning_rate": 6.900930222802588e-05,
      "loss": 2.700998544692993,
      "step": 1273
    },
    {
      "epoch": 0.6134954553662795,
      "grad_norm": 229.94512939453125,
      "learning_rate": 6.886103203471337e-05,
      "loss": 2.4434549808502197,
      "step": 1274
    },
    {
      "epoch": 0.6139770059591886,
      "grad_norm": 8.266853332519531,
      "learning_rate": 6.871283760924665e-05,
      "loss": 2.4458773136138916,
      "step": 1275
    },
    {
      "epoch": 0.6144585565520978,
      "grad_norm": 30.796579360961914,
      "learning_rate": 6.856471931221478e-05,
      "loss": 2.514631986618042,
      "step": 1276
    },
    {
      "epoch": 0.6149401071450069,
      "grad_norm": 8.548853874206543,
      "learning_rate": 6.841667750402162e-05,
      "loss": 1.4263074398040771,
      "step": 1277
    },
    {
      "epoch": 0.6154216577379161,
      "grad_norm": 28.37224578857422,
      "learning_rate": 6.826871254488496e-05,
      "loss": 2.4018430709838867,
      "step": 1278
    },
    {
      "epoch": 0.6159032083308252,
      "grad_norm": 13.749380111694336,
      "learning_rate": 6.812082479483553e-05,
      "loss": 1.9165009260177612,
      "step": 1279
    },
    {
      "epoch": 0.6163847589237345,
      "grad_norm": 16.447437286376953,
      "learning_rate": 6.797301461371625e-05,
      "loss": 2.191324234008789,
      "step": 1280
    },
    {
      "epoch": 0.6168663095166436,
      "grad_norm": 16.03656005859375,
      "learning_rate": 6.782528236118124e-05,
      "loss": 2.1046218872070312,
      "step": 1281
    },
    {
      "epoch": 0.6173478601095528,
      "grad_norm": 35.96092224121094,
      "learning_rate": 6.767762839669503e-05,
      "loss": 2.20194673538208,
      "step": 1282
    },
    {
      "epoch": 0.6178294107024619,
      "grad_norm": 13.640375137329102,
      "learning_rate": 6.753005307953167e-05,
      "loss": 1.7570641040802002,
      "step": 1283
    },
    {
      "epoch": 0.6183109612953711,
      "grad_norm": 102.05022430419922,
      "learning_rate": 6.738255676877381e-05,
      "loss": 2.1494338512420654,
      "step": 1284
    },
    {
      "epoch": 0.6187925118882802,
      "grad_norm": 12.283950805664062,
      "learning_rate": 6.723513982331195e-05,
      "loss": 2.686990737915039,
      "step": 1285
    },
    {
      "epoch": 0.6192740624811894,
      "grad_norm": 24.881696701049805,
      "learning_rate": 6.708780260184333e-05,
      "loss": 2.545398473739624,
      "step": 1286
    },
    {
      "epoch": 0.6197556130740985,
      "grad_norm": 8.386361122131348,
      "learning_rate": 6.694054546287132e-05,
      "loss": 2.2660892009735107,
      "step": 1287
    },
    {
      "epoch": 0.6202371636670078,
      "grad_norm": 32.3531608581543,
      "learning_rate": 6.679336876470441e-05,
      "loss": 2.4050588607788086,
      "step": 1288
    },
    {
      "epoch": 0.620718714259917,
      "grad_norm": 18.48056411743164,
      "learning_rate": 6.664627286545535e-05,
      "loss": 2.7195281982421875,
      "step": 1289
    },
    {
      "epoch": 0.6212002648528261,
      "grad_norm": 12.983318328857422,
      "learning_rate": 6.649925812304025e-05,
      "loss": 1.7612711191177368,
      "step": 1290
    },
    {
      "epoch": 0.6216818154457353,
      "grad_norm": 31.652891159057617,
      "learning_rate": 6.635232489517782e-05,
      "loss": 2.293074131011963,
      "step": 1291
    },
    {
      "epoch": 0.6221633660386444,
      "grad_norm": 10.453925132751465,
      "learning_rate": 6.620547353938836e-05,
      "loss": 2.316542863845825,
      "step": 1292
    },
    {
      "epoch": 0.6226449166315536,
      "grad_norm": 791.4789428710938,
      "learning_rate": 6.605870441299302e-05,
      "loss": 2.5888671875,
      "step": 1293
    },
    {
      "epoch": 0.6231264672244627,
      "grad_norm": 76.34371185302734,
      "learning_rate": 6.591201787311285e-05,
      "loss": 3.108834981918335,
      "step": 1294
    },
    {
      "epoch": 0.623608017817372,
      "grad_norm": 14.513887405395508,
      "learning_rate": 6.57654142766679e-05,
      "loss": 1.803714394569397,
      "step": 1295
    },
    {
      "epoch": 0.6240895684102811,
      "grad_norm": 11.078219413757324,
      "learning_rate": 6.561889398037643e-05,
      "loss": 2.5376200675964355,
      "step": 1296
    },
    {
      "epoch": 0.6245711190031903,
      "grad_norm": 93.35169219970703,
      "learning_rate": 6.547245734075403e-05,
      "loss": 2.593148946762085,
      "step": 1297
    },
    {
      "epoch": 0.6250526695960994,
      "grad_norm": 23.44378089904785,
      "learning_rate": 6.532610471411274e-05,
      "loss": 2.314544439315796,
      "step": 1298
    },
    {
      "epoch": 0.6255342201890086,
      "grad_norm": 12.786527633666992,
      "learning_rate": 6.517983645656014e-05,
      "loss": 2.1688215732574463,
      "step": 1299
    },
    {
      "epoch": 0.6260157707819177,
      "grad_norm": 16.822490692138672,
      "learning_rate": 6.503365292399857e-05,
      "loss": 2.1257145404815674,
      "step": 1300
    },
    {
      "epoch": 0.6264973213748269,
      "grad_norm": 268.8167419433594,
      "learning_rate": 6.488755447212418e-05,
      "loss": 2.089125394821167,
      "step": 1301
    },
    {
      "epoch": 0.6269788719677362,
      "grad_norm": 19.888486862182617,
      "learning_rate": 6.474154145642612e-05,
      "loss": 1.8739361763000488,
      "step": 1302
    },
    {
      "epoch": 0.6274604225606453,
      "grad_norm": 6.685331344604492,
      "learning_rate": 6.459561423218561e-05,
      "loss": 2.1901845932006836,
      "step": 1303
    },
    {
      "epoch": 0.6279419731535545,
      "grad_norm": 32.919307708740234,
      "learning_rate": 6.444977315447521e-05,
      "loss": 2.3540401458740234,
      "step": 1304
    },
    {
      "epoch": 0.6284235237464636,
      "grad_norm": 17.233673095703125,
      "learning_rate": 6.430401857815776e-05,
      "loss": 2.3354194164276123,
      "step": 1305
    },
    {
      "epoch": 0.6289050743393728,
      "grad_norm": 70.68090057373047,
      "learning_rate": 6.415835085788575e-05,
      "loss": 2.683459758758545,
      "step": 1306
    },
    {
      "epoch": 0.6293866249322819,
      "grad_norm": 14.703972816467285,
      "learning_rate": 6.401277034810017e-05,
      "loss": 2.0076792240142822,
      "step": 1307
    },
    {
      "epoch": 0.6298681755251911,
      "grad_norm": 189.67921447753906,
      "learning_rate": 6.386727740302994e-05,
      "loss": 2.484280824661255,
      "step": 1308
    },
    {
      "epoch": 0.6303497261181003,
      "grad_norm": 15.169654846191406,
      "learning_rate": 6.37218723766909e-05,
      "loss": 2.607064723968506,
      "step": 1309
    },
    {
      "epoch": 0.6308312767110095,
      "grad_norm": 40.14570999145508,
      "learning_rate": 6.357655562288488e-05,
      "loss": 2.305037498474121,
      "step": 1310
    },
    {
      "epoch": 0.6313128273039186,
      "grad_norm": 55.35014724731445,
      "learning_rate": 6.343132749519902e-05,
      "loss": 2.4636099338531494,
      "step": 1311
    },
    {
      "epoch": 0.6317943778968278,
      "grad_norm": 32.983917236328125,
      "learning_rate": 6.328618834700474e-05,
      "loss": 2.118178129196167,
      "step": 1312
    },
    {
      "epoch": 0.6322759284897369,
      "grad_norm": 22.722248077392578,
      "learning_rate": 6.314113853145703e-05,
      "loss": 2.011446237564087,
      "step": 1313
    },
    {
      "epoch": 0.6327574790826461,
      "grad_norm": 30.797718048095703,
      "learning_rate": 6.299617840149349e-05,
      "loss": 2.1654739379882812,
      "step": 1314
    },
    {
      "epoch": 0.6332390296755553,
      "grad_norm": 121.77246856689453,
      "learning_rate": 6.285130830983339e-05,
      "loss": 2.40348482131958,
      "step": 1315
    },
    {
      "epoch": 0.6337205802684645,
      "grad_norm": 19.90184783935547,
      "learning_rate": 6.270652860897704e-05,
      "loss": 2.173053026199341,
      "step": 1316
    },
    {
      "epoch": 0.6342021308613737,
      "grad_norm": 18.192501068115234,
      "learning_rate": 6.25618396512048e-05,
      "loss": 1.7719213962554932,
      "step": 1317
    },
    {
      "epoch": 0.6346836814542828,
      "grad_norm": 20.45534896850586,
      "learning_rate": 6.24172417885762e-05,
      "loss": 1.8881733417510986,
      "step": 1318
    },
    {
      "epoch": 0.635165232047192,
      "grad_norm": 295.6603698730469,
      "learning_rate": 6.227273537292911e-05,
      "loss": 2.3603031635284424,
      "step": 1319
    },
    {
      "epoch": 0.6356467826401011,
      "grad_norm": 17.45140266418457,
      "learning_rate": 6.212832075587891e-05,
      "loss": 2.0259287357330322,
      "step": 1320
    },
    {
      "epoch": 0.6361283332330103,
      "grad_norm": 31.32977294921875,
      "learning_rate": 6.19839982888176e-05,
      "loss": 1.6289196014404297,
      "step": 1321
    },
    {
      "epoch": 0.6366098838259194,
      "grad_norm": 28.619815826416016,
      "learning_rate": 6.183976832291296e-05,
      "loss": 2.3794469833374023,
      "step": 1322
    },
    {
      "epoch": 0.6370914344188287,
      "grad_norm": 122.67472076416016,
      "learning_rate": 6.169563120910775e-05,
      "loss": 2.5747995376586914,
      "step": 1323
    },
    {
      "epoch": 0.6375729850117378,
      "grad_norm": 24.478042602539062,
      "learning_rate": 6.155158729811867e-05,
      "loss": 2.7360591888427734,
      "step": 1324
    },
    {
      "epoch": 0.638054535604647,
      "grad_norm": 64.31078338623047,
      "learning_rate": 6.140763694043578e-05,
      "loss": 1.8781907558441162,
      "step": 1325
    },
    {
      "epoch": 0.6385360861975561,
      "grad_norm": 7.416792869567871,
      "learning_rate": 6.126378048632139e-05,
      "loss": 1.8610559701919556,
      "step": 1326
    },
    {
      "epoch": 0.6390176367904653,
      "grad_norm": 18.87049674987793,
      "learning_rate": 6.112001828580944e-05,
      "loss": 2.6838080883026123,
      "step": 1327
    },
    {
      "epoch": 0.6394991873833744,
      "grad_norm": 49.56711196899414,
      "learning_rate": 6.0976350688704455e-05,
      "loss": 1.9311238527297974,
      "step": 1328
    },
    {
      "epoch": 0.6399807379762836,
      "grad_norm": 17.113311767578125,
      "learning_rate": 6.083277804458072e-05,
      "loss": 2.0785200595855713,
      "step": 1329
    },
    {
      "epoch": 0.6404622885691929,
      "grad_norm": 11.392815589904785,
      "learning_rate": 6.068930070278159e-05,
      "loss": 1.9060535430908203,
      "step": 1330
    },
    {
      "epoch": 0.640943839162102,
      "grad_norm": 53.930519104003906,
      "learning_rate": 6.054591901241846e-05,
      "loss": 2.1154263019561768,
      "step": 1331
    },
    {
      "epoch": 0.6414253897550112,
      "grad_norm": 26.979841232299805,
      "learning_rate": 6.040263332237002e-05,
      "loss": 2.3511486053466797,
      "step": 1332
    },
    {
      "epoch": 0.6419069403479203,
      "grad_norm": 8.07104778289795,
      "learning_rate": 6.025944398128137e-05,
      "loss": 2.1998953819274902,
      "step": 1333
    },
    {
      "epoch": 0.6423884909408295,
      "grad_norm": 17.194171905517578,
      "learning_rate": 6.011635133756309e-05,
      "loss": 2.2196006774902344,
      "step": 1334
    },
    {
      "epoch": 0.6428700415337386,
      "grad_norm": 20.1202449798584,
      "learning_rate": 5.99733557393906e-05,
      "loss": 2.1045033931732178,
      "step": 1335
    },
    {
      "epoch": 0.6433515921266478,
      "grad_norm": 25.75075912475586,
      "learning_rate": 5.983045753470308e-05,
      "loss": 2.8343985080718994,
      "step": 1336
    },
    {
      "epoch": 0.6438331427195569,
      "grad_norm": 7.4643940925598145,
      "learning_rate": 5.96876570712028e-05,
      "loss": 2.0355002880096436,
      "step": 1337
    },
    {
      "epoch": 0.6443146933124662,
      "grad_norm": 55.19389343261719,
      "learning_rate": 5.954495469635417e-05,
      "loss": 2.52773380279541,
      "step": 1338
    },
    {
      "epoch": 0.6447962439053753,
      "grad_norm": 24.717439651489258,
      "learning_rate": 5.940235075738296e-05,
      "loss": 2.4677577018737793,
      "step": 1339
    },
    {
      "epoch": 0.6452777944982845,
      "grad_norm": 11.159093856811523,
      "learning_rate": 5.925984560127542e-05,
      "loss": 2.6207892894744873,
      "step": 1340
    },
    {
      "epoch": 0.6457593450911936,
      "grad_norm": 12.797510147094727,
      "learning_rate": 5.911743957477739e-05,
      "loss": 2.583030939102173,
      "step": 1341
    },
    {
      "epoch": 0.6462408956841028,
      "grad_norm": 18.24090003967285,
      "learning_rate": 5.897513302439355e-05,
      "loss": 2.110109806060791,
      "step": 1342
    },
    {
      "epoch": 0.646722446277012,
      "grad_norm": 9.327401161193848,
      "learning_rate": 5.883292629638651e-05,
      "loss": 2.4518837928771973,
      "step": 1343
    },
    {
      "epoch": 0.6472039968699211,
      "grad_norm": 16.610471725463867,
      "learning_rate": 5.869081973677604e-05,
      "loss": 2.103522777557373,
      "step": 1344
    },
    {
      "epoch": 0.6476855474628304,
      "grad_norm": 10.04007625579834,
      "learning_rate": 5.8548813691338134e-05,
      "loss": 2.151047706604004,
      "step": 1345
    },
    {
      "epoch": 0.6481670980557395,
      "grad_norm": 13.547382354736328,
      "learning_rate": 5.84069085056042e-05,
      "loss": 2.5440502166748047,
      "step": 1346
    },
    {
      "epoch": 0.6486486486486487,
      "grad_norm": 14.391494750976562,
      "learning_rate": 5.826510452486027e-05,
      "loss": 2.4164719581604004,
      "step": 1347
    },
    {
      "epoch": 0.6491301992415578,
      "grad_norm": 11.3621244430542,
      "learning_rate": 5.81234020941461e-05,
      "loss": 2.1638290882110596,
      "step": 1348
    },
    {
      "epoch": 0.649611749834467,
      "grad_norm": 23.873714447021484,
      "learning_rate": 5.798180155825437e-05,
      "loss": 1.5566887855529785,
      "step": 1349
    },
    {
      "epoch": 0.6500933004273761,
      "grad_norm": 29.425859451293945,
      "learning_rate": 5.784030326172981e-05,
      "loss": 2.2582240104675293,
      "step": 1350
    },
    {
      "epoch": 0.6505748510202853,
      "grad_norm": 13.132271766662598,
      "learning_rate": 5.7698907548868395e-05,
      "loss": 1.830911636352539,
      "step": 1351
    },
    {
      "epoch": 0.6510564016131944,
      "grad_norm": 9.837907791137695,
      "learning_rate": 5.755761476371653e-05,
      "loss": 1.6203837394714355,
      "step": 1352
    },
    {
      "epoch": 0.6515379522061037,
      "grad_norm": 18.093618392944336,
      "learning_rate": 5.741642525007003e-05,
      "loss": 2.4737489223480225,
      "step": 1353
    },
    {
      "epoch": 0.6520195027990128,
      "grad_norm": 12.30912971496582,
      "learning_rate": 5.727533935147359e-05,
      "loss": 2.6758015155792236,
      "step": 1354
    },
    {
      "epoch": 0.652501053391922,
      "grad_norm": 37.572383880615234,
      "learning_rate": 5.713435741121975e-05,
      "loss": 2.250786542892456,
      "step": 1355
    },
    {
      "epoch": 0.6529826039848312,
      "grad_norm": 32.09661102294922,
      "learning_rate": 5.699347977234799e-05,
      "loss": 1.7232040166854858,
      "step": 1356
    },
    {
      "epoch": 0.6534641545777403,
      "grad_norm": 26.769393920898438,
      "learning_rate": 5.685270677764412e-05,
      "loss": 1.8753536939620972,
      "step": 1357
    },
    {
      "epoch": 0.6539457051706495,
      "grad_norm": 9.951456069946289,
      "learning_rate": 5.671203876963931e-05,
      "loss": 2.0826632976531982,
      "step": 1358
    },
    {
      "epoch": 0.6544272557635586,
      "grad_norm": 38.21604919433594,
      "learning_rate": 5.657147609060924e-05,
      "loss": 1.9251338243484497,
      "step": 1359
    },
    {
      "epoch": 0.6549088063564679,
      "grad_norm": 22.809223175048828,
      "learning_rate": 5.643101908257333e-05,
      "loss": 1.8415642976760864,
      "step": 1360
    },
    {
      "epoch": 0.655390356949377,
      "grad_norm": 7.28424072265625,
      "learning_rate": 5.629066808729385e-05,
      "loss": 2.0468292236328125,
      "step": 1361
    },
    {
      "epoch": 0.6558719075422862,
      "grad_norm": 9.687171936035156,
      "learning_rate": 5.6150423446275144e-05,
      "loss": 2.135054588317871,
      "step": 1362
    },
    {
      "epoch": 0.6563534581351953,
      "grad_norm": 19.837295532226562,
      "learning_rate": 5.601028550076277e-05,
      "loss": 2.8038930892944336,
      "step": 1363
    },
    {
      "epoch": 0.6568350087281045,
      "grad_norm": 33.94176483154297,
      "learning_rate": 5.587025459174271e-05,
      "loss": 2.0950329303741455,
      "step": 1364
    },
    {
      "epoch": 0.6573165593210136,
      "grad_norm": 6.44937801361084,
      "learning_rate": 5.573033105994038e-05,
      "loss": 1.8111432790756226,
      "step": 1365
    },
    {
      "epoch": 0.6577981099139228,
      "grad_norm": 15.400513648986816,
      "learning_rate": 5.559051524582002e-05,
      "loss": 2.0143699645996094,
      "step": 1366
    },
    {
      "epoch": 0.658279660506832,
      "grad_norm": 13.314600944519043,
      "learning_rate": 5.5450807489583777e-05,
      "loss": 1.9432076215744019,
      "step": 1367
    },
    {
      "epoch": 0.6587612110997412,
      "grad_norm": 9.786284446716309,
      "learning_rate": 5.531120813117085e-05,
      "loss": 1.7496002912521362,
      "step": 1368
    },
    {
      "epoch": 0.6592427616926503,
      "grad_norm": 18.975370407104492,
      "learning_rate": 5.517171751025667e-05,
      "loss": 2.4910192489624023,
      "step": 1369
    },
    {
      "epoch": 0.6597243122855595,
      "grad_norm": 27.542009353637695,
      "learning_rate": 5.5032335966252103e-05,
      "loss": 1.6665278673171997,
      "step": 1370
    },
    {
      "epoch": 0.6602058628784687,
      "grad_norm": 76.08589935302734,
      "learning_rate": 5.489306383830258e-05,
      "loss": 2.6495883464813232,
      "step": 1371
    },
    {
      "epoch": 0.6606874134713778,
      "grad_norm": 5.521340847015381,
      "learning_rate": 5.475390146528738e-05,
      "loss": 2.0278210639953613,
      "step": 1372
    },
    {
      "epoch": 0.661168964064287,
      "grad_norm": 17.686767578125,
      "learning_rate": 5.461484918581858e-05,
      "loss": 2.3557350635528564,
      "step": 1373
    },
    {
      "epoch": 0.6616505146571962,
      "grad_norm": 17.052732467651367,
      "learning_rate": 5.4475907338240494e-05,
      "loss": 1.839455485343933,
      "step": 1374
    },
    {
      "epoch": 0.6621320652501054,
      "grad_norm": 31.841890335083008,
      "learning_rate": 5.43370762606287e-05,
      "loss": 3.115117073059082,
      "step": 1375
    },
    {
      "epoch": 0.6626136158430145,
      "grad_norm": 7.034003257751465,
      "learning_rate": 5.4198356290789276e-05,
      "loss": 2.1519393920898438,
      "step": 1376
    },
    {
      "epoch": 0.6630951664359237,
      "grad_norm": 11.795368194580078,
      "learning_rate": 5.405974776625785e-05,
      "loss": 1.3862327337265015,
      "step": 1377
    },
    {
      "epoch": 0.6635767170288328,
      "grad_norm": 14.378402709960938,
      "learning_rate": 5.392125102429899e-05,
      "loss": 2.0001323223114014,
      "step": 1378
    },
    {
      "epoch": 0.664058267621742,
      "grad_norm": 26.478343963623047,
      "learning_rate": 5.378286640190522e-05,
      "loss": 1.7846403121948242,
      "step": 1379
    },
    {
      "epoch": 0.6645398182146511,
      "grad_norm": 28.261091232299805,
      "learning_rate": 5.364459423579629e-05,
      "loss": 2.427395820617676,
      "step": 1380
    },
    {
      "epoch": 0.6650213688075604,
      "grad_norm": 17.126924514770508,
      "learning_rate": 5.350643486241825e-05,
      "loss": 2.276582956314087,
      "step": 1381
    },
    {
      "epoch": 0.6655029194004695,
      "grad_norm": 11.664289474487305,
      "learning_rate": 5.33683886179428e-05,
      "loss": 2.1981446743011475,
      "step": 1382
    },
    {
      "epoch": 0.6659844699933787,
      "grad_norm": 25.05721664428711,
      "learning_rate": 5.3230455838266266e-05,
      "loss": 2.4785866737365723,
      "step": 1383
    },
    {
      "epoch": 0.6664660205862879,
      "grad_norm": 5.526149272918701,
      "learning_rate": 5.309263685900898e-05,
      "loss": 1.9356056451797485,
      "step": 1384
    },
    {
      "epoch": 0.666947571179197,
      "grad_norm": 18.4484920501709,
      "learning_rate": 5.295493201551433e-05,
      "loss": 1.7978616952896118,
      "step": 1385
    },
    {
      "epoch": 0.6674291217721062,
      "grad_norm": 12.470459938049316,
      "learning_rate": 5.281734164284802e-05,
      "loss": 2.7486159801483154,
      "step": 1386
    },
    {
      "epoch": 0.6679106723650153,
      "grad_norm": 17.01667022705078,
      "learning_rate": 5.26798660757971e-05,
      "loss": 3.2195160388946533,
      "step": 1387
    },
    {
      "epoch": 0.6683922229579246,
      "grad_norm": 19.266998291015625,
      "learning_rate": 5.2542505648869434e-05,
      "loss": 1.9141626358032227,
      "step": 1388
    },
    {
      "epoch": 0.6688737735508337,
      "grad_norm": 18.82830810546875,
      "learning_rate": 5.240526069629265e-05,
      "loss": 2.018866777420044,
      "step": 1389
    },
    {
      "epoch": 0.6693553241437429,
      "grad_norm": 41.583770751953125,
      "learning_rate": 5.22681315520134e-05,
      "loss": 2.309873104095459,
      "step": 1390
    },
    {
      "epoch": 0.669836874736652,
      "grad_norm": 14.264725685119629,
      "learning_rate": 5.213111854969661e-05,
      "loss": 2.3256587982177734,
      "step": 1391
    },
    {
      "epoch": 0.6703184253295612,
      "grad_norm": 12.354305267333984,
      "learning_rate": 5.199422202272448e-05,
      "loss": 2.118558645248413,
      "step": 1392
    },
    {
      "epoch": 0.6707999759224703,
      "grad_norm": 10.3298978805542,
      "learning_rate": 5.185744230419589e-05,
      "loss": 2.828246831893921,
      "step": 1393
    },
    {
      "epoch": 0.6712815265153795,
      "grad_norm": 26.800037384033203,
      "learning_rate": 5.172077972692553e-05,
      "loss": 2.0189967155456543,
      "step": 1394
    },
    {
      "epoch": 0.6717630771082886,
      "grad_norm": 13.435601234436035,
      "learning_rate": 5.1584234623442974e-05,
      "loss": 1.6227072477340698,
      "step": 1395
    },
    {
      "epoch": 0.6722446277011979,
      "grad_norm": 9.993821144104004,
      "learning_rate": 5.1447807325992025e-05,
      "loss": 1.391953468322754,
      "step": 1396
    },
    {
      "epoch": 0.6727261782941071,
      "grad_norm": 9.751219749450684,
      "learning_rate": 5.13114981665298e-05,
      "loss": 1.764944314956665,
      "step": 1397
    },
    {
      "epoch": 0.6732077288870162,
      "grad_norm": 12.79586124420166,
      "learning_rate": 5.117530747672603e-05,
      "loss": 1.642270803451538,
      "step": 1398
    },
    {
      "epoch": 0.6736892794799254,
      "grad_norm": 5.9880571365356445,
      "learning_rate": 5.103923558796203e-05,
      "loss": 2.1377248764038086,
      "step": 1399
    },
    {
      "epoch": 0.6741708300728345,
      "grad_norm": 11.674127578735352,
      "learning_rate": 5.090328283133019e-05,
      "loss": 2.549734354019165,
      "step": 1400
    },
    {
      "epoch": 0.6746523806657437,
      "grad_norm": 11.739805221557617,
      "learning_rate": 5.0767449537632986e-05,
      "loss": 2.4866278171539307,
      "step": 1401
    },
    {
      "epoch": 0.6751339312586528,
      "grad_norm": 55.038604736328125,
      "learning_rate": 5.06317360373822e-05,
      "loss": 2.3475303649902344,
      "step": 1402
    },
    {
      "epoch": 0.6756154818515621,
      "grad_norm": 6.5804524421691895,
      "learning_rate": 5.049614266079813e-05,
      "loss": 2.1680283546447754,
      "step": 1403
    },
    {
      "epoch": 0.6760970324444712,
      "grad_norm": 18.370853424072266,
      "learning_rate": 5.036066973780882e-05,
      "loss": 2.5669546127319336,
      "step": 1404
    },
    {
      "epoch": 0.6765785830373804,
      "grad_norm": 32.63365936279297,
      "learning_rate": 5.022531759804918e-05,
      "loss": 2.0771801471710205,
      "step": 1405
    },
    {
      "epoch": 0.6770601336302895,
      "grad_norm": 24.169818878173828,
      "learning_rate": 5.009008657086025e-05,
      "loss": 1.8822633028030396,
      "step": 1406
    },
    {
      "epoch": 0.6775416842231987,
      "grad_norm": 17.178329467773438,
      "learning_rate": 4.9954976985288395e-05,
      "loss": 2.7332992553710938,
      "step": 1407
    },
    {
      "epoch": 0.6780232348161078,
      "grad_norm": 23.985057830810547,
      "learning_rate": 4.981998917008448e-05,
      "loss": 1.8720582723617554,
      "step": 1408
    },
    {
      "epoch": 0.678504785409017,
      "grad_norm": 14.91263484954834,
      "learning_rate": 4.9685123453703e-05,
      "loss": 2.1097958087921143,
      "step": 1409
    },
    {
      "epoch": 0.6789863360019261,
      "grad_norm": 13.565987586975098,
      "learning_rate": 4.955038016430149e-05,
      "loss": 1.7091691493988037,
      "step": 1410
    },
    {
      "epoch": 0.6794678865948354,
      "grad_norm": 16.942079544067383,
      "learning_rate": 4.9415759629739455e-05,
      "loss": 1.9388729333877563,
      "step": 1411
    },
    {
      "epoch": 0.6799494371877446,
      "grad_norm": 42.697635650634766,
      "learning_rate": 4.928126217757782e-05,
      "loss": 2.432218074798584,
      "step": 1412
    },
    {
      "epoch": 0.6804309877806537,
      "grad_norm": 31.785444259643555,
      "learning_rate": 4.914688813507797e-05,
      "loss": 2.5593619346618652,
      "step": 1413
    },
    {
      "epoch": 0.6809125383735629,
      "grad_norm": 81.679443359375,
      "learning_rate": 4.901263782920105e-05,
      "loss": 2.438509464263916,
      "step": 1414
    },
    {
      "epoch": 0.681394088966472,
      "grad_norm": 16.786428451538086,
      "learning_rate": 4.887851158660706e-05,
      "loss": 2.7449638843536377,
      "step": 1415
    },
    {
      "epoch": 0.6818756395593812,
      "grad_norm": 16.288908004760742,
      "learning_rate": 4.8744509733654184e-05,
      "loss": 1.975643515586853,
      "step": 1416
    },
    {
      "epoch": 0.6823571901522903,
      "grad_norm": 7.350008964538574,
      "learning_rate": 4.861063259639793e-05,
      "loss": 2.2706692218780518,
      "step": 1417
    },
    {
      "epoch": 0.6828387407451996,
      "grad_norm": 192.86476135253906,
      "learning_rate": 4.847688050059033e-05,
      "loss": 2.157881021499634,
      "step": 1418
    },
    {
      "epoch": 0.6833202913381087,
      "grad_norm": 24.421960830688477,
      "learning_rate": 4.8343253771679155e-05,
      "loss": 2.7118067741394043,
      "step": 1419
    },
    {
      "epoch": 0.6838018419310179,
      "grad_norm": 9.706432342529297,
      "learning_rate": 4.82097527348072e-05,
      "loss": 2.7223708629608154,
      "step": 1420
    },
    {
      "epoch": 0.684283392523927,
      "grad_norm": 10.458221435546875,
      "learning_rate": 4.8076377714811284e-05,
      "loss": 1.9215338230133057,
      "step": 1421
    },
    {
      "epoch": 0.6847649431168362,
      "grad_norm": 12.958203315734863,
      "learning_rate": 4.7943129036221735e-05,
      "loss": 2.1612842082977295,
      "step": 1422
    },
    {
      "epoch": 0.6852464937097453,
      "grad_norm": 16.542993545532227,
      "learning_rate": 4.781000702326142e-05,
      "loss": 2.25095272064209,
      "step": 1423
    },
    {
      "epoch": 0.6857280443026545,
      "grad_norm": 83.37271881103516,
      "learning_rate": 4.767701199984497e-05,
      "loss": 2.0848870277404785,
      "step": 1424
    },
    {
      "epoch": 0.6862095948955638,
      "grad_norm": 12.66709041595459,
      "learning_rate": 4.7544144289578066e-05,
      "loss": 2.4191136360168457,
      "step": 1425
    },
    {
      "epoch": 0.6866911454884729,
      "grad_norm": 14.972891807556152,
      "learning_rate": 4.7411404215756594e-05,
      "loss": 1.9324803352355957,
      "step": 1426
    },
    {
      "epoch": 0.6871726960813821,
      "grad_norm": 14.781145095825195,
      "learning_rate": 4.7278792101365866e-05,
      "loss": 1.9340885877609253,
      "step": 1427
    },
    {
      "epoch": 0.6876542466742912,
      "grad_norm": 10.392034530639648,
      "learning_rate": 4.714630826907985e-05,
      "loss": 1.738418698310852,
      "step": 1428
    },
    {
      "epoch": 0.6881357972672004,
      "grad_norm": 5.056974411010742,
      "learning_rate": 4.701395304126038e-05,
      "loss": 2.078994035720825,
      "step": 1429
    },
    {
      "epoch": 0.6886173478601095,
      "grad_norm": 12.553384780883789,
      "learning_rate": 4.6881726739956375e-05,
      "loss": 2.2202930450439453,
      "step": 1430
    },
    {
      "epoch": 0.6890988984530187,
      "grad_norm": 25.005224227905273,
      "learning_rate": 4.6749629686902984e-05,
      "loss": 2.7338788509368896,
      "step": 1431
    },
    {
      "epoch": 0.6895804490459279,
      "grad_norm": 13.652777671813965,
      "learning_rate": 4.661766220352097e-05,
      "loss": 1.9549331665039062,
      "step": 1432
    },
    {
      "epoch": 0.6900619996388371,
      "grad_norm": 19.781715393066406,
      "learning_rate": 4.64858246109157e-05,
      "loss": 2.5229272842407227,
      "step": 1433
    },
    {
      "epoch": 0.6905435502317462,
      "grad_norm": 18.912235260009766,
      "learning_rate": 4.63541172298766e-05,
      "loss": 2.162497043609619,
      "step": 1434
    },
    {
      "epoch": 0.6910251008246554,
      "grad_norm": 59.32202911376953,
      "learning_rate": 4.622254038087622e-05,
      "loss": 2.2407443523406982,
      "step": 1435
    },
    {
      "epoch": 0.6915066514175645,
      "grad_norm": 7.081389904022217,
      "learning_rate": 4.60910943840695e-05,
      "loss": 2.3668904304504395,
      "step": 1436
    },
    {
      "epoch": 0.6919882020104737,
      "grad_norm": 20.587221145629883,
      "learning_rate": 4.5959779559292985e-05,
      "loss": 2.433422088623047,
      "step": 1437
    },
    {
      "epoch": 0.6924697526033828,
      "grad_norm": 30.798105239868164,
      "learning_rate": 4.582859622606406e-05,
      "loss": 2.1285624504089355,
      "step": 1438
    },
    {
      "epoch": 0.6929513031962921,
      "grad_norm": 34.20453643798828,
      "learning_rate": 4.569754470358014e-05,
      "loss": 2.6314432621002197,
      "step": 1439
    },
    {
      "epoch": 0.6934328537892013,
      "grad_norm": 22.450437545776367,
      "learning_rate": 4.556662531071796e-05,
      "loss": 2.419893264770508,
      "step": 1440
    },
    {
      "epoch": 0.6939144043821104,
      "grad_norm": 8.757513999938965,
      "learning_rate": 4.54358383660327e-05,
      "loss": 1.8962067365646362,
      "step": 1441
    },
    {
      "epoch": 0.6943959549750196,
      "grad_norm": 9.200644493103027,
      "learning_rate": 4.530518418775733e-05,
      "loss": 1.9078731536865234,
      "step": 1442
    },
    {
      "epoch": 0.6948775055679287,
      "grad_norm": 15.392255783081055,
      "learning_rate": 4.5174663093801674e-05,
      "loss": 1.7717167139053345,
      "step": 1443
    },
    {
      "epoch": 0.6953590561608379,
      "grad_norm": 23.680418014526367,
      "learning_rate": 4.504427540175181e-05,
      "loss": 1.5447163581848145,
      "step": 1444
    },
    {
      "epoch": 0.695840606753747,
      "grad_norm": 7.589408874511719,
      "learning_rate": 4.491402142886922e-05,
      "loss": 1.5998762845993042,
      "step": 1445
    },
    {
      "epoch": 0.6963221573466563,
      "grad_norm": 21.94112205505371,
      "learning_rate": 4.4783901492089984e-05,
      "loss": 2.382538080215454,
      "step": 1446
    },
    {
      "epoch": 0.6968037079395654,
      "grad_norm": 9.969077110290527,
      "learning_rate": 4.465391590802407e-05,
      "loss": 2.272739887237549,
      "step": 1447
    },
    {
      "epoch": 0.6972852585324746,
      "grad_norm": 181.54678344726562,
      "learning_rate": 4.4524064992954516e-05,
      "loss": 1.993480920791626,
      "step": 1448
    },
    {
      "epoch": 0.6977668091253837,
      "grad_norm": 45.154666900634766,
      "learning_rate": 4.4394349062836736e-05,
      "loss": 1.907798171043396,
      "step": 1449
    },
    {
      "epoch": 0.6982483597182929,
      "grad_norm": 37.768856048583984,
      "learning_rate": 4.4264768433297565e-05,
      "loss": 1.4964039325714111,
      "step": 1450
    },
    {
      "epoch": 0.698729910311202,
      "grad_norm": 70.52316284179688,
      "learning_rate": 4.4135323419634766e-05,
      "loss": 2.4861152172088623,
      "step": 1451
    },
    {
      "epoch": 0.6992114609041112,
      "grad_norm": 11.62633228302002,
      "learning_rate": 4.4006014336816035e-05,
      "loss": 2.3712968826293945,
      "step": 1452
    },
    {
      "epoch": 0.6996930114970205,
      "grad_norm": 18.961963653564453,
      "learning_rate": 4.387684149947837e-05,
      "loss": 1.5992252826690674,
      "step": 1453
    },
    {
      "epoch": 0.7001745620899296,
      "grad_norm": 7.445619583129883,
      "learning_rate": 4.374780522192726e-05,
      "loss": 2.0693721771240234,
      "step": 1454
    },
    {
      "epoch": 0.7006561126828388,
      "grad_norm": 15.765836715698242,
      "learning_rate": 4.3618905818135805e-05,
      "loss": 2.5738046169281006,
      "step": 1455
    },
    {
      "epoch": 0.7011376632757479,
      "grad_norm": 6.312963962554932,
      "learning_rate": 4.349014360174417e-05,
      "loss": 2.2423593997955322,
      "step": 1456
    },
    {
      "epoch": 0.7016192138686571,
      "grad_norm": 15.382637023925781,
      "learning_rate": 4.336151888605871e-05,
      "loss": 1.8969299793243408,
      "step": 1457
    },
    {
      "epoch": 0.7021007644615662,
      "grad_norm": 7.944668292999268,
      "learning_rate": 4.323303198405117e-05,
      "loss": 2.3287582397460938,
      "step": 1458
    },
    {
      "epoch": 0.7025823150544754,
      "grad_norm": 27.80340576171875,
      "learning_rate": 4.310468320835796e-05,
      "loss": 2.093217134475708,
      "step": 1459
    },
    {
      "epoch": 0.7030638656473845,
      "grad_norm": 13.201991081237793,
      "learning_rate": 4.297647287127946e-05,
      "loss": 2.134157180786133,
      "step": 1460
    },
    {
      "epoch": 0.7035454162402938,
      "grad_norm": 142.84713745117188,
      "learning_rate": 4.284840128477913e-05,
      "loss": 2.502084732055664,
      "step": 1461
    },
    {
      "epoch": 0.7040269668332029,
      "grad_norm": 6.651097774505615,
      "learning_rate": 4.2720468760482854e-05,
      "loss": 2.607974052429199,
      "step": 1462
    },
    {
      "epoch": 0.7045085174261121,
      "grad_norm": 5.559552192687988,
      "learning_rate": 4.2592675609678135e-05,
      "loss": 2.239809513092041,
      "step": 1463
    },
    {
      "epoch": 0.7049900680190212,
      "grad_norm": 48.10198974609375,
      "learning_rate": 4.24650221433134e-05,
      "loss": 2.4234225749969482,
      "step": 1464
    },
    {
      "epoch": 0.7054716186119304,
      "grad_norm": 9.000733375549316,
      "learning_rate": 4.2337508671997086e-05,
      "loss": 2.1721978187561035,
      "step": 1465
    },
    {
      "epoch": 0.7059531692048396,
      "grad_norm": 10.12733268737793,
      "learning_rate": 4.221013550599707e-05,
      "loss": 2.6770663261413574,
      "step": 1466
    },
    {
      "epoch": 0.7064347197977487,
      "grad_norm": 11.385762214660645,
      "learning_rate": 4.208290295523984e-05,
      "loss": 2.9267027378082275,
      "step": 1467
    },
    {
      "epoch": 0.706916270390658,
      "grad_norm": 23.580018997192383,
      "learning_rate": 4.1955811329309746e-05,
      "loss": 2.7104976177215576,
      "step": 1468
    },
    {
      "epoch": 0.7073978209835671,
      "grad_norm": 38.92906188964844,
      "learning_rate": 4.182886093744813e-05,
      "loss": 2.6511924266815186,
      "step": 1469
    },
    {
      "epoch": 0.7078793715764763,
      "grad_norm": 17.464834213256836,
      "learning_rate": 4.170205208855281e-05,
      "loss": 1.8844338655471802,
      "step": 1470
    },
    {
      "epoch": 0.7083609221693854,
      "grad_norm": 70.53472900390625,
      "learning_rate": 4.157538509117714e-05,
      "loss": 2.0565690994262695,
      "step": 1471
    },
    {
      "epoch": 0.7088424727622946,
      "grad_norm": 7.582005500793457,
      "learning_rate": 4.144886025352934e-05,
      "loss": 2.259084939956665,
      "step": 1472
    },
    {
      "epoch": 0.7093240233552037,
      "grad_norm": 11.209492683410645,
      "learning_rate": 4.13224778834717e-05,
      "loss": 1.9308503866195679,
      "step": 1473
    },
    {
      "epoch": 0.7098055739481129,
      "grad_norm": 64.82707214355469,
      "learning_rate": 4.1196238288519874e-05,
      "loss": 2.4262218475341797,
      "step": 1474
    },
    {
      "epoch": 0.710287124541022,
      "grad_norm": 91.13782501220703,
      "learning_rate": 4.107014177584211e-05,
      "loss": 2.5771119594573975,
      "step": 1475
    },
    {
      "epoch": 0.7107686751339313,
      "grad_norm": 19.077207565307617,
      "learning_rate": 4.094418865225853e-05,
      "loss": 1.4043588638305664,
      "step": 1476
    },
    {
      "epoch": 0.7112502257268404,
      "grad_norm": 18.34503173828125,
      "learning_rate": 4.081837922424027e-05,
      "loss": 2.446030616760254,
      "step": 1477
    },
    {
      "epoch": 0.7117317763197496,
      "grad_norm": 8.645740509033203,
      "learning_rate": 4.069271379790891e-05,
      "loss": 1.7507277727127075,
      "step": 1478
    },
    {
      "epoch": 0.7122133269126587,
      "grad_norm": 18.188371658325195,
      "learning_rate": 4.0567192679035636e-05,
      "loss": 2.2922494411468506,
      "step": 1479
    },
    {
      "epoch": 0.7126948775055679,
      "grad_norm": 19.108051300048828,
      "learning_rate": 4.044181617304048e-05,
      "loss": 2.2913360595703125,
      "step": 1480
    },
    {
      "epoch": 0.7131764280984771,
      "grad_norm": 35.44940948486328,
      "learning_rate": 4.03165845849916e-05,
      "loss": 2.3721864223480225,
      "step": 1481
    },
    {
      "epoch": 0.7136579786913863,
      "grad_norm": 9.53534984588623,
      "learning_rate": 4.019149821960455e-05,
      "loss": 2.1024296283721924,
      "step": 1482
    },
    {
      "epoch": 0.7141395292842955,
      "grad_norm": 20.990598678588867,
      "learning_rate": 4.006655738124152e-05,
      "loss": 2.3655967712402344,
      "step": 1483
    },
    {
      "epoch": 0.7146210798772046,
      "grad_norm": 9.005951881408691,
      "learning_rate": 3.9941762373910586e-05,
      "loss": 1.6189895868301392,
      "step": 1484
    },
    {
      "epoch": 0.7151026304701138,
      "grad_norm": 8.42679214477539,
      "learning_rate": 3.9817113501265016e-05,
      "loss": 2.318294048309326,
      "step": 1485
    },
    {
      "epoch": 0.7155841810630229,
      "grad_norm": 19.99054718017578,
      "learning_rate": 3.9692611066602516e-05,
      "loss": 2.2441742420196533,
      "step": 1486
    },
    {
      "epoch": 0.7160657316559321,
      "grad_norm": 7.419765949249268,
      "learning_rate": 3.956825537286436e-05,
      "loss": 1.6528128385543823,
      "step": 1487
    },
    {
      "epoch": 0.7165472822488412,
      "grad_norm": 58.591392517089844,
      "learning_rate": 3.944404672263494e-05,
      "loss": 2.3731653690338135,
      "step": 1488
    },
    {
      "epoch": 0.7170288328417505,
      "grad_norm": 26.949689865112305,
      "learning_rate": 3.931998541814069e-05,
      "loss": 2.4814236164093018,
      "step": 1489
    },
    {
      "epoch": 0.7175103834346596,
      "grad_norm": 25.022497177124023,
      "learning_rate": 3.919607176124966e-05,
      "loss": 2.6564781665802,
      "step": 1490
    },
    {
      "epoch": 0.7179919340275688,
      "grad_norm": 4.740179538726807,
      "learning_rate": 3.9072306053470566e-05,
      "loss": 2.3195433616638184,
      "step": 1491
    },
    {
      "epoch": 0.7184734846204779,
      "grad_norm": 13.819572448730469,
      "learning_rate": 3.8948688595952164e-05,
      "loss": 2.498307943344116,
      "step": 1492
    },
    {
      "epoch": 0.7189550352133871,
      "grad_norm": 9.59683609008789,
      "learning_rate": 3.882521968948246e-05,
      "loss": 2.2620394229888916,
      "step": 1493
    },
    {
      "epoch": 0.7194365858062963,
      "grad_norm": 7.224384784698486,
      "learning_rate": 3.8701899634488014e-05,
      "loss": 1.7881441116333008,
      "step": 1494
    },
    {
      "epoch": 0.7199181363992054,
      "grad_norm": 12.735471725463867,
      "learning_rate": 3.857872873103322e-05,
      "loss": 1.7333070039749146,
      "step": 1495
    },
    {
      "epoch": 0.7203996869921147,
      "grad_norm": 17.88201332092285,
      "learning_rate": 3.8455707278819507e-05,
      "loss": 2.0490880012512207,
      "step": 1496
    },
    {
      "epoch": 0.7208812375850238,
      "grad_norm": 16.397171020507812,
      "learning_rate": 3.833283557718471e-05,
      "loss": 2.425509214401245,
      "step": 1497
    },
    {
      "epoch": 0.721362788177933,
      "grad_norm": 13.588607788085938,
      "learning_rate": 3.821011392510228e-05,
      "loss": 1.9037139415740967,
      "step": 1498
    },
    {
      "epoch": 0.7218443387708421,
      "grad_norm": 16.43208885192871,
      "learning_rate": 3.808754262118046e-05,
      "loss": 2.138278007507324,
      "step": 1499
    },
    {
      "epoch": 0.7223258893637513,
      "grad_norm": 192.81752014160156,
      "learning_rate": 3.796512196366182e-05,
      "loss": 2.2428977489471436,
      "step": 1500
    },
    {
      "epoch": 0.7228074399566604,
      "grad_norm": 39.06431579589844,
      "learning_rate": 3.784285225042229e-05,
      "loss": 2.4307000637054443,
      "step": 1501
    },
    {
      "epoch": 0.7232889905495696,
      "grad_norm": 31.56878089904785,
      "learning_rate": 3.772073377897052e-05,
      "loss": 2.327298641204834,
      "step": 1502
    },
    {
      "epoch": 0.7237705411424787,
      "grad_norm": 14.632074356079102,
      "learning_rate": 3.7598766846447184e-05,
      "loss": 2.6031336784362793,
      "step": 1503
    },
    {
      "epoch": 0.724252091735388,
      "grad_norm": 5.640593528747559,
      "learning_rate": 3.747695174962423e-05,
      "loss": 2.467409610748291,
      "step": 1504
    },
    {
      "epoch": 0.7247336423282971,
      "grad_norm": 17.237293243408203,
      "learning_rate": 3.7355288784904116e-05,
      "loss": 2.2946698665618896,
      "step": 1505
    },
    {
      "epoch": 0.7252151929212063,
      "grad_norm": 8.112110137939453,
      "learning_rate": 3.7233778248319176e-05,
      "loss": 2.638425588607788,
      "step": 1506
    },
    {
      "epoch": 0.7256967435141155,
      "grad_norm": 12.742579460144043,
      "learning_rate": 3.7112420435530845e-05,
      "loss": 1.7296956777572632,
      "step": 1507
    },
    {
      "epoch": 0.7261782941070246,
      "grad_norm": 111.6516342163086,
      "learning_rate": 3.69912156418289e-05,
      "loss": 2.7637572288513184,
      "step": 1508
    },
    {
      "epoch": 0.7266598446999338,
      "grad_norm": 14.5833740234375,
      "learning_rate": 3.687016416213084e-05,
      "loss": 2.1540496349334717,
      "step": 1509
    },
    {
      "epoch": 0.7271413952928429,
      "grad_norm": 8.658010482788086,
      "learning_rate": 3.674926629098113e-05,
      "loss": 2.8271074295043945,
      "step": 1510
    },
    {
      "epoch": 0.7276229458857522,
      "grad_norm": 10.634912490844727,
      "learning_rate": 3.6628522322550394e-05,
      "loss": 2.1914515495300293,
      "step": 1511
    },
    {
      "epoch": 0.7281044964786613,
      "grad_norm": 5.653548240661621,
      "learning_rate": 3.6507932550634846e-05,
      "loss": 2.125309467315674,
      "step": 1512
    },
    {
      "epoch": 0.7285860470715705,
      "grad_norm": 13.417054176330566,
      "learning_rate": 3.638749726865552e-05,
      "loss": 1.987107276916504,
      "step": 1513
    },
    {
      "epoch": 0.7290675976644796,
      "grad_norm": 5.847234725952148,
      "learning_rate": 3.6267216769657485e-05,
      "loss": 1.9518831968307495,
      "step": 1514
    },
    {
      "epoch": 0.7295491482573888,
      "grad_norm": 12.182299613952637,
      "learning_rate": 3.6147091346309224e-05,
      "loss": 2.52665114402771,
      "step": 1515
    },
    {
      "epoch": 0.7300306988502979,
      "grad_norm": 34.5836296081543,
      "learning_rate": 3.602712129090189e-05,
      "loss": 1.9339662790298462,
      "step": 1516
    },
    {
      "epoch": 0.7305122494432071,
      "grad_norm": 9.502908706665039,
      "learning_rate": 3.590730689534857e-05,
      "loss": 1.6167874336242676,
      "step": 1517
    },
    {
      "epoch": 0.7309938000361162,
      "grad_norm": 7.830199718475342,
      "learning_rate": 3.578764845118362e-05,
      "loss": 2.40655255317688,
      "step": 1518
    },
    {
      "epoch": 0.7314753506290255,
      "grad_norm": 28.030784606933594,
      "learning_rate": 3.566814624956194e-05,
      "loss": 2.359697103500366,
      "step": 1519
    },
    {
      "epoch": 0.7319569012219346,
      "grad_norm": 22.836265563964844,
      "learning_rate": 3.554880058125819e-05,
      "loss": 2.230571746826172,
      "step": 1520
    },
    {
      "epoch": 0.7324384518148438,
      "grad_norm": 13.582202911376953,
      "learning_rate": 3.5429611736666235e-05,
      "loss": 1.7670902013778687,
      "step": 1521
    },
    {
      "epoch": 0.732920002407753,
      "grad_norm": 17.70282554626465,
      "learning_rate": 3.53105800057983e-05,
      "loss": 2.0803306102752686,
      "step": 1522
    },
    {
      "epoch": 0.7334015530006621,
      "grad_norm": 14.103548049926758,
      "learning_rate": 3.519170567828435e-05,
      "loss": 2.5934903621673584,
      "step": 1523
    },
    {
      "epoch": 0.7338831035935713,
      "grad_norm": 24.45741844177246,
      "learning_rate": 3.507298904337134e-05,
      "loss": 2.4898297786712646,
      "step": 1524
    },
    {
      "epoch": 0.7343646541864804,
      "grad_norm": 11.994695663452148,
      "learning_rate": 3.495443038992253e-05,
      "loss": 1.771467685699463,
      "step": 1525
    },
    {
      "epoch": 0.7348462047793897,
      "grad_norm": 61.02875900268555,
      "learning_rate": 3.4836030006416775e-05,
      "loss": 1.9682687520980835,
      "step": 1526
    },
    {
      "epoch": 0.7353277553722988,
      "grad_norm": 8.060063362121582,
      "learning_rate": 3.471778818094785e-05,
      "loss": 1.610675573348999,
      "step": 1527
    },
    {
      "epoch": 0.735809305965208,
      "grad_norm": 16.053476333618164,
      "learning_rate": 3.459970520122364e-05,
      "loss": 1.7292819023132324,
      "step": 1528
    },
    {
      "epoch": 0.7362908565581171,
      "grad_norm": 11.723716735839844,
      "learning_rate": 3.44817813545656e-05,
      "loss": 1.9194117784500122,
      "step": 1529
    },
    {
      "epoch": 0.7367724071510263,
      "grad_norm": 9.054965019226074,
      "learning_rate": 3.4364016927907974e-05,
      "loss": 2.726928472518921,
      "step": 1530
    },
    {
      "epoch": 0.7372539577439354,
      "grad_norm": 14.120265007019043,
      "learning_rate": 3.424641220779711e-05,
      "loss": 1.9882646799087524,
      "step": 1531
    },
    {
      "epoch": 0.7377355083368446,
      "grad_norm": 9.440196990966797,
      "learning_rate": 3.412896748039067e-05,
      "loss": 2.0584452152252197,
      "step": 1532
    },
    {
      "epoch": 0.7382170589297538,
      "grad_norm": 18.545263290405273,
      "learning_rate": 3.401168303145713e-05,
      "loss": 2.1726882457733154,
      "step": 1533
    },
    {
      "epoch": 0.738698609522663,
      "grad_norm": 7.7046122550964355,
      "learning_rate": 3.3894559146374924e-05,
      "loss": 1.4596697092056274,
      "step": 1534
    },
    {
      "epoch": 0.7391801601155722,
      "grad_norm": 6.135489463806152,
      "learning_rate": 3.37775961101318e-05,
      "loss": 1.5094640254974365,
      "step": 1535
    },
    {
      "epoch": 0.7396617107084813,
      "grad_norm": 59.63398361206055,
      "learning_rate": 3.366079420732413e-05,
      "loss": 2.2892379760742188,
      "step": 1536
    },
    {
      "epoch": 0.7401432613013905,
      "grad_norm": 16.89608383178711,
      "learning_rate": 3.3544153722156216e-05,
      "loss": 2.517340660095215,
      "step": 1537
    },
    {
      "epoch": 0.7406248118942996,
      "grad_norm": 10.142233848571777,
      "learning_rate": 3.3427674938439594e-05,
      "loss": 1.8671008348464966,
      "step": 1538
    },
    {
      "epoch": 0.7411063624872088,
      "grad_norm": 15.466540336608887,
      "learning_rate": 3.3311358139592317e-05,
      "loss": 1.9030804634094238,
      "step": 1539
    },
    {
      "epoch": 0.741587913080118,
      "grad_norm": 7.137239933013916,
      "learning_rate": 3.319520360863837e-05,
      "loss": 1.9294942617416382,
      "step": 1540
    },
    {
      "epoch": 0.7420694636730272,
      "grad_norm": 15.936016082763672,
      "learning_rate": 3.3079211628206854e-05,
      "loss": 2.6687564849853516,
      "step": 1541
    },
    {
      "epoch": 0.7425510142659363,
      "grad_norm": 37.54365921020508,
      "learning_rate": 3.296338248053129e-05,
      "loss": 1.9344918727874756,
      "step": 1542
    },
    {
      "epoch": 0.7430325648588455,
      "grad_norm": 21.039995193481445,
      "learning_rate": 3.2847716447449096e-05,
      "loss": 2.1293156147003174,
      "step": 1543
    },
    {
      "epoch": 0.7435141154517546,
      "grad_norm": 53.69261169433594,
      "learning_rate": 3.2732213810400745e-05,
      "loss": 2.030240535736084,
      "step": 1544
    },
    {
      "epoch": 0.7439956660446638,
      "grad_norm": 10.95885181427002,
      "learning_rate": 3.261687485042915e-05,
      "loss": 2.564906120300293,
      "step": 1545
    },
    {
      "epoch": 0.7444772166375729,
      "grad_norm": 13.629457473754883,
      "learning_rate": 3.250169984817897e-05,
      "loss": 1.9856513738632202,
      "step": 1546
    },
    {
      "epoch": 0.7449587672304822,
      "grad_norm": 38.24711990356445,
      "learning_rate": 3.238668908389586e-05,
      "loss": 2.672027826309204,
      "step": 1547
    },
    {
      "epoch": 0.7454403178233914,
      "grad_norm": 28.7163143157959,
      "learning_rate": 3.227184283742591e-05,
      "loss": 1.9615859985351562,
      "step": 1548
    },
    {
      "epoch": 0.7459218684163005,
      "grad_norm": 19.77630615234375,
      "learning_rate": 3.215716138821488e-05,
      "loss": 2.5826303958892822,
      "step": 1549
    },
    {
      "epoch": 0.7464034190092097,
      "grad_norm": 9.196318626403809,
      "learning_rate": 3.204264501530756e-05,
      "loss": 1.9490114450454712,
      "step": 1550
    },
    {
      "epoch": 0.7468849696021188,
      "grad_norm": 53.615142822265625,
      "learning_rate": 3.192829399734706e-05,
      "loss": 2.319967269897461,
      "step": 1551
    },
    {
      "epoch": 0.747366520195028,
      "grad_norm": 42.86952209472656,
      "learning_rate": 3.181410861257413e-05,
      "loss": 1.8679351806640625,
      "step": 1552
    },
    {
      "epoch": 0.7478480707879371,
      "grad_norm": 143.00579833984375,
      "learning_rate": 3.170008913882656e-05,
      "loss": 2.2735583782196045,
      "step": 1553
    },
    {
      "epoch": 0.7483296213808464,
      "grad_norm": 27.326858520507812,
      "learning_rate": 3.1586235853538325e-05,
      "loss": 1.9953429698944092,
      "step": 1554
    },
    {
      "epoch": 0.7488111719737555,
      "grad_norm": 18.8415584564209,
      "learning_rate": 3.1472549033739126e-05,
      "loss": 2.228888988494873,
      "step": 1555
    },
    {
      "epoch": 0.7492927225666647,
      "grad_norm": 15.21926498413086,
      "learning_rate": 3.1359028956053615e-05,
      "loss": 1.8453739881515503,
      "step": 1556
    },
    {
      "epoch": 0.7497742731595738,
      "grad_norm": 19.302515029907227,
      "learning_rate": 3.1245675896700685e-05,
      "loss": 1.9102338552474976,
      "step": 1557
    },
    {
      "epoch": 0.750255823752483,
      "grad_norm": 86.55175018310547,
      "learning_rate": 3.113249013149284e-05,
      "loss": 2.146324396133423,
      "step": 1558
    },
    {
      "epoch": 0.7507373743453921,
      "grad_norm": 14.49450969696045,
      "learning_rate": 3.101947193583557e-05,
      "loss": 2.2975220680236816,
      "step": 1559
    },
    {
      "epoch": 0.7512189249383013,
      "grad_norm": 2819.07275390625,
      "learning_rate": 3.0906621584726546e-05,
      "loss": 2.5023317337036133,
      "step": 1560
    },
    {
      "epoch": 0.7517004755312104,
      "grad_norm": 18.12244415283203,
      "learning_rate": 3.079393935275513e-05,
      "loss": 1.9322973489761353,
      "step": 1561
    },
    {
      "epoch": 0.7521820261241197,
      "grad_norm": 42.833866119384766,
      "learning_rate": 3.068142551410155e-05,
      "loss": 2.0392727851867676,
      "step": 1562
    },
    {
      "epoch": 0.7526635767170289,
      "grad_norm": 16.5412654876709,
      "learning_rate": 3.0569080342536347e-05,
      "loss": 2.2490084171295166,
      "step": 1563
    },
    {
      "epoch": 0.753145127309938,
      "grad_norm": 20.066001892089844,
      "learning_rate": 3.0456904111419572e-05,
      "loss": 2.36694598197937,
      "step": 1564
    },
    {
      "epoch": 0.7536266779028472,
      "grad_norm": 10.5184965133667,
      "learning_rate": 3.034489709370033e-05,
      "loss": 2.5775938034057617,
      "step": 1565
    },
    {
      "epoch": 0.7541082284957563,
      "grad_norm": 47.03486633300781,
      "learning_rate": 3.0233059561915855e-05,
      "loss": 2.722463607788086,
      "step": 1566
    },
    {
      "epoch": 0.7545897790886655,
      "grad_norm": 13.940997123718262,
      "learning_rate": 3.01213917881911e-05,
      "loss": 1.9597033262252808,
      "step": 1567
    },
    {
      "epoch": 0.7550713296815746,
      "grad_norm": 21.505788803100586,
      "learning_rate": 3.0009894044237907e-05,
      "loss": 2.484783172607422,
      "step": 1568
    },
    {
      "epoch": 0.7555528802744839,
      "grad_norm": 18.22726821899414,
      "learning_rate": 2.9898566601354418e-05,
      "loss": 2.3711235523223877,
      "step": 1569
    },
    {
      "epoch": 0.756034430867393,
      "grad_norm": 201.85919189453125,
      "learning_rate": 2.9787409730424374e-05,
      "loss": 2.0997979640960693,
      "step": 1570
    },
    {
      "epoch": 0.7565159814603022,
      "grad_norm": 9.195306777954102,
      "learning_rate": 2.96764237019165e-05,
      "loss": 2.1944193840026855,
      "step": 1571
    },
    {
      "epoch": 0.7569975320532113,
      "grad_norm": 11.51015567779541,
      "learning_rate": 2.9565608785883815e-05,
      "loss": 2.2182741165161133,
      "step": 1572
    },
    {
      "epoch": 0.7574790826461205,
      "grad_norm": 25.4970645904541,
      "learning_rate": 2.9454965251962973e-05,
      "loss": 2.395439624786377,
      "step": 1573
    },
    {
      "epoch": 0.7579606332390296,
      "grad_norm": 13.97570514678955,
      "learning_rate": 2.9344493369373637e-05,
      "loss": 2.921539068222046,
      "step": 1574
    },
    {
      "epoch": 0.7584421838319388,
      "grad_norm": 10.376752853393555,
      "learning_rate": 2.9234193406917833e-05,
      "loss": 2.8483335971832275,
      "step": 1575
    },
    {
      "epoch": 0.7589237344248481,
      "grad_norm": 15.122527122497559,
      "learning_rate": 2.912406563297916e-05,
      "loss": 2.2518889904022217,
      "step": 1576
    },
    {
      "epoch": 0.7594052850177572,
      "grad_norm": 9.84664249420166,
      "learning_rate": 2.901411031552236e-05,
      "loss": 1.6758619546890259,
      "step": 1577
    },
    {
      "epoch": 0.7598868356106664,
      "grad_norm": 32.190399169921875,
      "learning_rate": 2.8904327722092495e-05,
      "loss": 1.9989359378814697,
      "step": 1578
    },
    {
      "epoch": 0.7603683862035755,
      "grad_norm": 11.352503776550293,
      "learning_rate": 2.879471811981437e-05,
      "loss": 2.291527032852173,
      "step": 1579
    },
    {
      "epoch": 0.7608499367964847,
      "grad_norm": 60.98366165161133,
      "learning_rate": 2.868528177539187e-05,
      "loss": 2.334860324859619,
      "step": 1580
    },
    {
      "epoch": 0.7613314873893938,
      "grad_norm": 44.983375549316406,
      "learning_rate": 2.8576018955107285e-05,
      "loss": 2.6476898193359375,
      "step": 1581
    },
    {
      "epoch": 0.761813037982303,
      "grad_norm": 9.542588233947754,
      "learning_rate": 2.8466929924820705e-05,
      "loss": 2.5645203590393066,
      "step": 1582
    },
    {
      "epoch": 0.7622945885752122,
      "grad_norm": 27.344079971313477,
      "learning_rate": 2.8358014949969334e-05,
      "loss": 2.3792803287506104,
      "step": 1583
    },
    {
      "epoch": 0.7627761391681214,
      "grad_norm": 8.19046401977539,
      "learning_rate": 2.8249274295566864e-05,
      "loss": 1.821667194366455,
      "step": 1584
    },
    {
      "epoch": 0.7632576897610305,
      "grad_norm": 19.27357292175293,
      "learning_rate": 2.8140708226202884e-05,
      "loss": 2.52309513092041,
      "step": 1585
    },
    {
      "epoch": 0.7637392403539397,
      "grad_norm": 85.54947662353516,
      "learning_rate": 2.803231700604204e-05,
      "loss": 1.9930731058120728,
      "step": 1586
    },
    {
      "epoch": 0.7642207909468488,
      "grad_norm": 65.08124542236328,
      "learning_rate": 2.7924100898823702e-05,
      "loss": 1.8118879795074463,
      "step": 1587
    },
    {
      "epoch": 0.764702341539758,
      "grad_norm": 29.591135025024414,
      "learning_rate": 2.7816060167861002e-05,
      "loss": 1.7089391946792603,
      "step": 1588
    },
    {
      "epoch": 0.7651838921326672,
      "grad_norm": 39.546897888183594,
      "learning_rate": 2.7708195076040445e-05,
      "loss": 1.9873093366622925,
      "step": 1589
    },
    {
      "epoch": 0.7656654427255764,
      "grad_norm": 13.003232955932617,
      "learning_rate": 2.760050588582114e-05,
      "loss": 2.2568342685699463,
      "step": 1590
    },
    {
      "epoch": 0.7661469933184856,
      "grad_norm": 28.90044403076172,
      "learning_rate": 2.749299285923417e-05,
      "loss": 1.997179627418518,
      "step": 1591
    },
    {
      "epoch": 0.7666285439113947,
      "grad_norm": 130.5333251953125,
      "learning_rate": 2.7385656257881997e-05,
      "loss": 2.8112165927886963,
      "step": 1592
    },
    {
      "epoch": 0.7671100945043039,
      "grad_norm": 57.600738525390625,
      "learning_rate": 2.7278496342937788e-05,
      "loss": 1.9667147397994995,
      "step": 1593
    },
    {
      "epoch": 0.767591645097213,
      "grad_norm": 15.130563735961914,
      "learning_rate": 2.717151337514482e-05,
      "loss": 2.05840802192688,
      "step": 1594
    },
    {
      "epoch": 0.7680731956901222,
      "grad_norm": 25.671083450317383,
      "learning_rate": 2.7064707614815776e-05,
      "loss": 1.5969617366790771,
      "step": 1595
    },
    {
      "epoch": 0.7685547462830313,
      "grad_norm": 241.4725341796875,
      "learning_rate": 2.6958079321832185e-05,
      "loss": 2.567415952682495,
      "step": 1596
    },
    {
      "epoch": 0.7690362968759406,
      "grad_norm": 53.493648529052734,
      "learning_rate": 2.6851628755643776e-05,
      "loss": 2.4957938194274902,
      "step": 1597
    },
    {
      "epoch": 0.7695178474688497,
      "grad_norm": 200.00228881835938,
      "learning_rate": 2.6745356175267765e-05,
      "loss": 1.7085001468658447,
      "step": 1598
    },
    {
      "epoch": 0.7699993980617589,
      "grad_norm": 21.864439010620117,
      "learning_rate": 2.6639261839288343e-05,
      "loss": 2.8812034130096436,
      "step": 1599
    },
    {
      "epoch": 0.770480948654668,
      "grad_norm": 15.793027877807617,
      "learning_rate": 2.6533346005855987e-05,
      "loss": 2.273818016052246,
      "step": 1600
    },
    {
      "epoch": 0.7709624992475772,
      "grad_norm": 19.191354751586914,
      "learning_rate": 2.6427608932686843e-05,
      "loss": 2.2759861946105957,
      "step": 1601
    },
    {
      "epoch": 0.7714440498404863,
      "grad_norm": 36.91248321533203,
      "learning_rate": 2.6322050877062064e-05,
      "loss": 2.3563413619995117,
      "step": 1602
    },
    {
      "epoch": 0.7719256004333955,
      "grad_norm": 52.511009216308594,
      "learning_rate": 2.6216672095827266e-05,
      "loss": 1.623268723487854,
      "step": 1603
    },
    {
      "epoch": 0.7724071510263048,
      "grad_norm": 82.75418090820312,
      "learning_rate": 2.6111472845391827e-05,
      "loss": 3.2713024616241455,
      "step": 1604
    },
    {
      "epoch": 0.7728887016192139,
      "grad_norm": 404.479736328125,
      "learning_rate": 2.6006453381728236e-05,
      "loss": 2.4687321186065674,
      "step": 1605
    },
    {
      "epoch": 0.7733702522121231,
      "grad_norm": 46.64566421508789,
      "learning_rate": 2.5901613960371585e-05,
      "loss": 2.209409236907959,
      "step": 1606
    },
    {
      "epoch": 0.7738518028050322,
      "grad_norm": 21.279287338256836,
      "learning_rate": 2.5796954836418884e-05,
      "loss": 2.8811182975769043,
      "step": 1607
    },
    {
      "epoch": 0.7743333533979414,
      "grad_norm": 55.70894241333008,
      "learning_rate": 2.569247626452842e-05,
      "loss": 2.2511744499206543,
      "step": 1608
    },
    {
      "epoch": 0.7748149039908505,
      "grad_norm": 46.6888542175293,
      "learning_rate": 2.558817849891918e-05,
      "loss": 2.360551595687866,
      "step": 1609
    },
    {
      "epoch": 0.7752964545837597,
      "grad_norm": 16.736732482910156,
      "learning_rate": 2.548406179337015e-05,
      "loss": 1.6041576862335205,
      "step": 1610
    },
    {
      "epoch": 0.7757780051766688,
      "grad_norm": 150.83419799804688,
      "learning_rate": 2.5380126401219807e-05,
      "loss": 2.424070358276367,
      "step": 1611
    },
    {
      "epoch": 0.7762595557695781,
      "grad_norm": 502.006103515625,
      "learning_rate": 2.527637257536547e-05,
      "loss": 2.64465594291687,
      "step": 1612
    },
    {
      "epoch": 0.7767411063624872,
      "grad_norm": 235.59349060058594,
      "learning_rate": 2.517280056826262e-05,
      "loss": 2.3847734928131104,
      "step": 1613
    },
    {
      "epoch": 0.7772226569553964,
      "grad_norm": 10.92882251739502,
      "learning_rate": 2.5069410631924385e-05,
      "loss": 2.2138547897338867,
      "step": 1614
    },
    {
      "epoch": 0.7777042075483055,
      "grad_norm": 7.828449249267578,
      "learning_rate": 2.4966203017920818e-05,
      "loss": 1.6806570291519165,
      "step": 1615
    },
    {
      "epoch": 0.7781857581412147,
      "grad_norm": 23.864967346191406,
      "learning_rate": 2.4863177977378392e-05,
      "loss": 2.4067115783691406,
      "step": 1616
    },
    {
      "epoch": 0.7786673087341239,
      "grad_norm": 13.39703369140625,
      "learning_rate": 2.4760335760979312e-05,
      "loss": 1.6126511096954346,
      "step": 1617
    },
    {
      "epoch": 0.779148859327033,
      "grad_norm": 17.97138214111328,
      "learning_rate": 2.4657676618960944e-05,
      "loss": 2.194777250289917,
      "step": 1618
    },
    {
      "epoch": 0.7796304099199423,
      "grad_norm": 11.644425392150879,
      "learning_rate": 2.455520080111522e-05,
      "loss": 2.972615957260132,
      "step": 1619
    },
    {
      "epoch": 0.7801119605128514,
      "grad_norm": 16.89405632019043,
      "learning_rate": 2.4452908556787912e-05,
      "loss": 2.1951513290405273,
      "step": 1620
    },
    {
      "epoch": 0.7805935111057606,
      "grad_norm": 29.776460647583008,
      "learning_rate": 2.4350800134878203e-05,
      "loss": 2.687551736831665,
      "step": 1621
    },
    {
      "epoch": 0.7810750616986697,
      "grad_norm": 138.51393127441406,
      "learning_rate": 2.4248875783837987e-05,
      "loss": 2.44097638130188,
      "step": 1622
    },
    {
      "epoch": 0.7815566122915789,
      "grad_norm": 142.41143798828125,
      "learning_rate": 2.414713575167129e-05,
      "loss": 1.8788847923278809,
      "step": 1623
    },
    {
      "epoch": 0.782038162884488,
      "grad_norm": 37.700042724609375,
      "learning_rate": 2.4045580285933557e-05,
      "loss": 2.609987735748291,
      "step": 1624
    },
    {
      "epoch": 0.7825197134773972,
      "grad_norm": 13.05506420135498,
      "learning_rate": 2.3944209633731242e-05,
      "loss": 2.865351438522339,
      "step": 1625
    },
    {
      "epoch": 0.7830012640703063,
      "grad_norm": 21.45610809326172,
      "learning_rate": 2.3843024041721053e-05,
      "loss": 1.9029873609542847,
      "step": 1626
    },
    {
      "epoch": 0.7834828146632156,
      "grad_norm": 217.3386688232422,
      "learning_rate": 2.3742023756109456e-05,
      "loss": 2.4037930965423584,
      "step": 1627
    },
    {
      "epoch": 0.7839643652561247,
      "grad_norm": 24.854291915893555,
      "learning_rate": 2.3641209022651976e-05,
      "loss": 2.6236095428466797,
      "step": 1628
    },
    {
      "epoch": 0.7844459158490339,
      "grad_norm": 27.568981170654297,
      "learning_rate": 2.3540580086652675e-05,
      "loss": 1.749655842781067,
      "step": 1629
    },
    {
      "epoch": 0.7849274664419431,
      "grad_norm": 34.141563415527344,
      "learning_rate": 2.344013719296353e-05,
      "loss": 2.1378445625305176,
      "step": 1630
    },
    {
      "epoch": 0.7854090170348522,
      "grad_norm": 82.98526763916016,
      "learning_rate": 2.3339880585983842e-05,
      "loss": 2.159066677093506,
      "step": 1631
    },
    {
      "epoch": 0.7858905676277614,
      "grad_norm": 37.28023147583008,
      "learning_rate": 2.3239810509659597e-05,
      "loss": 2.4547805786132812,
      "step": 1632
    },
    {
      "epoch": 0.7863721182206705,
      "grad_norm": 269.4442443847656,
      "learning_rate": 2.313992720748295e-05,
      "loss": 2.5271646976470947,
      "step": 1633
    },
    {
      "epoch": 0.7868536688135798,
      "grad_norm": 29.091293334960938,
      "learning_rate": 2.304023092249159e-05,
      "loss": 2.649186611175537,
      "step": 1634
    },
    {
      "epoch": 0.7873352194064889,
      "grad_norm": 62.720638275146484,
      "learning_rate": 2.2940721897268136e-05,
      "loss": 1.9474279880523682,
      "step": 1635
    },
    {
      "epoch": 0.7878167699993981,
      "grad_norm": 20.864410400390625,
      "learning_rate": 2.2841400373939592e-05,
      "loss": 1.7905054092407227,
      "step": 1636
    },
    {
      "epoch": 0.7882983205923072,
      "grad_norm": 28.337385177612305,
      "learning_rate": 2.274226659417671e-05,
      "loss": 2.581101655960083,
      "step": 1637
    },
    {
      "epoch": 0.7887798711852164,
      "grad_norm": 18.090791702270508,
      "learning_rate": 2.2643320799193402e-05,
      "loss": 2.2226545810699463,
      "step": 1638
    },
    {
      "epoch": 0.7892614217781255,
      "grad_norm": 19.358997344970703,
      "learning_rate": 2.2544563229746218e-05,
      "loss": 2.4501492977142334,
      "step": 1639
    },
    {
      "epoch": 0.7897429723710347,
      "grad_norm": 13.759553909301758,
      "learning_rate": 2.2445994126133708e-05,
      "loss": 2.7139623165130615,
      "step": 1640
    },
    {
      "epoch": 0.7902245229639439,
      "grad_norm": 22.47949981689453,
      "learning_rate": 2.234761372819577e-05,
      "loss": 1.9592139720916748,
      "step": 1641
    },
    {
      "epoch": 0.7907060735568531,
      "grad_norm": 211.6905975341797,
      "learning_rate": 2.2249422275313214e-05,
      "loss": 2.1870486736297607,
      "step": 1642
    },
    {
      "epoch": 0.7911876241497622,
      "grad_norm": 56.047969818115234,
      "learning_rate": 2.215142000640714e-05,
      "loss": 1.832180142402649,
      "step": 1643
    },
    {
      "epoch": 0.7916691747426714,
      "grad_norm": 82.63705444335938,
      "learning_rate": 2.2053607159938195e-05,
      "loss": 3.014643669128418,
      "step": 1644
    },
    {
      "epoch": 0.7921507253355806,
      "grad_norm": 14.407215118408203,
      "learning_rate": 2.1955983973906236e-05,
      "loss": 2.1513161659240723,
      "step": 1645
    },
    {
      "epoch": 0.7926322759284897,
      "grad_norm": 27.240333557128906,
      "learning_rate": 2.1858550685849578e-05,
      "loss": 2.250056505203247,
      "step": 1646
    },
    {
      "epoch": 0.793113826521399,
      "grad_norm": 18.774436950683594,
      "learning_rate": 2.17613075328445e-05,
      "loss": 2.164506435394287,
      "step": 1647
    },
    {
      "epoch": 0.793595377114308,
      "grad_norm": 18.297117233276367,
      "learning_rate": 2.1664254751504642e-05,
      "loss": 2.3322231769561768,
      "step": 1648
    },
    {
      "epoch": 0.7940769277072173,
      "grad_norm": 17.829038619995117,
      "learning_rate": 2.1567392577980393e-05,
      "loss": 1.302966594696045,
      "step": 1649
    },
    {
      "epoch": 0.7945584783001264,
      "grad_norm": 69.4352798461914,
      "learning_rate": 2.1470721247958404e-05,
      "loss": 2.282017469406128,
      "step": 1650
    },
    {
      "epoch": 0.7950400288930356,
      "grad_norm": 21.57246208190918,
      "learning_rate": 2.137424099666091e-05,
      "loss": 1.7980762720108032,
      "step": 1651
    },
    {
      "epoch": 0.7955215794859447,
      "grad_norm": 888.8842163085938,
      "learning_rate": 2.1277952058845284e-05,
      "loss": 1.6456063985824585,
      "step": 1652
    },
    {
      "epoch": 0.7960031300788539,
      "grad_norm": 44.89912796020508,
      "learning_rate": 2.118185466880327e-05,
      "loss": 1.6127712726593018,
      "step": 1653
    },
    {
      "epoch": 0.796484680671763,
      "grad_norm": 9.141228675842285,
      "learning_rate": 2.1085949060360654e-05,
      "loss": 2.2674872875213623,
      "step": 1654
    },
    {
      "epoch": 0.7969662312646723,
      "grad_norm": 16.947711944580078,
      "learning_rate": 2.0990235466876517e-05,
      "loss": 1.7761516571044922,
      "step": 1655
    },
    {
      "epoch": 0.7974477818575814,
      "grad_norm": 44.309688568115234,
      "learning_rate": 2.089471412124274e-05,
      "loss": 2.190779685974121,
      "step": 1656
    },
    {
      "epoch": 0.7979293324504906,
      "grad_norm": 26.228700637817383,
      "learning_rate": 2.079938525588342e-05,
      "loss": 1.87485933303833,
      "step": 1657
    },
    {
      "epoch": 0.7984108830433998,
      "grad_norm": 296.59112548828125,
      "learning_rate": 2.0704249102754324e-05,
      "loss": 2.072669744491577,
      "step": 1658
    },
    {
      "epoch": 0.7988924336363089,
      "grad_norm": 105.4520263671875,
      "learning_rate": 2.0609305893342278e-05,
      "loss": 2.7704403400421143,
      "step": 1659
    },
    {
      "epoch": 0.7993739842292181,
      "grad_norm": 9.372925758361816,
      "learning_rate": 2.0514555858664663e-05,
      "loss": 2.3646323680877686,
      "step": 1660
    },
    {
      "epoch": 0.7998555348221272,
      "grad_norm": 36.40127182006836,
      "learning_rate": 2.0419999229268805e-05,
      "loss": 2.3758928775787354,
      "step": 1661
    },
    {
      "epoch": 0.8003370854150365,
      "grad_norm": 19.521509170532227,
      "learning_rate": 2.032563623523147e-05,
      "loss": 2.3414809703826904,
      "step": 1662
    },
    {
      "epoch": 0.8008186360079456,
      "grad_norm": 8.2715425491333,
      "learning_rate": 2.0231467106158186e-05,
      "loss": 1.7054967880249023,
      "step": 1663
    },
    {
      "epoch": 0.8013001866008548,
      "grad_norm": 6.587419033050537,
      "learning_rate": 2.0137492071182863e-05,
      "loss": 2.0715582370758057,
      "step": 1664
    },
    {
      "epoch": 0.8017817371937639,
      "grad_norm": 28.19876480102539,
      "learning_rate": 2.0043711358967043e-05,
      "loss": 2.7612059116363525,
      "step": 1665
    },
    {
      "epoch": 0.8022632877866731,
      "grad_norm": 30.944412231445312,
      "learning_rate": 1.9950125197699508e-05,
      "loss": 1.9050592184066772,
      "step": 1666
    },
    {
      "epoch": 0.8027448383795822,
      "grad_norm": 793.4000244140625,
      "learning_rate": 1.985673381509565e-05,
      "loss": 2.558663845062256,
      "step": 1667
    },
    {
      "epoch": 0.8032263889724914,
      "grad_norm": 56.31897735595703,
      "learning_rate": 1.9763537438396894e-05,
      "loss": 1.8769971132278442,
      "step": 1668
    },
    {
      "epoch": 0.8037079395654005,
      "grad_norm": 15.887137413024902,
      "learning_rate": 1.96705362943702e-05,
      "loss": 1.6313334703445435,
      "step": 1669
    },
    {
      "epoch": 0.8041894901583098,
      "grad_norm": 39.088565826416016,
      "learning_rate": 1.9577730609307454e-05,
      "loss": 2.132922887802124,
      "step": 1670
    },
    {
      "epoch": 0.804671040751219,
      "grad_norm": 128.1447296142578,
      "learning_rate": 1.9485120609024975e-05,
      "loss": 1.8112378120422363,
      "step": 1671
    },
    {
      "epoch": 0.8051525913441281,
      "grad_norm": 27.235652923583984,
      "learning_rate": 1.9392706518862935e-05,
      "loss": 1.7875444889068604,
      "step": 1672
    },
    {
      "epoch": 0.8056341419370373,
      "grad_norm": 15.68264389038086,
      "learning_rate": 1.9300488563684804e-05,
      "loss": 1.574759840965271,
      "step": 1673
    },
    {
      "epoch": 0.8061156925299464,
      "grad_norm": 11.892011642456055,
      "learning_rate": 1.920846696787684e-05,
      "loss": 1.4733136892318726,
      "step": 1674
    },
    {
      "epoch": 0.8065972431228556,
      "grad_norm": 12.105127334594727,
      "learning_rate": 1.9116641955347446e-05,
      "loss": 1.7335654497146606,
      "step": 1675
    },
    {
      "epoch": 0.8070787937157647,
      "grad_norm": 10.337924003601074,
      "learning_rate": 1.9025013749526767e-05,
      "loss": 1.8187847137451172,
      "step": 1676
    },
    {
      "epoch": 0.807560344308674,
      "grad_norm": 161.31893920898438,
      "learning_rate": 1.8933582573366036e-05,
      "loss": 2.5574941635131836,
      "step": 1677
    },
    {
      "epoch": 0.8080418949015831,
      "grad_norm": 16.34262466430664,
      "learning_rate": 1.8842348649337116e-05,
      "loss": 2.616194486618042,
      "step": 1678
    },
    {
      "epoch": 0.8085234454944923,
      "grad_norm": 15.288034439086914,
      "learning_rate": 1.875131219943187e-05,
      "loss": 2.161318778991699,
      "step": 1679
    },
    {
      "epoch": 0.8090049960874014,
      "grad_norm": 652.957275390625,
      "learning_rate": 1.8660473445161663e-05,
      "loss": 1.9636789560317993,
      "step": 1680
    },
    {
      "epoch": 0.8094865466803106,
      "grad_norm": 15.1727294921875,
      "learning_rate": 1.856983260755686e-05,
      "loss": 2.323918581008911,
      "step": 1681
    },
    {
      "epoch": 0.8099680972732197,
      "grad_norm": 15.314567565917969,
      "learning_rate": 1.8479389907166223e-05,
      "loss": 2.6391103267669678,
      "step": 1682
    },
    {
      "epoch": 0.8104496478661289,
      "grad_norm": 107.36561584472656,
      "learning_rate": 1.8389145564056387e-05,
      "loss": 1.9152196645736694,
      "step": 1683
    },
    {
      "epoch": 0.810931198459038,
      "grad_norm": 75.9604263305664,
      "learning_rate": 1.829909979781137e-05,
      "loss": 1.8736522197723389,
      "step": 1684
    },
    {
      "epoch": 0.8114127490519473,
      "grad_norm": 15.61462688446045,
      "learning_rate": 1.820925282753201e-05,
      "loss": 1.545247197151184,
      "step": 1685
    },
    {
      "epoch": 0.8118942996448565,
      "grad_norm": 254.4397735595703,
      "learning_rate": 1.8119604871835437e-05,
      "loss": 2.042907476425171,
      "step": 1686
    },
    {
      "epoch": 0.8123758502377656,
      "grad_norm": 49.77496337890625,
      "learning_rate": 1.8030156148854492e-05,
      "loss": 1.9419856071472168,
      "step": 1687
    },
    {
      "epoch": 0.8128574008306748,
      "grad_norm": 13.223454475402832,
      "learning_rate": 1.7940906876237284e-05,
      "loss": 2.3655800819396973,
      "step": 1688
    },
    {
      "epoch": 0.8133389514235839,
      "grad_norm": 6.405287742614746,
      "learning_rate": 1.78518572711466e-05,
      "loss": 1.7890498638153076,
      "step": 1689
    },
    {
      "epoch": 0.8138205020164931,
      "grad_norm": 19.170087814331055,
      "learning_rate": 1.776300755025939e-05,
      "loss": 1.9310275316238403,
      "step": 1690
    },
    {
      "epoch": 0.8143020526094022,
      "grad_norm": 13.048337936401367,
      "learning_rate": 1.767435792976626e-05,
      "loss": 2.693742036819458,
      "step": 1691
    },
    {
      "epoch": 0.8147836032023115,
      "grad_norm": 20.831899642944336,
      "learning_rate": 1.7585908625370905e-05,
      "loss": 1.6789766550064087,
      "step": 1692
    },
    {
      "epoch": 0.8152651537952206,
      "grad_norm": 15.9860258102417,
      "learning_rate": 1.749765985228963e-05,
      "loss": 1.7934186458587646,
      "step": 1693
    },
    {
      "epoch": 0.8157467043881298,
      "grad_norm": 73.03245544433594,
      "learning_rate": 1.740961182525077e-05,
      "loss": 2.3263254165649414,
      "step": 1694
    },
    {
      "epoch": 0.8162282549810389,
      "grad_norm": 7.20707368850708,
      "learning_rate": 1.7321764758494252e-05,
      "loss": 2.5864484310150146,
      "step": 1695
    },
    {
      "epoch": 0.8167098055739481,
      "grad_norm": 6.920950889587402,
      "learning_rate": 1.7234118865770987e-05,
      "loss": 2.7824034690856934,
      "step": 1696
    },
    {
      "epoch": 0.8171913561668572,
      "grad_norm": 21.115922927856445,
      "learning_rate": 1.7146674360342373e-05,
      "loss": 2.9465112686157227,
      "step": 1697
    },
    {
      "epoch": 0.8176729067597664,
      "grad_norm": 119.02877807617188,
      "learning_rate": 1.7059431454979824e-05,
      "loss": 2.8785901069641113,
      "step": 1698
    },
    {
      "epoch": 0.8181544573526757,
      "grad_norm": 24.940753936767578,
      "learning_rate": 1.6972390361964195e-05,
      "loss": 2.3310813903808594,
      "step": 1699
    },
    {
      "epoch": 0.8186360079455848,
      "grad_norm": 14.924835205078125,
      "learning_rate": 1.688555129308531e-05,
      "loss": 2.286611318588257,
      "step": 1700
    },
    {
      "epoch": 0.819117558538494,
      "grad_norm": 673.6541748046875,
      "learning_rate": 1.6798914459641434e-05,
      "loss": 2.1311590671539307,
      "step": 1701
    },
    {
      "epoch": 0.8195991091314031,
      "grad_norm": 58.93374252319336,
      "learning_rate": 1.6712480072438662e-05,
      "loss": 2.237276315689087,
      "step": 1702
    },
    {
      "epoch": 0.8200806597243123,
      "grad_norm": 19.234285354614258,
      "learning_rate": 1.6626248341790596e-05,
      "loss": 2.480745315551758,
      "step": 1703
    },
    {
      "epoch": 0.8205622103172214,
      "grad_norm": 14.931989669799805,
      "learning_rate": 1.6540219477517684e-05,
      "loss": 2.736008644104004,
      "step": 1704
    },
    {
      "epoch": 0.8210437609101306,
      "grad_norm": 17.632658004760742,
      "learning_rate": 1.6454393688946767e-05,
      "loss": 2.2615182399749756,
      "step": 1705
    },
    {
      "epoch": 0.8215253115030398,
      "grad_norm": 10.56692123413086,
      "learning_rate": 1.6368771184910557e-05,
      "loss": 2.6031484603881836,
      "step": 1706
    },
    {
      "epoch": 0.822006862095949,
      "grad_norm": 39.547489166259766,
      "learning_rate": 1.6283352173747145e-05,
      "loss": 2.6019723415374756,
      "step": 1707
    },
    {
      "epoch": 0.8224884126888581,
      "grad_norm": 59.730751037597656,
      "learning_rate": 1.619813686329946e-05,
      "loss": 2.1798150539398193,
      "step": 1708
    },
    {
      "epoch": 0.8229699632817673,
      "grad_norm": 26.919492721557617,
      "learning_rate": 1.611312546091476e-05,
      "loss": 2.5908329486846924,
      "step": 1709
    },
    {
      "epoch": 0.8234515138746764,
      "grad_norm": 13.531916618347168,
      "learning_rate": 1.6028318173444202e-05,
      "loss": 2.6482393741607666,
      "step": 1710
    },
    {
      "epoch": 0.8239330644675856,
      "grad_norm": 12.53933334350586,
      "learning_rate": 1.594371520724226e-05,
      "loss": 2.2038941383361816,
      "step": 1711
    },
    {
      "epoch": 0.8244146150604948,
      "grad_norm": 25.19108772277832,
      "learning_rate": 1.5859316768166244e-05,
      "loss": 1.853979229927063,
      "step": 1712
    },
    {
      "epoch": 0.824896165653404,
      "grad_norm": 9.68689250946045,
      "learning_rate": 1.5775123061575836e-05,
      "loss": 2.67887544631958,
      "step": 1713
    },
    {
      "epoch": 0.8253777162463132,
      "grad_norm": 10.160589218139648,
      "learning_rate": 1.569113429233252e-05,
      "loss": 1.9435373544692993,
      "step": 1714
    },
    {
      "epoch": 0.8258592668392223,
      "grad_norm": 25.849435806274414,
      "learning_rate": 1.5607350664799157e-05,
      "loss": 2.2453348636627197,
      "step": 1715
    },
    {
      "epoch": 0.8263408174321315,
      "grad_norm": 24.244094848632812,
      "learning_rate": 1.552377238283943e-05,
      "loss": 2.2064523696899414,
      "step": 1716
    },
    {
      "epoch": 0.8268223680250406,
      "grad_norm": 12.355450630187988,
      "learning_rate": 1.5440399649817385e-05,
      "loss": 2.209178924560547,
      "step": 1717
    },
    {
      "epoch": 0.8273039186179498,
      "grad_norm": 9.689085960388184,
      "learning_rate": 1.5357232668596933e-05,
      "loss": 1.8386706113815308,
      "step": 1718
    },
    {
      "epoch": 0.8277854692108589,
      "grad_norm": 16.319616317749023,
      "learning_rate": 1.5274271641541295e-05,
      "loss": 2.426541805267334,
      "step": 1719
    },
    {
      "epoch": 0.8282670198037682,
      "grad_norm": 12.601068496704102,
      "learning_rate": 1.5191516770512649e-05,
      "loss": 2.2863762378692627,
      "step": 1720
    },
    {
      "epoch": 0.8287485703966773,
      "grad_norm": 7.795726299285889,
      "learning_rate": 1.5108968256871437e-05,
      "loss": 2.278886556625366,
      "step": 1721
    },
    {
      "epoch": 0.8292301209895865,
      "grad_norm": 31.01878547668457,
      "learning_rate": 1.5026626301476087e-05,
      "loss": 2.061758279800415,
      "step": 1722
    },
    {
      "epoch": 0.8297116715824956,
      "grad_norm": 132.04676818847656,
      "learning_rate": 1.4944491104682379e-05,
      "loss": 2.1558616161346436,
      "step": 1723
    },
    {
      "epoch": 0.8301932221754048,
      "grad_norm": 59.68663787841797,
      "learning_rate": 1.4862562866343034e-05,
      "loss": 2.5534725189208984,
      "step": 1724
    },
    {
      "epoch": 0.8306747727683139,
      "grad_norm": 77.88793182373047,
      "learning_rate": 1.4780841785807164e-05,
      "loss": 2.726958751678467,
      "step": 1725
    },
    {
      "epoch": 0.8311563233612231,
      "grad_norm": 40.7107048034668,
      "learning_rate": 1.4699328061919848e-05,
      "loss": 2.7683303356170654,
      "step": 1726
    },
    {
      "epoch": 0.8316378739541324,
      "grad_norm": 25.056602478027344,
      "learning_rate": 1.4618021893021605e-05,
      "loss": 2.900151252746582,
      "step": 1727
    },
    {
      "epoch": 0.8321194245470415,
      "grad_norm": 42.86872482299805,
      "learning_rate": 1.453692347694794e-05,
      "loss": 2.3790745735168457,
      "step": 1728
    },
    {
      "epoch": 0.8326009751399507,
      "grad_norm": 41.763851165771484,
      "learning_rate": 1.4456033011028835e-05,
      "loss": 2.1202027797698975,
      "step": 1729
    },
    {
      "epoch": 0.8330825257328598,
      "grad_norm": 26.37014389038086,
      "learning_rate": 1.437535069208833e-05,
      "loss": 2.226165294647217,
      "step": 1730
    },
    {
      "epoch": 0.833564076325769,
      "grad_norm": 32.331626892089844,
      "learning_rate": 1.4294876716443906e-05,
      "loss": 1.8176771402359009,
      "step": 1731
    },
    {
      "epoch": 0.8340456269186781,
      "grad_norm": 21.676651000976562,
      "learning_rate": 1.4214611279906187e-05,
      "loss": 2.1915485858917236,
      "step": 1732
    },
    {
      "epoch": 0.8345271775115873,
      "grad_norm": 12.385884284973145,
      "learning_rate": 1.4134554577778337e-05,
      "loss": 2.130030870437622,
      "step": 1733
    },
    {
      "epoch": 0.8350087281044964,
      "grad_norm": 52.20359420776367,
      "learning_rate": 1.4054706804855634e-05,
      "loss": 2.4495089054107666,
      "step": 1734
    },
    {
      "epoch": 0.8354902786974057,
      "grad_norm": 55.49383544921875,
      "learning_rate": 1.3975068155424976e-05,
      "loss": 2.375330924987793,
      "step": 1735
    },
    {
      "epoch": 0.8359718292903148,
      "grad_norm": 9.762662887573242,
      "learning_rate": 1.3895638823264446e-05,
      "loss": 2.1161153316497803,
      "step": 1736
    },
    {
      "epoch": 0.836453379883224,
      "grad_norm": 20.450448989868164,
      "learning_rate": 1.3816419001642777e-05,
      "loss": 3.2676515579223633,
      "step": 1737
    },
    {
      "epoch": 0.8369349304761331,
      "grad_norm": 12.188035011291504,
      "learning_rate": 1.3737408883318948e-05,
      "loss": 2.6651337146759033,
      "step": 1738
    },
    {
      "epoch": 0.8374164810690423,
      "grad_norm": 18.78900718688965,
      "learning_rate": 1.365860866054165e-05,
      "loss": 1.6782009601593018,
      "step": 1739
    },
    {
      "epoch": 0.8378980316619515,
      "grad_norm": 79.50492095947266,
      "learning_rate": 1.358001852504891e-05,
      "loss": 2.2190253734588623,
      "step": 1740
    },
    {
      "epoch": 0.8383795822548606,
      "grad_norm": 12.550873756408691,
      "learning_rate": 1.3501638668067485e-05,
      "loss": 2.6884520053863525,
      "step": 1741
    },
    {
      "epoch": 0.8388611328477699,
      "grad_norm": 22.618450164794922,
      "learning_rate": 1.3423469280312562e-05,
      "loss": 2.368995189666748,
      "step": 1742
    },
    {
      "epoch": 0.839342683440679,
      "grad_norm": 121.9845199584961,
      "learning_rate": 1.3345510551987128e-05,
      "loss": 1.9224029779434204,
      "step": 1743
    },
    {
      "epoch": 0.8398242340335882,
      "grad_norm": 49.248138427734375,
      "learning_rate": 1.326776267278167e-05,
      "loss": 2.556452989578247,
      "step": 1744
    },
    {
      "epoch": 0.8403057846264973,
      "grad_norm": 53.63905715942383,
      "learning_rate": 1.3190225831873581e-05,
      "loss": 1.9962321519851685,
      "step": 1745
    },
    {
      "epoch": 0.8407873352194065,
      "grad_norm": 9.97878360748291,
      "learning_rate": 1.3112900217926782e-05,
      "loss": 1.8901442289352417,
      "step": 1746
    },
    {
      "epoch": 0.8412688858123156,
      "grad_norm": 29.118379592895508,
      "learning_rate": 1.3035786019091223e-05,
      "loss": 2.38714337348938,
      "step": 1747
    },
    {
      "epoch": 0.8417504364052248,
      "grad_norm": 33.04164123535156,
      "learning_rate": 1.2958883423002422e-05,
      "loss": 2.3598837852478027,
      "step": 1748
    },
    {
      "epoch": 0.842231986998134,
      "grad_norm": 18.927539825439453,
      "learning_rate": 1.288219261678103e-05,
      "loss": 2.332634210586548,
      "step": 1749
    },
    {
      "epoch": 0.8427135375910432,
      "grad_norm": 27.491254806518555,
      "learning_rate": 1.2805713787032381e-05,
      "loss": 2.2470650672912598,
      "step": 1750
    },
    {
      "epoch": 0.8431950881839523,
      "grad_norm": 45.392086029052734,
      "learning_rate": 1.2729447119846016e-05,
      "loss": 1.636900782585144,
      "step": 1751
    },
    {
      "epoch": 0.8436766387768615,
      "grad_norm": 19.090238571166992,
      "learning_rate": 1.265339280079525e-05,
      "loss": 2.036206007003784,
      "step": 1752
    },
    {
      "epoch": 0.8441581893697706,
      "grad_norm": 33.57917404174805,
      "learning_rate": 1.257755101493665e-05,
      "loss": 2.15912127494812,
      "step": 1753
    },
    {
      "epoch": 0.8446397399626798,
      "grad_norm": 70.89653015136719,
      "learning_rate": 1.2501921946809714e-05,
      "loss": 2.12508225440979,
      "step": 1754
    },
    {
      "epoch": 0.845121290555589,
      "grad_norm": 13.496138572692871,
      "learning_rate": 1.2426505780436326e-05,
      "loss": 2.0174143314361572,
      "step": 1755
    },
    {
      "epoch": 0.8456028411484982,
      "grad_norm": 23.939395904541016,
      "learning_rate": 1.2351302699320332e-05,
      "loss": 2.171762704849243,
      "step": 1756
    },
    {
      "epoch": 0.8460843917414074,
      "grad_norm": 125.99385833740234,
      "learning_rate": 1.2276312886447106e-05,
      "loss": 2.496366024017334,
      "step": 1757
    },
    {
      "epoch": 0.8465659423343165,
      "grad_norm": 184.74864196777344,
      "learning_rate": 1.2201536524283074e-05,
      "loss": 1.5828849077224731,
      "step": 1758
    },
    {
      "epoch": 0.8470474929272257,
      "grad_norm": 29.959121704101562,
      "learning_rate": 1.2126973794775343e-05,
      "loss": 2.406085252761841,
      "step": 1759
    },
    {
      "epoch": 0.8475290435201348,
      "grad_norm": 10.193273544311523,
      "learning_rate": 1.2052624879351104e-05,
      "loss": 1.8349852561950684,
      "step": 1760
    },
    {
      "epoch": 0.848010594113044,
      "grad_norm": 19.860424041748047,
      "learning_rate": 1.1978489958917382e-05,
      "loss": 2.735891103744507,
      "step": 1761
    },
    {
      "epoch": 0.8484921447059531,
      "grad_norm": 52.98332595825195,
      "learning_rate": 1.1904569213860472e-05,
      "loss": 2.3324196338653564,
      "step": 1762
    },
    {
      "epoch": 0.8489736952988624,
      "grad_norm": 44.120819091796875,
      "learning_rate": 1.1830862824045552e-05,
      "loss": 2.035529136657715,
      "step": 1763
    },
    {
      "epoch": 0.8494552458917715,
      "grad_norm": 103.28450012207031,
      "learning_rate": 1.1757370968816217e-05,
      "loss": 2.140004873275757,
      "step": 1764
    },
    {
      "epoch": 0.8499367964846807,
      "grad_norm": 775.3075561523438,
      "learning_rate": 1.1684093826994024e-05,
      "loss": 2.020720958709717,
      "step": 1765
    },
    {
      "epoch": 0.8504183470775898,
      "grad_norm": 9.194197654724121,
      "learning_rate": 1.1611031576878117e-05,
      "loss": 2.0987462997436523,
      "step": 1766
    },
    {
      "epoch": 0.850899897670499,
      "grad_norm": 16.45020294189453,
      "learning_rate": 1.1538184396244778e-05,
      "loss": 1.6809371709823608,
      "step": 1767
    },
    {
      "epoch": 0.8513814482634082,
      "grad_norm": 22.138771057128906,
      "learning_rate": 1.146555246234694e-05,
      "loss": 2.1221797466278076,
      "step": 1768
    },
    {
      "epoch": 0.8518629988563173,
      "grad_norm": 49.573387145996094,
      "learning_rate": 1.1393135951913824e-05,
      "loss": 2.026611328125,
      "step": 1769
    },
    {
      "epoch": 0.8523445494492266,
      "grad_norm": 104.6827621459961,
      "learning_rate": 1.132093504115046e-05,
      "loss": 2.4885098934173584,
      "step": 1770
    },
    {
      "epoch": 0.8528261000421357,
      "grad_norm": 15.384543418884277,
      "learning_rate": 1.1248949905737283e-05,
      "loss": 2.230156183242798,
      "step": 1771
    },
    {
      "epoch": 0.8533076506350449,
      "grad_norm": 17.310894012451172,
      "learning_rate": 1.1177180720829694e-05,
      "loss": 2.2457916736602783,
      "step": 1772
    },
    {
      "epoch": 0.853789201227954,
      "grad_norm": 17.21358871459961,
      "learning_rate": 1.1105627661057671e-05,
      "loss": 1.9556950330734253,
      "step": 1773
    },
    {
      "epoch": 0.8542707518208632,
      "grad_norm": 13.73275089263916,
      "learning_rate": 1.103429090052528e-05,
      "loss": 2.4906153678894043,
      "step": 1774
    },
    {
      "epoch": 0.8547523024137723,
      "grad_norm": 12.734793663024902,
      "learning_rate": 1.096317061281027e-05,
      "loss": 1.8558112382888794,
      "step": 1775
    },
    {
      "epoch": 0.8552338530066815,
      "grad_norm": 11.212618827819824,
      "learning_rate": 1.0892266970963704e-05,
      "loss": 2.2882015705108643,
      "step": 1776
    },
    {
      "epoch": 0.8557154035995906,
      "grad_norm": 11.96105670928955,
      "learning_rate": 1.082158014750948e-05,
      "loss": 2.5987117290496826,
      "step": 1777
    },
    {
      "epoch": 0.8561969541924999,
      "grad_norm": 34.44051742553711,
      "learning_rate": 1.0751110314443958e-05,
      "loss": 2.7807278633117676,
      "step": 1778
    },
    {
      "epoch": 0.856678504785409,
      "grad_norm": 22.92962646484375,
      "learning_rate": 1.0680857643235431e-05,
      "loss": 2.271592140197754,
      "step": 1779
    },
    {
      "epoch": 0.8571600553783182,
      "grad_norm": 10.802287101745605,
      "learning_rate": 1.0610822304823887e-05,
      "loss": 2.506913900375366,
      "step": 1780
    },
    {
      "epoch": 0.8576416059712274,
      "grad_norm": 26.7536678314209,
      "learning_rate": 1.0541004469620452e-05,
      "loss": 2.6156816482543945,
      "step": 1781
    },
    {
      "epoch": 0.8581231565641365,
      "grad_norm": 10.344137191772461,
      "learning_rate": 1.0471404307507016e-05,
      "loss": 2.364393949508667,
      "step": 1782
    },
    {
      "epoch": 0.8586047071570457,
      "grad_norm": 19.775697708129883,
      "learning_rate": 1.0402021987835831e-05,
      "loss": 2.981182336807251,
      "step": 1783
    },
    {
      "epoch": 0.8590862577499548,
      "grad_norm": 50.16659164428711,
      "learning_rate": 1.0332857679429098e-05,
      "loss": 2.4871888160705566,
      "step": 1784
    },
    {
      "epoch": 0.8595678083428641,
      "grad_norm": 35.830474853515625,
      "learning_rate": 1.0263911550578531e-05,
      "loss": 2.1594748497009277,
      "step": 1785
    },
    {
      "epoch": 0.8600493589357732,
      "grad_norm": 20.722991943359375,
      "learning_rate": 1.0195183769045013e-05,
      "loss": 1.9320988655090332,
      "step": 1786
    },
    {
      "epoch": 0.8605309095286824,
      "grad_norm": 31.27431297302246,
      "learning_rate": 1.0126674502058054e-05,
      "loss": 2.4648470878601074,
      "step": 1787
    },
    {
      "epoch": 0.8610124601215915,
      "grad_norm": 47.39214324951172,
      "learning_rate": 1.005838391631555e-05,
      "loss": 2.3427817821502686,
      "step": 1788
    },
    {
      "epoch": 0.8614940107145007,
      "grad_norm": 15.365518569946289,
      "learning_rate": 9.990312177983263e-06,
      "loss": 2.128221035003662,
      "step": 1789
    },
    {
      "epoch": 0.8619755613074098,
      "grad_norm": 119.07408142089844,
      "learning_rate": 9.922459452694466e-06,
      "loss": 2.505864143371582,
      "step": 1790
    },
    {
      "epoch": 0.862457111900319,
      "grad_norm": 28.9871883392334,
      "learning_rate": 9.854825905549503e-06,
      "loss": 2.0256214141845703,
      "step": 1791
    },
    {
      "epoch": 0.8629386624932281,
      "grad_norm": 20.011960983276367,
      "learning_rate": 9.787411701115456e-06,
      "loss": 2.309047222137451,
      "step": 1792
    },
    {
      "epoch": 0.8634202130861374,
      "grad_norm": 14.734883308410645,
      "learning_rate": 9.720217003425647e-06,
      "loss": 2.5731613636016846,
      "step": 1793
    },
    {
      "epoch": 0.8639017636790465,
      "grad_norm": 10.657934188842773,
      "learning_rate": 9.65324197597931e-06,
      "loss": 2.248896598815918,
      "step": 1794
    },
    {
      "epoch": 0.8643833142719557,
      "grad_norm": 13.048442840576172,
      "learning_rate": 9.58648678174121e-06,
      "loss": 1.3597673177719116,
      "step": 1795
    },
    {
      "epoch": 0.8648648648648649,
      "grad_norm": 263.8738098144531,
      "learning_rate": 9.51995158314113e-06,
      "loss": 1.9085001945495605,
      "step": 1796
    },
    {
      "epoch": 0.865346415457774,
      "grad_norm": 625.9962158203125,
      "learning_rate": 9.45363654207363e-06,
      "loss": 2.385519504547119,
      "step": 1797
    },
    {
      "epoch": 0.8658279660506832,
      "grad_norm": 16.73454475402832,
      "learning_rate": 9.387541819897549e-06,
      "loss": 2.023244619369507,
      "step": 1798
    },
    {
      "epoch": 0.8663095166435923,
      "grad_norm": 56.181453704833984,
      "learning_rate": 9.321667577435634e-06,
      "loss": 2.599057197570801,
      "step": 1799
    },
    {
      "epoch": 0.8667910672365016,
      "grad_norm": 14.19897174835205,
      "learning_rate": 9.256013974974175e-06,
      "loss": 1.800797462463379,
      "step": 1800
    },
    {
      "epoch": 0.8672726178294107,
      "grad_norm": 27.449462890625,
      "learning_rate": 9.19058117226258e-06,
      "loss": 2.3053979873657227,
      "step": 1801
    },
    {
      "epoch": 0.8677541684223199,
      "grad_norm": 20.3785400390625,
      "learning_rate": 9.125369328513034e-06,
      "loss": 2.3519537448883057,
      "step": 1802
    },
    {
      "epoch": 0.868235719015229,
      "grad_norm": 31.549026489257812,
      "learning_rate": 9.060378602400054e-06,
      "loss": 2.7452261447906494,
      "step": 1803
    },
    {
      "epoch": 0.8687172696081382,
      "grad_norm": 57.20258331298828,
      "learning_rate": 8.995609152060136e-06,
      "loss": 2.6669228076934814,
      "step": 1804
    },
    {
      "epoch": 0.8691988202010473,
      "grad_norm": 10.981582641601562,
      "learning_rate": 8.931061135091357e-06,
      "loss": 2.4050519466400146,
      "step": 1805
    },
    {
      "epoch": 0.8696803707939565,
      "grad_norm": 13.443211555480957,
      "learning_rate": 8.866734708553015e-06,
      "loss": 2.720674753189087,
      "step": 1806
    },
    {
      "epoch": 0.8701619213868657,
      "grad_norm": 85.9037857055664,
      "learning_rate": 8.802630028965242e-06,
      "loss": 1.8400115966796875,
      "step": 1807
    },
    {
      "epoch": 0.8706434719797749,
      "grad_norm": 49.75071334838867,
      "learning_rate": 8.738747252308555e-06,
      "loss": 2.399500608444214,
      "step": 1808
    },
    {
      "epoch": 0.8711250225726841,
      "grad_norm": 14.994614601135254,
      "learning_rate": 8.675086534023591e-06,
      "loss": 1.781163215637207,
      "step": 1809
    },
    {
      "epoch": 0.8716065731655932,
      "grad_norm": 104.98004150390625,
      "learning_rate": 8.611648029010643e-06,
      "loss": 2.268181562423706,
      "step": 1810
    },
    {
      "epoch": 0.8720881237585024,
      "grad_norm": 69.60195922851562,
      "learning_rate": 8.548431891629316e-06,
      "loss": 2.08390474319458,
      "step": 1811
    },
    {
      "epoch": 0.8725696743514115,
      "grad_norm": 71.62710571289062,
      "learning_rate": 8.485438275698154e-06,
      "loss": 2.1462807655334473,
      "step": 1812
    },
    {
      "epoch": 0.8730512249443207,
      "grad_norm": 8.39852523803711,
      "learning_rate": 8.422667334494249e-06,
      "loss": 2.168865203857422,
      "step": 1813
    },
    {
      "epoch": 0.8735327755372299,
      "grad_norm": 21.37485122680664,
      "learning_rate": 8.360119220752893e-06,
      "loss": 2.427374839782715,
      "step": 1814
    },
    {
      "epoch": 0.8740143261301391,
      "grad_norm": 12.292753219604492,
      "learning_rate": 8.297794086667165e-06,
      "loss": 1.5472667217254639,
      "step": 1815
    },
    {
      "epoch": 0.8744958767230482,
      "grad_norm": 26.986820220947266,
      "learning_rate": 8.235692083887613e-06,
      "loss": 2.1685938835144043,
      "step": 1816
    },
    {
      "epoch": 0.8749774273159574,
      "grad_norm": 11.217364311218262,
      "learning_rate": 8.173813363521843e-06,
      "loss": 2.1815266609191895,
      "step": 1817
    },
    {
      "epoch": 0.8754589779088665,
      "grad_norm": 44.48500061035156,
      "learning_rate": 8.112158076134157e-06,
      "loss": 2.2966487407684326,
      "step": 1818
    },
    {
      "epoch": 0.8759405285017757,
      "grad_norm": 23.248502731323242,
      "learning_rate": 8.05072637174522e-06,
      "loss": 2.016343116760254,
      "step": 1819
    },
    {
      "epoch": 0.8764220790946848,
      "grad_norm": 353.4232482910156,
      "learning_rate": 7.989518399831641e-06,
      "loss": 2.0148723125457764,
      "step": 1820
    },
    {
      "epoch": 0.876903629687594,
      "grad_norm": 9.667152404785156,
      "learning_rate": 7.928534309325675e-06,
      "loss": 2.1355738639831543,
      "step": 1821
    },
    {
      "epoch": 0.8773851802805033,
      "grad_norm": 163.0660400390625,
      "learning_rate": 7.8677742486148e-06,
      "loss": 2.135883331298828,
      "step": 1822
    },
    {
      "epoch": 0.8778667308734124,
      "grad_norm": 33.93982696533203,
      "learning_rate": 7.807238365541391e-06,
      "loss": 2.0563571453094482,
      "step": 1823
    },
    {
      "epoch": 0.8783482814663216,
      "grad_norm": 42.838340759277344,
      "learning_rate": 7.746926807402344e-06,
      "loss": 2.215298891067505,
      "step": 1824
    },
    {
      "epoch": 0.8788298320592307,
      "grad_norm": 10.279427528381348,
      "learning_rate": 7.686839720948736e-06,
      "loss": 2.497452735900879,
      "step": 1825
    },
    {
      "epoch": 0.8793113826521399,
      "grad_norm": 55.821617126464844,
      "learning_rate": 7.6269772523854365e-06,
      "loss": 2.10611629486084,
      "step": 1826
    },
    {
      "epoch": 0.879792933245049,
      "grad_norm": 20.165124893188477,
      "learning_rate": 7.567339547370789e-06,
      "loss": 2.3833611011505127,
      "step": 1827
    },
    {
      "epoch": 0.8802744838379583,
      "grad_norm": 13.183784484863281,
      "learning_rate": 7.507926751016248e-06,
      "loss": 1.9828100204467773,
      "step": 1828
    },
    {
      "epoch": 0.8807560344308674,
      "grad_norm": 36.07333755493164,
      "learning_rate": 7.4487390078859855e-06,
      "loss": 1.802869439125061,
      "step": 1829
    },
    {
      "epoch": 0.8812375850237766,
      "grad_norm": 17.62275505065918,
      "learning_rate": 7.389776461996578e-06,
      "loss": 2.176570177078247,
      "step": 1830
    },
    {
      "epoch": 0.8817191356166857,
      "grad_norm": 28.91176414489746,
      "learning_rate": 7.331039256816663e-06,
      "loss": 2.8018016815185547,
      "step": 1831
    },
    {
      "epoch": 0.8822006862095949,
      "grad_norm": 40.658634185791016,
      "learning_rate": 7.27252753526656e-06,
      "loss": 2.2399063110351562,
      "step": 1832
    },
    {
      "epoch": 0.882682236802504,
      "grad_norm": 10.302037239074707,
      "learning_rate": 7.214241439717962e-06,
      "loss": 2.4459478855133057,
      "step": 1833
    },
    {
      "epoch": 0.8831637873954132,
      "grad_norm": 17.88764190673828,
      "learning_rate": 7.1561811119935425e-06,
      "loss": 2.563390016555786,
      "step": 1834
    },
    {
      "epoch": 0.8836453379883223,
      "grad_norm": 20.619155883789062,
      "learning_rate": 7.098346693366642e-06,
      "loss": 2.381293296813965,
      "step": 1835
    },
    {
      "epoch": 0.8841268885812316,
      "grad_norm": 6.883835315704346,
      "learning_rate": 7.0407383245609136e-06,
      "loss": 1.9837093353271484,
      "step": 1836
    },
    {
      "epoch": 0.8846084391741408,
      "grad_norm": 31.909618377685547,
      "learning_rate": 6.983356145749975e-06,
      "loss": 1.757419228553772,
      "step": 1837
    },
    {
      "epoch": 0.8850899897670499,
      "grad_norm": 13.818788528442383,
      "learning_rate": 6.9262002965570835e-06,
      "loss": 2.4753684997558594,
      "step": 1838
    },
    {
      "epoch": 0.8855715403599591,
      "grad_norm": 21.634008407592773,
      "learning_rate": 6.869270916054782e-06,
      "loss": 1.6436930894851685,
      "step": 1839
    },
    {
      "epoch": 0.8860530909528682,
      "grad_norm": 33.229251861572266,
      "learning_rate": 6.812568142764575e-06,
      "loss": 2.386049747467041,
      "step": 1840
    },
    {
      "epoch": 0.8865346415457774,
      "grad_norm": 10.53374195098877,
      "learning_rate": 6.756092114656587e-06,
      "loss": 2.1908419132232666,
      "step": 1841
    },
    {
      "epoch": 0.8870161921386865,
      "grad_norm": 87.08748626708984,
      "learning_rate": 6.699842969149195e-06,
      "loss": 2.330183982849121,
      "step": 1842
    },
    {
      "epoch": 0.8874977427315958,
      "grad_norm": 119.005615234375,
      "learning_rate": 6.64382084310875e-06,
      "loss": 2.224841356277466,
      "step": 1843
    },
    {
      "epoch": 0.8879792933245049,
      "grad_norm": 12.903675079345703,
      "learning_rate": 6.5880258728491905e-06,
      "loss": 1.9411665201187134,
      "step": 1844
    },
    {
      "epoch": 0.8884608439174141,
      "grad_norm": 15.647506713867188,
      "learning_rate": 6.532458194131763e-06,
      "loss": 1.7138543128967285,
      "step": 1845
    },
    {
      "epoch": 0.8889423945103232,
      "grad_norm": 16.521108627319336,
      "learning_rate": 6.477117942164657e-06,
      "loss": 2.553086280822754,
      "step": 1846
    },
    {
      "epoch": 0.8894239451032324,
      "grad_norm": 10.649314880371094,
      "learning_rate": 6.422005251602658e-06,
      "loss": 1.9098258018493652,
      "step": 1847
    },
    {
      "epoch": 0.8899054956961415,
      "grad_norm": 40.38874053955078,
      "learning_rate": 6.367120256546888e-06,
      "loss": 2.070422887802124,
      "step": 1848
    },
    {
      "epoch": 0.8903870462890507,
      "grad_norm": 4.736035346984863,
      "learning_rate": 6.312463090544396e-06,
      "loss": 1.6170673370361328,
      "step": 1849
    },
    {
      "epoch": 0.89086859688196,
      "grad_norm": 32.9072265625,
      "learning_rate": 6.258033886587911e-06,
      "loss": 2.734808921813965,
      "step": 1850
    },
    {
      "epoch": 0.8913501474748691,
      "grad_norm": 75.83673095703125,
      "learning_rate": 6.2038327771154485e-06,
      "loss": 2.1002702713012695,
      "step": 1851
    },
    {
      "epoch": 0.8918316980677783,
      "grad_norm": 15.051196098327637,
      "learning_rate": 6.1498598940100346e-06,
      "loss": 1.6587945222854614,
      "step": 1852
    },
    {
      "epoch": 0.8923132486606874,
      "grad_norm": 10.953733444213867,
      "learning_rate": 6.0961153685993646e-06,
      "loss": 2.1192615032196045,
      "step": 1853
    },
    {
      "epoch": 0.8927947992535966,
      "grad_norm": 32.58938217163086,
      "learning_rate": 6.0425993316554965e-06,
      "loss": 2.041821002960205,
      "step": 1854
    },
    {
      "epoch": 0.8932763498465057,
      "grad_norm": 31.042783737182617,
      "learning_rate": 5.989311913394546e-06,
      "loss": 2.1215896606445312,
      "step": 1855
    },
    {
      "epoch": 0.8937579004394149,
      "grad_norm": 14.42559814453125,
      "learning_rate": 5.93625324347632e-06,
      "loss": 1.7974070310592651,
      "step": 1856
    },
    {
      "epoch": 0.894239451032324,
      "grad_norm": 13.686972618103027,
      "learning_rate": 5.8834234510040335e-06,
      "loss": 1.8919804096221924,
      "step": 1857
    },
    {
      "epoch": 0.8947210016252333,
      "grad_norm": 39.59521484375,
      "learning_rate": 5.830822664523994e-06,
      "loss": 1.6902875900268555,
      "step": 1858
    },
    {
      "epoch": 0.8952025522181424,
      "grad_norm": 7.735287666320801,
      "learning_rate": 5.77845101202531e-06,
      "loss": 2.243987798690796,
      "step": 1859
    },
    {
      "epoch": 0.8956841028110516,
      "grad_norm": 9.256468772888184,
      "learning_rate": 5.726308620939536e-06,
      "loss": 1.5881404876708984,
      "step": 1860
    },
    {
      "epoch": 0.8961656534039607,
      "grad_norm": 53.057281494140625,
      "learning_rate": 5.674395618140393e-06,
      "loss": 2.166440725326538,
      "step": 1861
    },
    {
      "epoch": 0.8966472039968699,
      "grad_norm": 89.9701919555664,
      "learning_rate": 5.622712129943453e-06,
      "loss": 2.4564766883850098,
      "step": 1862
    },
    {
      "epoch": 0.8971287545897791,
      "grad_norm": 7.472794532775879,
      "learning_rate": 5.571258282105829e-06,
      "loss": 2.1239330768585205,
      "step": 1863
    },
    {
      "epoch": 0.8976103051826883,
      "grad_norm": 42.896610260009766,
      "learning_rate": 5.520034199825841e-06,
      "loss": 2.161186456680298,
      "step": 1864
    },
    {
      "epoch": 0.8980918557755975,
      "grad_norm": 49.448875427246094,
      "learning_rate": 5.469040007742776e-06,
      "loss": 1.4966919422149658,
      "step": 1865
    },
    {
      "epoch": 0.8985734063685066,
      "grad_norm": 8.561729431152344,
      "learning_rate": 5.418275829936537e-06,
      "loss": 2.4374399185180664,
      "step": 1866
    },
    {
      "epoch": 0.8990549569614158,
      "grad_norm": 16.040908813476562,
      "learning_rate": 5.36774178992735e-06,
      "loss": 2.230736494064331,
      "step": 1867
    },
    {
      "epoch": 0.8995365075543249,
      "grad_norm": 31.85891342163086,
      "learning_rate": 5.317438010675469e-06,
      "loss": 2.203746795654297,
      "step": 1868
    },
    {
      "epoch": 0.9000180581472341,
      "grad_norm": 12.181364059448242,
      "learning_rate": 5.267364614580861e-06,
      "loss": 2.4433701038360596,
      "step": 1869
    },
    {
      "epoch": 0.9004996087401432,
      "grad_norm": 27.01870346069336,
      "learning_rate": 5.217521723482943e-06,
      "loss": 1.7476731538772583,
      "step": 1870
    },
    {
      "epoch": 0.9009811593330525,
      "grad_norm": 10.05484676361084,
      "learning_rate": 5.167909458660258e-06,
      "loss": 2.0234534740448,
      "step": 1871
    },
    {
      "epoch": 0.9014627099259616,
      "grad_norm": 6.110547065734863,
      "learning_rate": 5.118527940830165e-06,
      "loss": 1.7689430713653564,
      "step": 1872
    },
    {
      "epoch": 0.9019442605188708,
      "grad_norm": 49.18951416015625,
      "learning_rate": 5.069377290148602e-06,
      "loss": 2.209658145904541,
      "step": 1873
    },
    {
      "epoch": 0.9024258111117799,
      "grad_norm": 24.131223678588867,
      "learning_rate": 5.020457626209707e-06,
      "loss": 1.9901940822601318,
      "step": 1874
    },
    {
      "epoch": 0.9029073617046891,
      "grad_norm": 60.522926330566406,
      "learning_rate": 4.971769068045628e-06,
      "loss": 2.2893788814544678,
      "step": 1875
    },
    {
      "epoch": 0.9033889122975982,
      "grad_norm": 8.565832138061523,
      "learning_rate": 4.923311734126135e-06,
      "loss": 1.7633949518203735,
      "step": 1876
    },
    {
      "epoch": 0.9038704628905074,
      "grad_norm": 16.79183006286621,
      "learning_rate": 4.875085742358432e-06,
      "loss": 2.1932575702667236,
      "step": 1877
    },
    {
      "epoch": 0.9043520134834167,
      "grad_norm": 30.315893173217773,
      "learning_rate": 4.827091210086776e-06,
      "loss": 2.435506582260132,
      "step": 1878
    },
    {
      "epoch": 0.9048335640763258,
      "grad_norm": 14.30863094329834,
      "learning_rate": 4.779328254092252e-06,
      "loss": 2.0386955738067627,
      "step": 1879
    },
    {
      "epoch": 0.905315114669235,
      "grad_norm": 1586.3836669921875,
      "learning_rate": 4.731796990592452e-06,
      "loss": 2.913670063018799,
      "step": 1880
    },
    {
      "epoch": 0.9057966652621441,
      "grad_norm": 31.395780563354492,
      "learning_rate": 4.68449753524125e-06,
      "loss": 2.1484696865081787,
      "step": 1881
    },
    {
      "epoch": 0.9062782158550533,
      "grad_norm": 194.7786865234375,
      "learning_rate": 4.637430003128429e-06,
      "loss": 1.9289233684539795,
      "step": 1882
    },
    {
      "epoch": 0.9067597664479624,
      "grad_norm": 42.657203674316406,
      "learning_rate": 4.5905945087794996e-06,
      "loss": 2.6290740966796875,
      "step": 1883
    },
    {
      "epoch": 0.9072413170408716,
      "grad_norm": 28.2672119140625,
      "learning_rate": 4.543991166155337e-06,
      "loss": 1.8920466899871826,
      "step": 1884
    },
    {
      "epoch": 0.9077228676337807,
      "grad_norm": 24.503767013549805,
      "learning_rate": 4.497620088651966e-06,
      "loss": 2.5785610675811768,
      "step": 1885
    },
    {
      "epoch": 0.90820441822669,
      "grad_norm": 101.07157135009766,
      "learning_rate": 4.451481389100232e-06,
      "loss": 2.0475687980651855,
      "step": 1886
    },
    {
      "epoch": 0.9086859688195991,
      "grad_norm": 58.47396469116211,
      "learning_rate": 4.405575179765586e-06,
      "loss": 2.431293249130249,
      "step": 1887
    },
    {
      "epoch": 0.9091675194125083,
      "grad_norm": 12.806317329406738,
      "learning_rate": 4.359901572347758e-06,
      "loss": 1.9634805917739868,
      "step": 1888
    },
    {
      "epoch": 0.9096490700054174,
      "grad_norm": 10.933167457580566,
      "learning_rate": 4.314460677980537e-06,
      "loss": 2.8796465396881104,
      "step": 1889
    },
    {
      "epoch": 0.9101306205983266,
      "grad_norm": 13.837821960449219,
      "learning_rate": 4.269252607231422e-06,
      "loss": 1.8437820672988892,
      "step": 1890
    },
    {
      "epoch": 0.9106121711912358,
      "grad_norm": 17.16478157043457,
      "learning_rate": 4.224277470101445e-06,
      "loss": 2.3884592056274414,
      "step": 1891
    },
    {
      "epoch": 0.9110937217841449,
      "grad_norm": 19.890758514404297,
      "learning_rate": 4.179535376024857e-06,
      "loss": 2.1411943435668945,
      "step": 1892
    },
    {
      "epoch": 0.9115752723770542,
      "grad_norm": 28.730344772338867,
      "learning_rate": 4.135026433868827e-06,
      "loss": 1.9595941305160522,
      "step": 1893
    },
    {
      "epoch": 0.9120568229699633,
      "grad_norm": 114.01387023925781,
      "learning_rate": 4.090750751933248e-06,
      "loss": 2.173799991607666,
      "step": 1894
    },
    {
      "epoch": 0.9125383735628725,
      "grad_norm": 24.017370223999023,
      "learning_rate": 4.046708437950464e-06,
      "loss": 2.3254764080047607,
      "step": 1895
    },
    {
      "epoch": 0.9130199241557816,
      "grad_norm": 21.790517807006836,
      "learning_rate": 4.0028995990849084e-06,
      "loss": 2.1315817832946777,
      "step": 1896
    },
    {
      "epoch": 0.9135014747486908,
      "grad_norm": 21.933189392089844,
      "learning_rate": 3.95932434193299e-06,
      "loss": 1.3415451049804688,
      "step": 1897
    },
    {
      "epoch": 0.9139830253415999,
      "grad_norm": 33.66352462768555,
      "learning_rate": 3.915982772522719e-06,
      "loss": 2.717515468597412,
      "step": 1898
    },
    {
      "epoch": 0.9144645759345091,
      "grad_norm": 42.704612731933594,
      "learning_rate": 3.872874996313513e-06,
      "loss": 2.3408310413360596,
      "step": 1899
    },
    {
      "epoch": 0.9149461265274182,
      "grad_norm": 18.800764083862305,
      "learning_rate": 3.830001118195936e-06,
      "loss": 2.314470052719116,
      "step": 1900
    },
    {
      "epoch": 0.9154276771203275,
      "grad_norm": 25.35037612915039,
      "learning_rate": 3.787361242491394e-06,
      "loss": 2.0337181091308594,
      "step": 1901
    },
    {
      "epoch": 0.9159092277132366,
      "grad_norm": 35.942665100097656,
      "learning_rate": 3.744955472951928e-06,
      "loss": 2.4466049671173096,
      "step": 1902
    },
    {
      "epoch": 0.9163907783061458,
      "grad_norm": 29.930850982666016,
      "learning_rate": 3.702783912759955e-06,
      "loss": 2.290684938430786,
      "step": 1903
    },
    {
      "epoch": 0.916872328899055,
      "grad_norm": 198.25140380859375,
      "learning_rate": 3.660846664528006e-06,
      "loss": 2.0863454341888428,
      "step": 1904
    },
    {
      "epoch": 0.9173538794919641,
      "grad_norm": 22.110591888427734,
      "learning_rate": 3.6191438302984772e-06,
      "loss": 2.7567031383514404,
      "step": 1905
    },
    {
      "epoch": 0.9178354300848733,
      "grad_norm": 31.0425968170166,
      "learning_rate": 3.577675511543388e-06,
      "loss": 2.028440237045288,
      "step": 1906
    },
    {
      "epoch": 0.9183169806777824,
      "grad_norm": 29.279407501220703,
      "learning_rate": 3.5364418091641373e-06,
      "loss": 3.019869565963745,
      "step": 1907
    },
    {
      "epoch": 0.9187985312706917,
      "grad_norm": 67.96923828125,
      "learning_rate": 3.495442823491224e-06,
      "loss": 2.5527541637420654,
      "step": 1908
    },
    {
      "epoch": 0.9192800818636008,
      "grad_norm": 14.775664329528809,
      "learning_rate": 3.4546786542840605e-06,
      "loss": 1.8357912302017212,
      "step": 1909
    },
    {
      "epoch": 0.91976163245651,
      "grad_norm": 63.909706115722656,
      "learning_rate": 3.4141494007306816e-06,
      "loss": 2.399369239807129,
      "step": 1910
    },
    {
      "epoch": 0.9202431830494191,
      "grad_norm": 42.50132369995117,
      "learning_rate": 3.373855161447548e-06,
      "loss": 2.6662278175354004,
      "step": 1911
    },
    {
      "epoch": 0.9207247336423283,
      "grad_norm": 17.58905601501465,
      "learning_rate": 3.333796034479242e-06,
      "loss": 2.699321746826172,
      "step": 1912
    },
    {
      "epoch": 0.9212062842352374,
      "grad_norm": 17.556442260742188,
      "learning_rate": 3.293972117298294e-06,
      "loss": 1.8133213520050049,
      "step": 1913
    },
    {
      "epoch": 0.9216878348281466,
      "grad_norm": 18.869897842407227,
      "learning_rate": 3.2543835068049255e-06,
      "loss": 2.0924360752105713,
      "step": 1914
    },
    {
      "epoch": 0.9221693854210558,
      "grad_norm": 1251.0399169921875,
      "learning_rate": 3.21503029932676e-06,
      "loss": 1.7048006057739258,
      "step": 1915
    },
    {
      "epoch": 0.922650936013965,
      "grad_norm": 63.76348876953125,
      "learning_rate": 3.1759125906186793e-06,
      "loss": 2.011449098587036,
      "step": 1916
    },
    {
      "epoch": 0.9231324866068741,
      "grad_norm": 40.91250991821289,
      "learning_rate": 3.137030475862535e-06,
      "loss": 1.9411365985870361,
      "step": 1917
    },
    {
      "epoch": 0.9236140371997833,
      "grad_norm": 21.720945358276367,
      "learning_rate": 3.098384049666925e-06,
      "loss": 2.4920778274536133,
      "step": 1918
    },
    {
      "epoch": 0.9240955877926925,
      "grad_norm": 14.017812728881836,
      "learning_rate": 3.059973406066963e-06,
      "loss": 2.2390174865722656,
      "step": 1919
    },
    {
      "epoch": 0.9245771383856016,
      "grad_norm": 30.95956039428711,
      "learning_rate": 3.0217986385240537e-06,
      "loss": 2.246201753616333,
      "step": 1920
    },
    {
      "epoch": 0.9250586889785108,
      "grad_norm": 39.85332489013672,
      "learning_rate": 2.983859839925662e-06,
      "loss": 1.9972989559173584,
      "step": 1921
    },
    {
      "epoch": 0.92554023957142,
      "grad_norm": 166.24850463867188,
      "learning_rate": 2.94615710258509e-06,
      "loss": 2.6342241764068604,
      "step": 1922
    },
    {
      "epoch": 0.9260217901643292,
      "grad_norm": 24.094608306884766,
      "learning_rate": 2.908690518241275e-06,
      "loss": 2.113696813583374,
      "step": 1923
    },
    {
      "epoch": 0.9265033407572383,
      "grad_norm": 8.129045486450195,
      "learning_rate": 2.8714601780584937e-06,
      "loss": 2.0441579818725586,
      "step": 1924
    },
    {
      "epoch": 0.9269848913501475,
      "grad_norm": 22.600740432739258,
      "learning_rate": 2.834466172626238e-06,
      "loss": 1.3354748487472534,
      "step": 1925
    },
    {
      "epoch": 0.9274664419430566,
      "grad_norm": 11.49627685546875,
      "learning_rate": 2.7977085919589254e-06,
      "loss": 2.112112045288086,
      "step": 1926
    },
    {
      "epoch": 0.9279479925359658,
      "grad_norm": 21.147592544555664,
      "learning_rate": 2.76118752549569e-06,
      "loss": 2.1954360008239746,
      "step": 1927
    },
    {
      "epoch": 0.9284295431288749,
      "grad_norm": 36.1245231628418,
      "learning_rate": 2.7249030621001924e-06,
      "loss": 2.0960018634796143,
      "step": 1928
    },
    {
      "epoch": 0.9289110937217842,
      "grad_norm": 43.56359100341797,
      "learning_rate": 2.688855290060399e-06,
      "loss": 2.799917459487915,
      "step": 1929
    },
    {
      "epoch": 0.9293926443146933,
      "grad_norm": 28.046796798706055,
      "learning_rate": 2.653044297088314e-06,
      "loss": 2.2113633155822754,
      "step": 1930
    },
    {
      "epoch": 0.9298741949076025,
      "grad_norm": 35.95022201538086,
      "learning_rate": 2.6174701703198468e-06,
      "loss": 1.9227635860443115,
      "step": 1931
    },
    {
      "epoch": 0.9303557455005117,
      "grad_norm": 97.79923248291016,
      "learning_rate": 2.5821329963145347e-06,
      "loss": 2.8213865756988525,
      "step": 1932
    },
    {
      "epoch": 0.9308372960934208,
      "grad_norm": 36.23052978515625,
      "learning_rate": 2.547032861055376e-06,
      "loss": 2.600921154022217,
      "step": 1933
    },
    {
      "epoch": 0.93131884668633,
      "grad_norm": 121.48738098144531,
      "learning_rate": 2.5121698499485757e-06,
      "loss": 2.086285352706909,
      "step": 1934
    },
    {
      "epoch": 0.9318003972792391,
      "grad_norm": 58.71704864501953,
      "learning_rate": 2.4775440478233993e-06,
      "loss": 2.111034393310547,
      "step": 1935
    },
    {
      "epoch": 0.9322819478721484,
      "grad_norm": 10.052783012390137,
      "learning_rate": 2.4431555389319074e-06,
      "loss": 1.9900968074798584,
      "step": 1936
    },
    {
      "epoch": 0.9327634984650575,
      "grad_norm": 26.118776321411133,
      "learning_rate": 2.4090044069487784e-06,
      "loss": 2.0652575492858887,
      "step": 1937
    },
    {
      "epoch": 0.9332450490579667,
      "grad_norm": 23.427587509155273,
      "learning_rate": 2.3750907349711084e-06,
      "loss": 2.211749792098999,
      "step": 1938
    },
    {
      "epoch": 0.9337265996508758,
      "grad_norm": 35.970088958740234,
      "learning_rate": 2.3414146055182106e-06,
      "loss": 2.1825449466705322,
      "step": 1939
    },
    {
      "epoch": 0.934208150243785,
      "grad_norm": 40.166839599609375,
      "learning_rate": 2.307976100531384e-06,
      "loss": 2.898446559906006,
      "step": 1940
    },
    {
      "epoch": 0.9346897008366941,
      "grad_norm": 10.43848991394043,
      "learning_rate": 2.274775301373744e-06,
      "loss": 2.0363521575927734,
      "step": 1941
    },
    {
      "epoch": 0.9351712514296033,
      "grad_norm": 52.165367126464844,
      "learning_rate": 2.241812288830003e-06,
      "loss": 2.7197790145874023,
      "step": 1942
    },
    {
      "epoch": 0.9356528020225124,
      "grad_norm": 11.094637870788574,
      "learning_rate": 2.2090871431063253e-06,
      "loss": 1.7552502155303955,
      "step": 1943
    },
    {
      "epoch": 0.9361343526154217,
      "grad_norm": 36.51998519897461,
      "learning_rate": 2.176599943830071e-06,
      "loss": 1.7154202461242676,
      "step": 1944
    },
    {
      "epoch": 0.9366159032083309,
      "grad_norm": 13.132946968078613,
      "learning_rate": 2.144350770049597e-06,
      "loss": 1.768015742301941,
      "step": 1945
    },
    {
      "epoch": 0.93709745380124,
      "grad_norm": 15.87300968170166,
      "learning_rate": 2.112339700234156e-06,
      "loss": 2.1072158813476562,
      "step": 1946
    },
    {
      "epoch": 0.9375790043941492,
      "grad_norm": 12.473407745361328,
      "learning_rate": 2.0805668122735767e-06,
      "loss": 2.338819980621338,
      "step": 1947
    },
    {
      "epoch": 0.9380605549870583,
      "grad_norm": 14.842493057250977,
      "learning_rate": 2.0490321834781833e-06,
      "loss": 1.8490805625915527,
      "step": 1948
    },
    {
      "epoch": 0.9385421055799675,
      "grad_norm": 56.81089782714844,
      "learning_rate": 2.0177358905785537e-06,
      "loss": 2.3557591438293457,
      "step": 1949
    },
    {
      "epoch": 0.9390236561728766,
      "grad_norm": 24.732330322265625,
      "learning_rate": 1.986678009725329e-06,
      "loss": 2.722177743911743,
      "step": 1950
    },
    {
      "epoch": 0.9395052067657859,
      "grad_norm": 25.13091278076172,
      "learning_rate": 1.955858616489059e-06,
      "loss": 2.2072181701660156,
      "step": 1951
    },
    {
      "epoch": 0.939986757358695,
      "grad_norm": 15.112622261047363,
      "learning_rate": 1.9252777858599915e-06,
      "loss": 2.199179172515869,
      "step": 1952
    },
    {
      "epoch": 0.9404683079516042,
      "grad_norm": 12.926004409790039,
      "learning_rate": 1.8949355922479151e-06,
      "loss": 2.162067413330078,
      "step": 1953
    },
    {
      "epoch": 0.9409498585445133,
      "grad_norm": 33.79193878173828,
      "learning_rate": 1.8648321094819287e-06,
      "loss": 2.9513967037200928,
      "step": 1954
    },
    {
      "epoch": 0.9414314091374225,
      "grad_norm": 12.362686157226562,
      "learning_rate": 1.8349674108103288e-06,
      "loss": 1.8950437307357788,
      "step": 1955
    },
    {
      "epoch": 0.9419129597303316,
      "grad_norm": 55.049930572509766,
      "learning_rate": 1.8053415689003872e-06,
      "loss": 2.353412628173828,
      "step": 1956
    },
    {
      "epoch": 0.9423945103232408,
      "grad_norm": 19.846500396728516,
      "learning_rate": 1.7759546558381967e-06,
      "loss": 1.8579858541488647,
      "step": 1957
    },
    {
      "epoch": 0.94287606091615,
      "grad_norm": 21.206253051757812,
      "learning_rate": 1.7468067431284707e-06,
      "loss": 2.0582149028778076,
      "step": 1958
    },
    {
      "epoch": 0.9433576115090592,
      "grad_norm": 12.491701126098633,
      "learning_rate": 1.7178979016943764e-06,
      "loss": 2.4236104488372803,
      "step": 1959
    },
    {
      "epoch": 0.9438391621019684,
      "grad_norm": 80.1990966796875,
      "learning_rate": 1.6892282018773908e-06,
      "loss": 2.007695436477661,
      "step": 1960
    },
    {
      "epoch": 0.9443207126948775,
      "grad_norm": 180.36056518554688,
      "learning_rate": 1.6607977134370789e-06,
      "loss": 3.2147603034973145,
      "step": 1961
    },
    {
      "epoch": 0.9448022632877867,
      "grad_norm": 23.851600646972656,
      "learning_rate": 1.6326065055510043e-06,
      "loss": 2.0905470848083496,
      "step": 1962
    },
    {
      "epoch": 0.9452838138806958,
      "grad_norm": 56.005348205566406,
      "learning_rate": 1.6046546468144407e-06,
      "loss": 1.8979460000991821,
      "step": 1963
    },
    {
      "epoch": 0.945765364473605,
      "grad_norm": 6.4673895835876465,
      "learning_rate": 1.576942205240317e-06,
      "loss": 2.247066020965576,
      "step": 1964
    },
    {
      "epoch": 0.9462469150665141,
      "grad_norm": 12.563465118408203,
      "learning_rate": 1.5494692482590057e-06,
      "loss": 1.7862352132797241,
      "step": 1965
    },
    {
      "epoch": 0.9467284656594234,
      "grad_norm": 17.85274314880371,
      "learning_rate": 1.522235842718156e-06,
      "loss": 1.9563432931900024,
      "step": 1966
    },
    {
      "epoch": 0.9472100162523325,
      "grad_norm": 22.442462921142578,
      "learning_rate": 1.4952420548825285e-06,
      "loss": 2.9954779148101807,
      "step": 1967
    },
    {
      "epoch": 0.9476915668452417,
      "grad_norm": 11.334410667419434,
      "learning_rate": 1.468487950433839e-06,
      "loss": 2.3425846099853516,
      "step": 1968
    },
    {
      "epoch": 0.9481731174381508,
      "grad_norm": 37.680992126464844,
      "learning_rate": 1.441973594470636e-06,
      "loss": 1.8482640981674194,
      "step": 1969
    },
    {
      "epoch": 0.94865466803106,
      "grad_norm": 6.412022113800049,
      "learning_rate": 1.415699051508068e-06,
      "loss": 2.2075445652008057,
      "step": 1970
    },
    {
      "epoch": 0.9491362186239691,
      "grad_norm": 17.832040786743164,
      "learning_rate": 1.3896643854777847e-06,
      "loss": 2.0605664253234863,
      "step": 1971
    },
    {
      "epoch": 0.9496177692168783,
      "grad_norm": 47.10072326660156,
      "learning_rate": 1.3638696597277679e-06,
      "loss": 2.1241676807403564,
      "step": 1972
    },
    {
      "epoch": 0.9500993198097876,
      "grad_norm": 30.611520767211914,
      "learning_rate": 1.3383149370221449e-06,
      "loss": 1.9927486181259155,
      "step": 1973
    },
    {
      "epoch": 0.9505808704026967,
      "grad_norm": 14.688706398010254,
      "learning_rate": 1.313000279541121e-06,
      "loss": 2.142468214035034,
      "step": 1974
    },
    {
      "epoch": 0.9510624209956059,
      "grad_norm": 16.909814834594727,
      "learning_rate": 1.287925748880703e-06,
      "loss": 2.3127191066741943,
      "step": 1975
    },
    {
      "epoch": 0.951543971588515,
      "grad_norm": 110.1868896484375,
      "learning_rate": 1.2630914060526522e-06,
      "loss": 2.8561222553253174,
      "step": 1976
    },
    {
      "epoch": 0.9520255221814242,
      "grad_norm": 13.730730056762695,
      "learning_rate": 1.2384973114843101e-06,
      "loss": 1.723078966140747,
      "step": 1977
    },
    {
      "epoch": 0.9525070727743333,
      "grad_norm": 17.550888061523438,
      "learning_rate": 1.2141435250184185e-06,
      "loss": 1.6527966260910034,
      "step": 1978
    },
    {
      "epoch": 0.9529886233672425,
      "grad_norm": 32.18467330932617,
      "learning_rate": 1.1900301059130093e-06,
      "loss": 1.9391158819198608,
      "step": 1979
    },
    {
      "epoch": 0.9534701739601517,
      "grad_norm": 13.833096504211426,
      "learning_rate": 1.1661571128412596e-06,
      "loss": 2.0655930042266846,
      "step": 1980
    },
    {
      "epoch": 0.9539517245530609,
      "grad_norm": 11.753029823303223,
      "learning_rate": 1.142524603891315e-06,
      "loss": 1.977735996246338,
      "step": 1981
    },
    {
      "epoch": 0.95443327514597,
      "grad_norm": 5.810583114624023,
      "learning_rate": 1.1191326365661892e-06,
      "loss": 1.9064180850982666,
      "step": 1982
    },
    {
      "epoch": 0.9549148257388792,
      "grad_norm": 15.158056259155273,
      "learning_rate": 1.0959812677835968e-06,
      "loss": 2.078388214111328,
      "step": 1983
    },
    {
      "epoch": 0.9553963763317883,
      "grad_norm": 62.080482482910156,
      "learning_rate": 1.0730705538758322e-06,
      "loss": 1.6790275573730469,
      "step": 1984
    },
    {
      "epoch": 0.9558779269246975,
      "grad_norm": 22.87973403930664,
      "learning_rate": 1.0504005505896141e-06,
      "loss": 1.9969377517700195,
      "step": 1985
    },
    {
      "epoch": 0.9563594775176067,
      "grad_norm": 10.44893741607666,
      "learning_rate": 1.0279713130859514e-06,
      "loss": 2.2060599327087402,
      "step": 1986
    },
    {
      "epoch": 0.9568410281105159,
      "grad_norm": 29.99155044555664,
      "learning_rate": 1.005782895940055e-06,
      "loss": 2.710313320159912,
      "step": 1987
    },
    {
      "epoch": 0.9573225787034251,
      "grad_norm": 44.77143859863281,
      "learning_rate": 9.838353531411272e-07,
      "loss": 1.948403000831604,
      "step": 1988
    },
    {
      "epoch": 0.9578041292963342,
      "grad_norm": 21.18767547607422,
      "learning_rate": 9.62128738092294e-07,
      "loss": 1.9304475784301758,
      "step": 1989
    },
    {
      "epoch": 0.9582856798892434,
      "grad_norm": 23.128162384033203,
      "learning_rate": 9.406631036104508e-07,
      "loss": 2.1398375034332275,
      "step": 1990
    },
    {
      "epoch": 0.9587672304821525,
      "grad_norm": 24.259889602661133,
      "learning_rate": 9.194385019261287e-07,
      "loss": 2.212322473526001,
      "step": 1991
    },
    {
      "epoch": 0.9592487810750617,
      "grad_norm": 12.365840911865234,
      "learning_rate": 8.984549846833612e-07,
      "loss": 2.4376609325408936,
      "step": 1992
    },
    {
      "epoch": 0.9597303316679708,
      "grad_norm": 18.16153907775879,
      "learning_rate": 8.777126029396065e-07,
      "loss": 2.0726394653320312,
      "step": 1993
    },
    {
      "epoch": 0.9602118822608801,
      "grad_norm": 93.86848449707031,
      "learning_rate": 8.572114071655479e-07,
      "loss": 2.248323917388916,
      "step": 1994
    },
    {
      "epoch": 0.9606934328537892,
      "grad_norm": 16.439420700073242,
      "learning_rate": 8.369514472450379e-07,
      "loss": 1.864798665046692,
      "step": 1995
    },
    {
      "epoch": 0.9611749834466984,
      "grad_norm": 57.86463928222656,
      "learning_rate": 8.169327724749543e-07,
      "loss": 3.0946967601776123,
      "step": 1996
    },
    {
      "epoch": 0.9616565340396075,
      "grad_norm": 143.62803649902344,
      "learning_rate": 7.971554315650442e-07,
      "loss": 2.1420068740844727,
      "step": 1997
    },
    {
      "epoch": 0.9621380846325167,
      "grad_norm": 21.849573135375977,
      "learning_rate": 7.776194726378583e-07,
      "loss": 1.755669355392456,
      "step": 1998
    },
    {
      "epoch": 0.9626196352254258,
      "grad_norm": 149.3284912109375,
      "learning_rate": 7.583249432286277e-07,
      "loss": 1.501633644104004,
      "step": 1999
    },
    {
      "epoch": 0.963101185818335,
      "grad_norm": 30.780540466308594,
      "learning_rate": 7.392718902850981e-07,
      "loss": 2.5893287658691406,
      "step": 2000
    },
    {
      "epoch": 0.9635827364112443,
      "grad_norm": 16.58056640625,
      "learning_rate": 7.204603601674853e-07,
      "loss": 2.9102697372436523,
      "step": 2001
    },
    {
      "epoch": 0.9640642870041534,
      "grad_norm": 885.3781127929688,
      "learning_rate": 7.018903986483083e-07,
      "loss": 2.300450086593628,
      "step": 2002
    },
    {
      "epoch": 0.9645458375970626,
      "grad_norm": 18.293195724487305,
      "learning_rate": 6.835620509122897e-07,
      "loss": 2.1224780082702637,
      "step": 2003
    },
    {
      "epoch": 0.9650273881899717,
      "grad_norm": 44.274391174316406,
      "learning_rate": 6.65475361556267e-07,
      "loss": 2.4352009296417236,
      "step": 2004
    },
    {
      "epoch": 0.9655089387828809,
      "grad_norm": 16.461580276489258,
      "learning_rate": 6.47630374589081e-07,
      "loss": 2.2931811809539795,
      "step": 2005
    },
    {
      "epoch": 0.96599048937579,
      "grad_norm": 8.748879432678223,
      "learning_rate": 6.300271334314434e-07,
      "loss": 2.0041396617889404,
      "step": 2006
    },
    {
      "epoch": 0.9664720399686992,
      "grad_norm": 14.340997695922852,
      "learning_rate": 6.126656809158359e-07,
      "loss": 2.0829880237579346,
      "step": 2007
    },
    {
      "epoch": 0.9669535905616083,
      "grad_norm": 207.8635711669922,
      "learning_rate": 5.955460592864337e-07,
      "loss": 1.7841203212738037,
      "step": 2008
    },
    {
      "epoch": 0.9674351411545176,
      "grad_norm": 89.33834838867188,
      "learning_rate": 5.78668310198982e-07,
      "loss": 2.38802433013916,
      "step": 2009
    },
    {
      "epoch": 0.9679166917474267,
      "grad_norm": 39.36704635620117,
      "learning_rate": 5.620324747207084e-07,
      "loss": 2.456850290298462,
      "step": 2010
    },
    {
      "epoch": 0.9683982423403359,
      "grad_norm": 46.56597137451172,
      "learning_rate": 5.456385933301777e-07,
      "loss": 2.931748390197754,
      "step": 2011
    },
    {
      "epoch": 0.968879792933245,
      "grad_norm": 9.23050308227539,
      "learning_rate": 5.294867059172592e-07,
      "loss": 2.325681209564209,
      "step": 2012
    },
    {
      "epoch": 0.9693613435261542,
      "grad_norm": 15.760897636413574,
      "learning_rate": 5.135768517829819e-07,
      "loss": 1.8451409339904785,
      "step": 2013
    },
    {
      "epoch": 0.9698428941190634,
      "grad_norm": 20.972869873046875,
      "learning_rate": 4.979090696394795e-07,
      "loss": 1.7424386739730835,
      "step": 2014
    },
    {
      "epoch": 0.9703244447119725,
      "grad_norm": 11.033096313476562,
      "learning_rate": 4.824833976098453e-07,
      "loss": 1.6550421714782715,
      "step": 2015
    },
    {
      "epoch": 0.9708059953048818,
      "grad_norm": 28.175426483154297,
      "learning_rate": 4.6729987322807757e-07,
      "loss": 2.245985269546509,
      "step": 2016
    },
    {
      "epoch": 0.9712875458977909,
      "grad_norm": 17.284460067749023,
      "learning_rate": 4.523585334389679e-07,
      "loss": 2.19952654838562,
      "step": 2017
    },
    {
      "epoch": 0.9717690964907001,
      "grad_norm": 43.336395263671875,
      "learning_rate": 4.3765941459804614e-07,
      "loss": 2.6563990116119385,
      "step": 2018
    },
    {
      "epoch": 0.9722506470836092,
      "grad_norm": 17.026710510253906,
      "learning_rate": 4.232025524714356e-07,
      "loss": 2.6179757118225098,
      "step": 2019
    },
    {
      "epoch": 0.9727321976765184,
      "grad_norm": 14.4814453125,
      "learning_rate": 4.0898798223582e-07,
      "loss": 2.2732598781585693,
      "step": 2020
    },
    {
      "epoch": 0.9732137482694275,
      "grad_norm": 13.024195671081543,
      "learning_rate": 3.950157384783104e-07,
      "loss": 1.9380193948745728,
      "step": 2021
    },
    {
      "epoch": 0.9736952988623367,
      "grad_norm": 44.50840377807617,
      "learning_rate": 3.8128585519640046e-07,
      "loss": 2.5293796062469482,
      "step": 2022
    },
    {
      "epoch": 0.9741768494552459,
      "grad_norm": 63.38443374633789,
      "learning_rate": 3.677983657978779e-07,
      "loss": 2.9504075050354004,
      "step": 2023
    },
    {
      "epoch": 0.9746584000481551,
      "grad_norm": 44.5324592590332,
      "learning_rate": 3.545533031007131e-07,
      "loss": 2.1063218116760254,
      "step": 2024
    },
    {
      "epoch": 0.9751399506410642,
      "grad_norm": 29.781055450439453,
      "learning_rate": 3.415506993330153e-07,
      "loss": 2.2531397342681885,
      "step": 2025
    },
    {
      "epoch": 0.9756215012339734,
      "grad_norm": 21.797683715820312,
      "learning_rate": 3.2879058613292105e-07,
      "loss": 2.732382297515869,
      "step": 2026
    },
    {
      "epoch": 0.9761030518268826,
      "grad_norm": 10.050985336303711,
      "learning_rate": 3.1627299454856095e-07,
      "loss": 1.5147898197174072,
      "step": 2027
    },
    {
      "epoch": 0.9765846024197917,
      "grad_norm": 7.51145601272583,
      "learning_rate": 3.0399795503793793e-07,
      "loss": 1.5709898471832275,
      "step": 2028
    },
    {
      "epoch": 0.9770661530127009,
      "grad_norm": 29.968917846679688,
      "learning_rate": 2.9196549746888235e-07,
      "loss": 2.351571559906006,
      "step": 2029
    },
    {
      "epoch": 0.97754770360561,
      "grad_norm": 31.16327667236328,
      "learning_rate": 2.801756511189524e-07,
      "loss": 2.2602460384368896,
      "step": 2030
    },
    {
      "epoch": 0.9780292541985193,
      "grad_norm": 17.132740020751953,
      "learning_rate": 2.686284446754006e-07,
      "loss": 2.6187827587127686,
      "step": 2031
    },
    {
      "epoch": 0.9785108047914284,
      "grad_norm": 14.757771492004395,
      "learning_rate": 2.573239062350963e-07,
      "loss": 2.280402660369873,
      "step": 2032
    },
    {
      "epoch": 0.9789923553843376,
      "grad_norm": 58.21686553955078,
      "learning_rate": 2.4626206330440326e-07,
      "loss": 2.0190393924713135,
      "step": 2033
    },
    {
      "epoch": 0.9794739059772467,
      "grad_norm": 8.738277435302734,
      "learning_rate": 2.3544294279918e-07,
      "loss": 1.9083871841430664,
      "step": 2034
    },
    {
      "epoch": 0.9799554565701559,
      "grad_norm": 48.8358268737793,
      "learning_rate": 2.2486657104471286e-07,
      "loss": 2.104325771331787,
      "step": 2035
    },
    {
      "epoch": 0.980437007163065,
      "grad_norm": 15.376314163208008,
      "learning_rate": 2.1453297377557191e-07,
      "loss": 2.0072691440582275,
      "step": 2036
    },
    {
      "epoch": 0.9809185577559743,
      "grad_norm": 21.630741119384766,
      "learning_rate": 2.044421761356552e-07,
      "loss": 2.4897079467773438,
      "step": 2037
    },
    {
      "epoch": 0.9814001083488834,
      "grad_norm": 10.657557487487793,
      "learning_rate": 1.9459420267804452e-07,
      "loss": 2.601041793823242,
      "step": 2038
    },
    {
      "epoch": 0.9818816589417926,
      "grad_norm": 23.391599655151367,
      "learning_rate": 1.8498907736499426e-07,
      "loss": 2.3580880165100098,
      "step": 2039
    },
    {
      "epoch": 0.9823632095347017,
      "grad_norm": 17.25575065612793,
      "learning_rate": 1.7562682356786487e-07,
      "loss": 2.2205557823181152,
      "step": 2040
    },
    {
      "epoch": 0.9828447601276109,
      "grad_norm": 544.0999145507812,
      "learning_rate": 1.665074640670228e-07,
      "loss": 2.1871299743652344,
      "step": 2041
    },
    {
      "epoch": 0.9833263107205201,
      "grad_norm": 15.216567993164062,
      "learning_rate": 1.576310210518517e-07,
      "loss": 1.7740360498428345,
      "step": 2042
    },
    {
      "epoch": 0.9838078613134292,
      "grad_norm": 7.3857245445251465,
      "learning_rate": 1.489975161206636e-07,
      "loss": 1.6604362726211548,
      "step": 2043
    },
    {
      "epoch": 0.9842894119063385,
      "grad_norm": 14.518994331359863,
      "learning_rate": 1.406069702806323e-07,
      "loss": 1.8585578203201294,
      "step": 2044
    },
    {
      "epoch": 0.9847709624992476,
      "grad_norm": 6078.52099609375,
      "learning_rate": 1.324594039477822e-07,
      "loss": 2.347435712814331,
      "step": 2045
    },
    {
      "epoch": 0.9852525130921568,
      "grad_norm": 20.992015838623047,
      "learning_rate": 1.2455483694689962e-07,
      "loss": 2.04612135887146,
      "step": 2046
    },
    {
      "epoch": 0.9857340636850659,
      "grad_norm": 25.48192024230957,
      "learning_rate": 1.1689328851151038e-07,
      "loss": 1.592198371887207,
      "step": 2047
    },
    {
      "epoch": 0.9862156142779751,
      "grad_norm": 11.0081787109375,
      "learning_rate": 1.0947477728381339e-07,
      "loss": 2.073349714279175,
      "step": 2048
    },
    {
      "epoch": 0.9866971648708842,
      "grad_norm": 29.665855407714844,
      "learning_rate": 1.0229932131465836e-07,
      "loss": 2.2474145889282227,
      "step": 2049
    },
    {
      "epoch": 0.9871787154637934,
      "grad_norm": 32.335853576660156,
      "learning_rate": 9.536693806347919e-08,
      "loss": 2.6354339122772217,
      "step": 2050
    },
    {
      "epoch": 0.9876602660567025,
      "grad_norm": 16.846311569213867,
      "learning_rate": 8.867764439826065e-08,
      "loss": 1.9451637268066406,
      "step": 2051
    },
    {
      "epoch": 0.9881418166496118,
      "grad_norm": 25.940858840942383,
      "learning_rate": 8.223145659550513e-08,
      "loss": 1.524754524230957,
      "step": 2052
    },
    {
      "epoch": 0.9886233672425209,
      "grad_norm": 8.313727378845215,
      "learning_rate": 7.602839034017706e-08,
      "loss": 2.381373882293701,
      "step": 2053
    },
    {
      "epoch": 0.9891049178354301,
      "grad_norm": 15.637900352478027,
      "learning_rate": 7.006846072568074e-08,
      "loss": 2.071277379989624,
      "step": 2054
    },
    {
      "epoch": 0.9895864684283393,
      "grad_norm": 41.94395065307617,
      "learning_rate": 6.435168225381594e-08,
      "loss": 2.356889009475708,
      "step": 2055
    },
    {
      "epoch": 0.9900680190212484,
      "grad_norm": 133.923583984375,
      "learning_rate": 5.887806883474456e-08,
      "loss": 2.3005757331848145,
      "step": 2056
    },
    {
      "epoch": 0.9905495696141576,
      "grad_norm": 30.060537338256836,
      "learning_rate": 5.364763378694626e-08,
      "loss": 2.3172037601470947,
      "step": 2057
    },
    {
      "epoch": 0.9910311202070667,
      "grad_norm": 11.838141441345215,
      "learning_rate": 4.8660389837207334e-08,
      "loss": 2.3206021785736084,
      "step": 2058
    },
    {
      "epoch": 0.991512670799976,
      "grad_norm": 26.10407066345215,
      "learning_rate": 4.391634912056519e-08,
      "loss": 2.3354482650756836,
      "step": 2059
    },
    {
      "epoch": 0.9919942213928851,
      "grad_norm": 79.33258819580078,
      "learning_rate": 3.9415523180297286e-08,
      "loss": 1.9041240215301514,
      "step": 2060
    },
    {
      "epoch": 0.9924757719857943,
      "grad_norm": 13.12371826171875,
      "learning_rate": 3.515792296789888e-08,
      "loss": 2.4203431606292725,
      "step": 2061
    },
    {
      "epoch": 0.9929573225787034,
      "grad_norm": 19.153486251831055,
      "learning_rate": 3.114355884301645e-08,
      "loss": 2.3018882274627686,
      "step": 2062
    },
    {
      "epoch": 0.9934388731716126,
      "grad_norm": 73.83121490478516,
      "learning_rate": 2.7372440573469883e-08,
      "loss": 2.4884068965911865,
      "step": 2063
    },
    {
      "epoch": 0.9939204237645217,
      "grad_norm": 23.212751388549805,
      "learning_rate": 2.384457733520806e-08,
      "loss": 2.3176987171173096,
      "step": 2064
    },
    {
      "epoch": 0.9944019743574309,
      "grad_norm": 12.5691499710083,
      "learning_rate": 2.0559977712297785e-08,
      "loss": 2.7043590545654297,
      "step": 2065
    },
    {
      "epoch": 0.99488352495034,
      "grad_norm": 14.416475296020508,
      "learning_rate": 1.7518649696857126e-08,
      "loss": 2.165778636932373,
      "step": 2066
    },
    {
      "epoch": 0.9953650755432493,
      "grad_norm": 83.97898864746094,
      "learning_rate": 1.4720600689110963e-08,
      "loss": 2.1930623054504395,
      "step": 2067
    },
    {
      "epoch": 0.9958466261361584,
      "grad_norm": 11.684106826782227,
      "learning_rate": 1.216583749731326e-08,
      "loss": 2.06412672996521,
      "step": 2068
    },
    {
      "epoch": 0.9963281767290676,
      "grad_norm": 8.96517276763916,
      "learning_rate": 9.854366337758159e-09,
      "loss": 2.257368564605713,
      "step": 2069
    },
    {
      "epoch": 0.9968097273219768,
      "grad_norm": 59.222904205322266,
      "learning_rate": 7.786192834746686e-09,
      "loss": 1.8355387449264526,
      "step": 2070
    },
    {
      "epoch": 0.9972912779148859,
      "grad_norm": 22.878347396850586,
      "learning_rate": 5.961322020608951e-09,
      "loss": 1.6282103061676025,
      "step": 2071
    },
    {
      "epoch": 0.9977728285077951,
      "grad_norm": 29.799440383911133,
      "learning_rate": 4.3797583356264275e-09,
      "loss": 1.9030158519744873,
      "step": 2072
    },
    {
      "epoch": 0.9982543791007042,
      "grad_norm": 62.5737419128418,
      "learning_rate": 3.0415056281096755e-09,
      "loss": 2.2005770206451416,
      "step": 2073
    },
    {
      "epoch": 0.9987359296936135,
      "grad_norm": 29.084320068359375,
      "learning_rate": 1.9465671543095197e-09,
      "loss": 2.6216471195220947,
      "step": 2074
    },
    {
      "epoch": 0.9992174802865226,
      "grad_norm": 30.733848571777344,
      "learning_rate": 1.094945578439255e-09,
      "loss": 2.3288826942443848,
      "step": 2075
    },
    {
      "epoch": 0.9996990308794318,
      "grad_norm": 20.24508285522461,
      "learning_rate": 4.866429726857469e-10,
      "loss": 1.9480527639389038,
      "step": 2076
    },
    {
      "epoch": 1.0,
      "grad_norm": 32.08734893798828,
      "learning_rate": 1.2166081717612797e-10,
      "loss": 2.036438465118408,
      "step": 2077
    },
    {
      "epoch": 1.0,
      "step": 2077,
      "total_flos": 8.07297066432725e+17,
      "train_loss": 2.353319917209572,
      "train_runtime": 7490.9269,
      "train_samples_per_second": 4.435,
      "train_steps_per_second": 0.277
    }
  ],
  "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": 8.07297066432725e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}