{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9990426041168023,
  "eval_steps": 500,
  "global_step": 348,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0057443752991862135,
      "grad_norm": 2.941743850708008,
      "learning_rate": 2.8571428571428575e-07,
      "loss": 0.7786,
      "step": 1
    },
    {
      "epoch": 0.011488750598372427,
      "grad_norm": 3.0483286380767822,
      "learning_rate": 5.714285714285715e-07,
      "loss": 0.8077,
      "step": 2
    },
    {
      "epoch": 0.01723312589755864,
      "grad_norm": 2.9768505096435547,
      "learning_rate": 8.571428571428572e-07,
      "loss": 0.8335,
      "step": 3
    },
    {
      "epoch": 0.022977501196744854,
      "grad_norm": 2.8796703815460205,
      "learning_rate": 1.142857142857143e-06,
      "loss": 0.8027,
      "step": 4
    },
    {
      "epoch": 0.028721876495931067,
      "grad_norm": 2.885287284851074,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.7972,
      "step": 5
    },
    {
      "epoch": 0.03446625179511728,
      "grad_norm": 2.892397403717041,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 0.7879,
      "step": 6
    },
    {
      "epoch": 0.040210627094303494,
      "grad_norm": 2.2109262943267822,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.7884,
      "step": 7
    },
    {
      "epoch": 0.04595500239348971,
      "grad_norm": 2.0732390880584717,
      "learning_rate": 2.285714285714286e-06,
      "loss": 0.7702,
      "step": 8
    },
    {
      "epoch": 0.05169937769267592,
      "grad_norm": 1.657523512840271,
      "learning_rate": 2.571428571428571e-06,
      "loss": 0.7247,
      "step": 9
    },
    {
      "epoch": 0.057443752991862135,
      "grad_norm": 1.6407607793807983,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.7748,
      "step": 10
    },
    {
      "epoch": 0.06318812829104835,
      "grad_norm": 1.5466963052749634,
      "learning_rate": 3.142857142857143e-06,
      "loss": 0.7613,
      "step": 11
    },
    {
      "epoch": 0.06893250359023456,
      "grad_norm": 1.069846510887146,
      "learning_rate": 3.428571428571429e-06,
      "loss": 0.6725,
      "step": 12
    },
    {
      "epoch": 0.07467687888942078,
      "grad_norm": 1.4985568523406982,
      "learning_rate": 3.7142857142857146e-06,
      "loss": 0.7338,
      "step": 13
    },
    {
      "epoch": 0.08042125418860699,
      "grad_norm": 1.4400984048843384,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.716,
      "step": 14
    },
    {
      "epoch": 0.0861656294877932,
      "grad_norm": 1.2160851955413818,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.7243,
      "step": 15
    },
    {
      "epoch": 0.09191000478697942,
      "grad_norm": 0.9179856181144714,
      "learning_rate": 4.571428571428572e-06,
      "loss": 0.7202,
      "step": 16
    },
    {
      "epoch": 0.09765438008616563,
      "grad_norm": 1.0979514122009277,
      "learning_rate": 4.857142857142858e-06,
      "loss": 0.6752,
      "step": 17
    },
    {
      "epoch": 0.10339875538535184,
      "grad_norm": 1.1197757720947266,
      "learning_rate": 5.142857142857142e-06,
      "loss": 0.6756,
      "step": 18
    },
    {
      "epoch": 0.10914313068453806,
      "grad_norm": 0.9753621220588684,
      "learning_rate": 5.428571428571429e-06,
      "loss": 0.6144,
      "step": 19
    },
    {
      "epoch": 0.11488750598372427,
      "grad_norm": 0.8363606929779053,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.6365,
      "step": 20
    },
    {
      "epoch": 0.12063188128291048,
      "grad_norm": 0.6905666589736938,
      "learning_rate": 6e-06,
      "loss": 0.6227,
      "step": 21
    },
    {
      "epoch": 0.1263762565820967,
      "grad_norm": 0.6746470928192139,
      "learning_rate": 6.285714285714286e-06,
      "loss": 0.6171,
      "step": 22
    },
    {
      "epoch": 0.13212063188128292,
      "grad_norm": 0.7786882519721985,
      "learning_rate": 6.571428571428572e-06,
      "loss": 0.5791,
      "step": 23
    },
    {
      "epoch": 0.13786500718046912,
      "grad_norm": 0.9305371642112732,
      "learning_rate": 6.857142857142858e-06,
      "loss": 0.6146,
      "step": 24
    },
    {
      "epoch": 0.14360938247965535,
      "grad_norm": 0.8036319017410278,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.6009,
      "step": 25
    },
    {
      "epoch": 0.14935375777884155,
      "grad_norm": 0.7687404751777649,
      "learning_rate": 7.428571428571429e-06,
      "loss": 0.5954,
      "step": 26
    },
    {
      "epoch": 0.15509813307802778,
      "grad_norm": 0.7354080677032471,
      "learning_rate": 7.714285714285716e-06,
      "loss": 0.6264,
      "step": 27
    },
    {
      "epoch": 0.16084250837721398,
      "grad_norm": 0.6613245606422424,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.5887,
      "step": 28
    },
    {
      "epoch": 0.1665868836764002,
      "grad_norm": 0.588733971118927,
      "learning_rate": 8.285714285714287e-06,
      "loss": 0.5776,
      "step": 29
    },
    {
      "epoch": 0.1723312589755864,
      "grad_norm": 0.608691930770874,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.5829,
      "step": 30
    },
    {
      "epoch": 0.17807563427477263,
      "grad_norm": 0.5666121244430542,
      "learning_rate": 8.857142857142858e-06,
      "loss": 0.5441,
      "step": 31
    },
    {
      "epoch": 0.18382000957395883,
      "grad_norm": 0.6083720922470093,
      "learning_rate": 9.142857142857144e-06,
      "loss": 0.5249,
      "step": 32
    },
    {
      "epoch": 0.18956438487314506,
      "grad_norm": 0.4732661545276642,
      "learning_rate": 9.42857142857143e-06,
      "loss": 0.5628,
      "step": 33
    },
    {
      "epoch": 0.19530876017233126,
      "grad_norm": 0.5184394121170044,
      "learning_rate": 9.714285714285715e-06,
      "loss": 0.5817,
      "step": 34
    },
    {
      "epoch": 0.20105313547151749,
      "grad_norm": 0.4829459488391876,
      "learning_rate": 1e-05,
      "loss": 0.5679,
      "step": 35
    },
    {
      "epoch": 0.20679751077070369,
      "grad_norm": 0.44880038499832153,
      "learning_rate": 9.999748146823376e-06,
      "loss": 0.5596,
      "step": 36
    },
    {
      "epoch": 0.2125418860698899,
      "grad_norm": 0.4486270248889923,
      "learning_rate": 9.99899261266551e-06,
      "loss": 0.5477,
      "step": 37
    },
    {
      "epoch": 0.2182862613690761,
      "grad_norm": 0.48706376552581787,
      "learning_rate": 9.997733473639876e-06,
      "loss": 0.5591,
      "step": 38
    },
    {
      "epoch": 0.22403063666826234,
      "grad_norm": 0.42943134903907776,
      "learning_rate": 9.995970856593739e-06,
      "loss": 0.5597,
      "step": 39
    },
    {
      "epoch": 0.22977501196744854,
      "grad_norm": 0.41754963994026184,
      "learning_rate": 9.993704939095376e-06,
      "loss": 0.5397,
      "step": 40
    },
    {
      "epoch": 0.23551938726663477,
      "grad_norm": 0.45870938897132874,
      "learning_rate": 9.9909359494162e-06,
      "loss": 0.5792,
      "step": 41
    },
    {
      "epoch": 0.24126376256582097,
      "grad_norm": 0.390557199716568,
      "learning_rate": 9.987664166507749e-06,
      "loss": 0.5204,
      "step": 42
    },
    {
      "epoch": 0.2470081378650072,
      "grad_norm": 0.4505020081996918,
      "learning_rate": 9.983889919973586e-06,
      "loss": 0.5538,
      "step": 43
    },
    {
      "epoch": 0.2527525131641934,
      "grad_norm": 0.42829522490501404,
      "learning_rate": 9.979613590036108e-06,
      "loss": 0.5677,
      "step": 44
    },
    {
      "epoch": 0.2584968884633796,
      "grad_norm": 0.39550063014030457,
      "learning_rate": 9.974835607498224e-06,
      "loss": 0.552,
      "step": 45
    },
    {
      "epoch": 0.26424126376256585,
      "grad_norm": 0.3860190808773041,
      "learning_rate": 9.969556453699966e-06,
      "loss": 0.5084,
      "step": 46
    },
    {
      "epoch": 0.26998563906175205,
      "grad_norm": 0.40885424613952637,
      "learning_rate": 9.963776660469996e-06,
      "loss": 0.5358,
      "step": 47
    },
    {
      "epoch": 0.27573001436093825,
      "grad_norm": 0.41485726833343506,
      "learning_rate": 9.957496810072027e-06,
      "loss": 0.5219,
      "step": 48
    },
    {
      "epoch": 0.28147438966012445,
      "grad_norm": 0.3923221230506897,
      "learning_rate": 9.95071753514617e-06,
      "loss": 0.5547,
      "step": 49
    },
    {
      "epoch": 0.2872187649593107,
      "grad_norm": 0.3466614782810211,
      "learning_rate": 9.943439518645193e-06,
      "loss": 0.5143,
      "step": 50
    },
    {
      "epoch": 0.2929631402584969,
      "grad_norm": 0.38951000571250916,
      "learning_rate": 9.935663493765726e-06,
      "loss": 0.5034,
      "step": 51
    },
    {
      "epoch": 0.2987075155576831,
      "grad_norm": 0.3788608908653259,
      "learning_rate": 9.9273902438744e-06,
      "loss": 0.542,
      "step": 52
    },
    {
      "epoch": 0.3044518908568693,
      "grad_norm": 0.34517818689346313,
      "learning_rate": 9.918620602428916e-06,
      "loss": 0.5185,
      "step": 53
    },
    {
      "epoch": 0.31019626615605556,
      "grad_norm": 0.3572729527950287,
      "learning_rate": 9.909355452894098e-06,
      "loss": 0.5099,
      "step": 54
    },
    {
      "epoch": 0.31594064145524176,
      "grad_norm": 0.3773225247859955,
      "learning_rate": 9.899595728652883e-06,
      "loss": 0.5126,
      "step": 55
    },
    {
      "epoch": 0.32168501675442795,
      "grad_norm": 0.37288057804107666,
      "learning_rate": 9.889342412912296e-06,
      "loss": 0.5199,
      "step": 56
    },
    {
      "epoch": 0.32742939205361415,
      "grad_norm": 0.3700467050075531,
      "learning_rate": 9.878596538604388e-06,
      "loss": 0.5286,
      "step": 57
    },
    {
      "epoch": 0.3331737673528004,
      "grad_norm": 0.37380489706993103,
      "learning_rate": 9.867359188282193e-06,
      "loss": 0.523,
      "step": 58
    },
    {
      "epoch": 0.3389181426519866,
      "grad_norm": 0.38175636529922485,
      "learning_rate": 9.855631494010661e-06,
      "loss": 0.4975,
      "step": 59
    },
    {
      "epoch": 0.3446625179511728,
      "grad_norm": 0.35676026344299316,
      "learning_rate": 9.843414637252615e-06,
      "loss": 0.5079,
      "step": 60
    },
    {
      "epoch": 0.350406893250359,
      "grad_norm": 0.38748887181282043,
      "learning_rate": 9.830709848749727e-06,
      "loss": 0.5222,
      "step": 61
    },
    {
      "epoch": 0.35615126854954526,
      "grad_norm": 0.3811897933483124,
      "learning_rate": 9.817518408398536e-06,
      "loss": 0.5305,
      "step": 62
    },
    {
      "epoch": 0.36189564384873146,
      "grad_norm": 0.3687308728694916,
      "learning_rate": 9.803841645121505e-06,
      "loss": 0.5168,
      "step": 63
    },
    {
      "epoch": 0.36764001914791766,
      "grad_norm": 0.4523705840110779,
      "learning_rate": 9.78968093673314e-06,
      "loss": 0.5015,
      "step": 64
    },
    {
      "epoch": 0.37338439444710386,
      "grad_norm": 0.3872551918029785,
      "learning_rate": 9.775037709801206e-06,
      "loss": 0.5199,
      "step": 65
    },
    {
      "epoch": 0.3791287697462901,
      "grad_norm": 0.38410401344299316,
      "learning_rate": 9.759913439502982e-06,
      "loss": 0.515,
      "step": 66
    },
    {
      "epoch": 0.3848731450454763,
      "grad_norm": 0.38912612199783325,
      "learning_rate": 9.74430964947668e-06,
      "loss": 0.4769,
      "step": 67
    },
    {
      "epoch": 0.3906175203446625,
      "grad_norm": 0.37264540791511536,
      "learning_rate": 9.728227911667934e-06,
      "loss": 0.5011,
      "step": 68
    },
    {
      "epoch": 0.3963618956438487,
      "grad_norm": 0.35123029351234436,
      "learning_rate": 9.711669846171443e-06,
      "loss": 0.5249,
      "step": 69
    },
    {
      "epoch": 0.40210627094303497,
      "grad_norm": 0.35720133781433105,
      "learning_rate": 9.694637121067764e-06,
      "loss": 0.4552,
      "step": 70
    },
    {
      "epoch": 0.40785064624222117,
      "grad_norm": 0.3890661597251892,
      "learning_rate": 9.677131452255272e-06,
      "loss": 0.5047,
      "step": 71
    },
    {
      "epoch": 0.41359502154140737,
      "grad_norm": 0.38108378648757935,
      "learning_rate": 9.659154603277283e-06,
      "loss": 0.4861,
      "step": 72
    },
    {
      "epoch": 0.41933939684059357,
      "grad_norm": 0.3942843973636627,
      "learning_rate": 9.640708385144403e-06,
      "loss": 0.5032,
      "step": 73
    },
    {
      "epoch": 0.4250837721397798,
      "grad_norm": 0.3866446018218994,
      "learning_rate": 9.62179465615209e-06,
      "loss": 0.5063,
      "step": 74
    },
    {
      "epoch": 0.430828147438966,
      "grad_norm": 0.3848111927509308,
      "learning_rate": 9.602415321693434e-06,
      "loss": 0.5193,
      "step": 75
    },
    {
      "epoch": 0.4365725227381522,
      "grad_norm": 0.39589178562164307,
      "learning_rate": 9.582572334067213e-06,
      "loss": 0.4956,
      "step": 76
    },
    {
      "epoch": 0.4423168980373384,
      "grad_norm": 0.3862696886062622,
      "learning_rate": 9.562267692281212e-06,
      "loss": 0.5197,
      "step": 77
    },
    {
      "epoch": 0.4480612733365247,
      "grad_norm": 0.4226361811161041,
      "learning_rate": 9.541503441850844e-06,
      "loss": 0.4993,
      "step": 78
    },
    {
      "epoch": 0.4538056486357109,
      "grad_norm": 0.3534332513809204,
      "learning_rate": 9.520281674593084e-06,
      "loss": 0.5062,
      "step": 79
    },
    {
      "epoch": 0.4595500239348971,
      "grad_norm": 0.3645087480545044,
      "learning_rate": 9.498604528415731e-06,
      "loss": 0.4919,
      "step": 80
    },
    {
      "epoch": 0.4652943992340833,
      "grad_norm": 0.4237683117389679,
      "learning_rate": 9.476474187102033e-06,
      "loss": 0.4944,
      "step": 81
    },
    {
      "epoch": 0.47103877453326953,
      "grad_norm": 0.4125359356403351,
      "learning_rate": 9.453892880090696e-06,
      "loss": 0.5123,
      "step": 82
    },
    {
      "epoch": 0.47678314983245573,
      "grad_norm": 0.3767677843570709,
      "learning_rate": 9.430862882251279e-06,
      "loss": 0.4984,
      "step": 83
    },
    {
      "epoch": 0.48252752513164193,
      "grad_norm": 0.4958581030368805,
      "learning_rate": 9.40738651365503e-06,
      "loss": 0.5235,
      "step": 84
    },
    {
      "epoch": 0.48827190043082813,
      "grad_norm": 0.41700416803359985,
      "learning_rate": 9.38346613934115e-06,
      "loss": 0.4861,
      "step": 85
    },
    {
      "epoch": 0.4940162757300144,
      "grad_norm": 0.4114746153354645,
      "learning_rate": 9.359104169078541e-06,
      "loss": 0.4995,
      "step": 86
    },
    {
      "epoch": 0.4997606510292006,
      "grad_norm": 0.39890262484550476,
      "learning_rate": 9.334303057123044e-06,
      "loss": 0.517,
      "step": 87
    },
    {
      "epoch": 0.5055050263283868,
      "grad_norm": 0.4252322018146515,
      "learning_rate": 9.309065301970193e-06,
      "loss": 0.5029,
      "step": 88
    },
    {
      "epoch": 0.511249401627573,
      "grad_norm": 0.3892343044281006,
      "learning_rate": 9.283393446103506e-06,
      "loss": 0.4965,
      "step": 89
    },
    {
      "epoch": 0.5169937769267592,
      "grad_norm": 0.38548123836517334,
      "learning_rate": 9.257290075738365e-06,
      "loss": 0.5072,
      "step": 90
    },
    {
      "epoch": 0.5227381522259454,
      "grad_norm": 0.3773181140422821,
      "learning_rate": 9.23075782056147e-06,
      "loss": 0.4871,
      "step": 91
    },
    {
      "epoch": 0.5284825275251317,
      "grad_norm": 0.40583336353302,
      "learning_rate": 9.20379935346592e-06,
      "loss": 0.4916,
      "step": 92
    },
    {
      "epoch": 0.5342269028243178,
      "grad_norm": 0.5210088491439819,
      "learning_rate": 9.176417390281944e-06,
      "loss": 0.4814,
      "step": 93
    },
    {
      "epoch": 0.5399712781235041,
      "grad_norm": 0.4008769989013672,
      "learning_rate": 9.148614689503307e-06,
      "loss": 0.5102,
      "step": 94
    },
    {
      "epoch": 0.5457156534226902,
      "grad_norm": 0.3989771604537964,
      "learning_rate": 9.120394052009412e-06,
      "loss": 0.4954,
      "step": 95
    },
    {
      "epoch": 0.5514600287218765,
      "grad_norm": 0.3988707959651947,
      "learning_rate": 9.091758320783139e-06,
      "loss": 0.5083,
      "step": 96
    },
    {
      "epoch": 0.5572044040210627,
      "grad_norm": 0.395672470331192,
      "learning_rate": 9.062710380624439e-06,
      "loss": 0.4826,
      "step": 97
    },
    {
      "epoch": 0.5629487793202489,
      "grad_norm": 0.4042396545410156,
      "learning_rate": 9.033253157859715e-06,
      "loss": 0.5217,
      "step": 98
    },
    {
      "epoch": 0.5686931546194351,
      "grad_norm": 0.3654581904411316,
      "learning_rate": 9.003389620047012e-06,
      "loss": 0.4801,
      "step": 99
    },
    {
      "epoch": 0.5744375299186214,
      "grad_norm": 0.4231509864330292,
      "learning_rate": 8.973122775677078e-06,
      "loss": 0.4849,
      "step": 100
    },
    {
      "epoch": 0.5801819052178075,
      "grad_norm": 0.4183957874774933,
      "learning_rate": 8.942455673870278e-06,
      "loss": 0.513,
      "step": 101
    },
    {
      "epoch": 0.5859262805169938,
      "grad_norm": 0.4512512981891632,
      "learning_rate": 8.91139140406941e-06,
      "loss": 0.5104,
      "step": 102
    },
    {
      "epoch": 0.59167065581618,
      "grad_norm": 0.40022507309913635,
      "learning_rate": 8.879933095728485e-06,
      "loss": 0.4967,
      "step": 103
    },
    {
      "epoch": 0.5974150311153662,
      "grad_norm": 0.35647737979888916,
      "learning_rate": 8.848083917997463e-06,
      "loss": 0.4819,
      "step": 104
    },
    {
      "epoch": 0.6031594064145525,
      "grad_norm": 0.4126330614089966,
      "learning_rate": 8.815847079402972e-06,
      "loss": 0.4817,
      "step": 105
    },
    {
      "epoch": 0.6089037817137386,
      "grad_norm": 0.4161527752876282,
      "learning_rate": 8.783225827525098e-06,
      "loss": 0.5018,
      "step": 106
    },
    {
      "epoch": 0.6146481570129249,
      "grad_norm": 0.3914158344268799,
      "learning_rate": 8.750223448670204e-06,
      "loss": 0.4971,
      "step": 107
    },
    {
      "epoch": 0.6203925323121111,
      "grad_norm": 0.3953137695789337,
      "learning_rate": 8.716843267539868e-06,
      "loss": 0.5123,
      "step": 108
    },
    {
      "epoch": 0.6261369076112973,
      "grad_norm": 0.4234555959701538,
      "learning_rate": 8.683088646895955e-06,
      "loss": 0.462,
      "step": 109
    },
    {
      "epoch": 0.6318812829104835,
      "grad_norm": 0.46630704402923584,
      "learning_rate": 8.648962987221837e-06,
      "loss": 0.526,
      "step": 110
    },
    {
      "epoch": 0.6376256582096697,
      "grad_norm": 0.38065338134765625,
      "learning_rate": 8.614469726379833e-06,
      "loss": 0.506,
      "step": 111
    },
    {
      "epoch": 0.6433700335088559,
      "grad_norm": 0.4356742799282074,
      "learning_rate": 8.579612339264867e-06,
      "loss": 0.5145,
      "step": 112
    },
    {
      "epoch": 0.6491144088080422,
      "grad_norm": 0.4542970061302185,
      "learning_rate": 8.544394337454409e-06,
      "loss": 0.4811,
      "step": 113
    },
    {
      "epoch": 0.6548587841072283,
      "grad_norm": 0.42011526226997375,
      "learning_rate": 8.508819268854713e-06,
      "loss": 0.5043,
      "step": 114
    },
    {
      "epoch": 0.6606031594064146,
      "grad_norm": 0.40397197008132935,
      "learning_rate": 8.472890717343391e-06,
      "loss": 0.5117,
      "step": 115
    },
    {
      "epoch": 0.6663475347056008,
      "grad_norm": 0.4361250698566437,
      "learning_rate": 8.436612302408376e-06,
      "loss": 0.4786,
      "step": 116
    },
    {
      "epoch": 0.672091910004787,
      "grad_norm": 0.4279170334339142,
      "learning_rate": 8.399987678783285e-06,
      "loss": 0.4954,
      "step": 117
    },
    {
      "epoch": 0.6778362853039732,
      "grad_norm": 0.37261563539505005,
      "learning_rate": 8.36302053607924e-06,
      "loss": 0.5162,
      "step": 118
    },
    {
      "epoch": 0.6835806606031594,
      "grad_norm": 0.4131385385990143,
      "learning_rate": 8.325714598413169e-06,
      "loss": 0.5187,
      "step": 119
    },
    {
      "epoch": 0.6893250359023456,
      "grad_norm": 0.38425323367118835,
      "learning_rate": 8.288073624032634e-06,
      "loss": 0.4855,
      "step": 120
    },
    {
      "epoch": 0.6950694112015319,
      "grad_norm": 0.3567887544631958,
      "learning_rate": 8.250101404937223e-06,
      "loss": 0.5031,
      "step": 121
    },
    {
      "epoch": 0.700813786500718,
      "grad_norm": 0.37229597568511963,
      "learning_rate": 8.211801766496537e-06,
      "loss": 0.4834,
      "step": 122
    },
    {
      "epoch": 0.7065581617999043,
      "grad_norm": 0.41608041524887085,
      "learning_rate": 8.17317856706482e-06,
      "loss": 0.4751,
      "step": 123
    },
    {
      "epoch": 0.7123025370990905,
      "grad_norm": 0.4281541705131531,
      "learning_rate": 8.13423569759226e-06,
      "loss": 0.5151,
      "step": 124
    },
    {
      "epoch": 0.7180469123982767,
      "grad_norm": 0.3908356726169586,
      "learning_rate": 8.094977081233006e-06,
      "loss": 0.5163,
      "step": 125
    },
    {
      "epoch": 0.7237912876974629,
      "grad_norm": 0.4016852378845215,
      "learning_rate": 8.055406672949957e-06,
      "loss": 0.4794,
      "step": 126
    },
    {
      "epoch": 0.7295356629966491,
      "grad_norm": 0.47066590189933777,
      "learning_rate": 8.015528459116321e-06,
      "loss": 0.4807,
      "step": 127
    },
    {
      "epoch": 0.7352800382958353,
      "grad_norm": 0.3807947039604187,
      "learning_rate": 7.975346457114034e-06,
      "loss": 0.4904,
      "step": 128
    },
    {
      "epoch": 0.7410244135950216,
      "grad_norm": 0.3645278513431549,
      "learning_rate": 7.934864714929036e-06,
      "loss": 0.4981,
      "step": 129
    },
    {
      "epoch": 0.7467687888942077,
      "grad_norm": 0.3861446678638458,
      "learning_rate": 7.894087310743468e-06,
      "loss": 0.5123,
      "step": 130
    },
    {
      "epoch": 0.752513164193394,
      "grad_norm": 0.34721535444259644,
      "learning_rate": 7.853018352524845e-06,
      "loss": 0.4778,
      "step": 131
    },
    {
      "epoch": 0.7582575394925802,
      "grad_norm": 0.3588525950908661,
      "learning_rate": 7.811661977612202e-06,
      "loss": 0.4828,
      "step": 132
    },
    {
      "epoch": 0.7640019147917664,
      "grad_norm": 0.38026273250579834,
      "learning_rate": 7.770022352299294e-06,
      "loss": 0.4768,
      "step": 133
    },
    {
      "epoch": 0.7697462900909526,
      "grad_norm": 0.4159945845603943,
      "learning_rate": 7.728103671414889e-06,
      "loss": 0.4676,
      "step": 134
    },
    {
      "epoch": 0.7754906653901388,
      "grad_norm": 0.41709959506988525,
      "learning_rate": 7.685910157900158e-06,
      "loss": 0.4854,
      "step": 135
    },
    {
      "epoch": 0.781235040689325,
      "grad_norm": 0.3677918612957001,
      "learning_rate": 7.643446062383273e-06,
      "loss": 0.4966,
      "step": 136
    },
    {
      "epoch": 0.7869794159885113,
      "grad_norm": 0.42224499583244324,
      "learning_rate": 7.600715662751166e-06,
      "loss": 0.4616,
      "step": 137
    },
    {
      "epoch": 0.7927237912876974,
      "grad_norm": 0.4179019331932068,
      "learning_rate": 7.557723263718596e-06,
      "loss": 0.5021,
      "step": 138
    },
    {
      "epoch": 0.7984681665868837,
      "grad_norm": 0.3866509795188904,
      "learning_rate": 7.514473196394467e-06,
      "loss": 0.5084,
      "step": 139
    },
    {
      "epoch": 0.8042125418860699,
      "grad_norm": 0.3728427290916443,
      "learning_rate": 7.470969817845518e-06,
      "loss": 0.5158,
      "step": 140
    },
    {
      "epoch": 0.8099569171852561,
      "grad_norm": 0.39587387442588806,
      "learning_rate": 7.427217510657383e-06,
      "loss": 0.4784,
      "step": 141
    },
    {
      "epoch": 0.8157012924844423,
      "grad_norm": 0.35877299308776855,
      "learning_rate": 7.383220682493081e-06,
      "loss": 0.5097,
      "step": 142
    },
    {
      "epoch": 0.8214456677836285,
      "grad_norm": 0.39193376898765564,
      "learning_rate": 7.338983765648985e-06,
      "loss": 0.471,
      "step": 143
    },
    {
      "epoch": 0.8271900430828147,
      "grad_norm": 0.3881797790527344,
      "learning_rate": 7.294511216608308e-06,
      "loss": 0.506,
      "step": 144
    },
    {
      "epoch": 0.832934418382001,
      "grad_norm": 0.36736828088760376,
      "learning_rate": 7.249807515592149e-06,
      "loss": 0.4894,
      "step": 145
    },
    {
      "epoch": 0.8386787936811871,
      "grad_norm": 0.3713340759277344,
      "learning_rate": 7.2048771661081515e-06,
      "loss": 0.4916,
      "step": 146
    },
    {
      "epoch": 0.8444231689803734,
      "grad_norm": 0.35520419478416443,
      "learning_rate": 7.159724694496815e-06,
      "loss": 0.4639,
      "step": 147
    },
    {
      "epoch": 0.8501675442795597,
      "grad_norm": 0.36185652017593384,
      "learning_rate": 7.114354649475499e-06,
      "loss": 0.4894,
      "step": 148
    },
    {
      "epoch": 0.8559119195787458,
      "grad_norm": 0.3789087235927582,
      "learning_rate": 7.068771601680191e-06,
      "loss": 0.4705,
      "step": 149
    },
    {
      "epoch": 0.861656294877932,
      "grad_norm": 0.36992397904396057,
      "learning_rate": 7.022980143205046e-06,
      "loss": 0.4934,
      "step": 150
    },
    {
      "epoch": 0.8674006701771182,
      "grad_norm": 0.3392237722873688,
      "learning_rate": 6.976984887139775e-06,
      "loss": 0.5065,
      "step": 151
    },
    {
      "epoch": 0.8731450454763044,
      "grad_norm": 0.3429148495197296,
      "learning_rate": 6.930790467104916e-06,
      "loss": 0.4436,
      "step": 152
    },
    {
      "epoch": 0.8788894207754907,
      "grad_norm": 0.3667795658111572,
      "learning_rate": 6.884401536785045e-06,
      "loss": 0.5015,
      "step": 153
    },
    {
      "epoch": 0.8846337960746768,
      "grad_norm": 0.36937567591667175,
      "learning_rate": 6.837822769459942e-06,
      "loss": 0.4662,
      "step": 154
    },
    {
      "epoch": 0.8903781713738631,
      "grad_norm": 0.406288743019104,
      "learning_rate": 6.791058857533814e-06,
      "loss": 0.4771,
      "step": 155
    },
    {
      "epoch": 0.8961225466730494,
      "grad_norm": 0.3828718960285187,
      "learning_rate": 6.744114512062571e-06,
      "loss": 0.4932,
      "step": 156
    },
    {
      "epoch": 0.9018669219722355,
      "grad_norm": 0.3526720106601715,
      "learning_rate": 6.696994462279223e-06,
      "loss": 0.4797,
      "step": 157
    },
    {
      "epoch": 0.9076112972714218,
      "grad_norm": 0.41513240337371826,
      "learning_rate": 6.6497034551174585e-06,
      "loss": 0.5064,
      "step": 158
    },
    {
      "epoch": 0.9133556725706079,
      "grad_norm": 0.3304041624069214,
      "learning_rate": 6.602246254733431e-06,
      "loss": 0.4729,
      "step": 159
    },
    {
      "epoch": 0.9191000478697942,
      "grad_norm": 0.34794774651527405,
      "learning_rate": 6.554627642025807e-06,
      "loss": 0.4799,
      "step": 160
    },
    {
      "epoch": 0.9248444231689804,
      "grad_norm": 0.3707845211029053,
      "learning_rate": 6.506852414154138e-06,
      "loss": 0.4972,
      "step": 161
    },
    {
      "epoch": 0.9305887984681666,
      "grad_norm": 0.3164322078227997,
      "learning_rate": 6.4589253840555856e-06,
      "loss": 0.4385,
      "step": 162
    },
    {
      "epoch": 0.9363331737673528,
      "grad_norm": 0.3256944417953491,
      "learning_rate": 6.41085137996006e-06,
      "loss": 0.4887,
      "step": 163
    },
    {
      "epoch": 0.9420775490665391,
      "grad_norm": 0.38169291615486145,
      "learning_rate": 6.362635244903818e-06,
      "loss": 0.5053,
      "step": 164
    },
    {
      "epoch": 0.9478219243657252,
      "grad_norm": 0.3810901939868927,
      "learning_rate": 6.314281836241573e-06,
      "loss": 0.503,
      "step": 165
    },
    {
      "epoch": 0.9535662996649115,
      "grad_norm": 0.36485424637794495,
      "learning_rate": 6.265796025157154e-06,
      "loss": 0.4573,
      "step": 166
    },
    {
      "epoch": 0.9593106749640976,
      "grad_norm": 0.356774240732193,
      "learning_rate": 6.217182696172776e-06,
      "loss": 0.4754,
      "step": 167
    },
    {
      "epoch": 0.9650550502632839,
      "grad_norm": 0.37172752618789673,
      "learning_rate": 6.168446746656973e-06,
      "loss": 0.4712,
      "step": 168
    },
    {
      "epoch": 0.9707994255624701,
      "grad_norm": 0.35497379302978516,
      "learning_rate": 6.119593086331225e-06,
      "loss": 0.4849,
      "step": 169
    },
    {
      "epoch": 0.9765438008616563,
      "grad_norm": 0.3691353499889374,
      "learning_rate": 6.070626636775349e-06,
      "loss": 0.4798,
      "step": 170
    },
    {
      "epoch": 0.9822881761608425,
      "grad_norm": 0.3360215723514557,
      "learning_rate": 6.021552330931693e-06,
      "loss": 0.5087,
      "step": 171
    },
    {
      "epoch": 0.9880325514600288,
      "grad_norm": 0.3349422812461853,
      "learning_rate": 5.972375112608182e-06,
      "loss": 0.4786,
      "step": 172
    },
    {
      "epoch": 0.9937769267592149,
      "grad_norm": 0.35586225986480713,
      "learning_rate": 5.923099935980278e-06,
      "loss": 0.5128,
      "step": 173
    },
    {
      "epoch": 0.9995213020584012,
      "grad_norm": 0.39223992824554443,
      "learning_rate": 5.8737317650918905e-06,
      "loss": 0.4976,
      "step": 174
    },
    {
      "epoch": 1.0052656773575874,
      "grad_norm": 1.1646690368652344,
      "learning_rate": 5.824275573355278e-06,
      "loss": 0.8332,
      "step": 175
    },
    {
      "epoch": 1.0110100526567736,
      "grad_norm": 0.4050624966621399,
      "learning_rate": 5.7747363430500395e-06,
      "loss": 0.4527,
      "step": 176
    },
    {
      "epoch": 1.0167544279559597,
      "grad_norm": 0.42426204681396484,
      "learning_rate": 5.725119064821185e-06,
      "loss": 0.4773,
      "step": 177
    },
    {
      "epoch": 1.022498803255146,
      "grad_norm": 0.3911304771900177,
      "learning_rate": 5.675428737176367e-06,
      "loss": 0.4203,
      "step": 178
    },
    {
      "epoch": 1.0282431785543322,
      "grad_norm": 0.40399208664894104,
      "learning_rate": 5.625670365982332e-06,
      "loss": 0.4058,
      "step": 179
    },
    {
      "epoch": 1.0339875538535184,
      "grad_norm": 0.4403814673423767,
      "learning_rate": 5.575848963960621e-06,
      "loss": 0.4834,
      "step": 180
    },
    {
      "epoch": 1.0397319291527047,
      "grad_norm": 0.38714340329170227,
      "learning_rate": 5.525969550182577e-06,
      "loss": 0.4371,
      "step": 181
    },
    {
      "epoch": 1.0454763044518909,
      "grad_norm": 0.418676495552063,
      "learning_rate": 5.4760371495637256e-06,
      "loss": 0.4361,
      "step": 182
    },
    {
      "epoch": 1.051220679751077,
      "grad_norm": 0.38185063004493713,
      "learning_rate": 5.426056792357552e-06,
      "loss": 0.4459,
      "step": 183
    },
    {
      "epoch": 1.0569650550502634,
      "grad_norm": 0.4303770661354065,
      "learning_rate": 5.376033513648743e-06,
      "loss": 0.437,
      "step": 184
    },
    {
      "epoch": 1.0627094303494495,
      "grad_norm": 0.40473535656929016,
      "learning_rate": 5.325972352845965e-06,
      "loss": 0.4539,
      "step": 185
    },
    {
      "epoch": 1.0684538056486357,
      "grad_norm": 0.38757988810539246,
      "learning_rate": 5.2758783531741655e-06,
      "loss": 0.4609,
      "step": 186
    },
    {
      "epoch": 1.0741981809478218,
      "grad_norm": 1.1067591905593872,
      "learning_rate": 5.225756561166521e-06,
      "loss": 0.4605,
      "step": 187
    },
    {
      "epoch": 1.0799425562470082,
      "grad_norm": 0.3715679943561554,
      "learning_rate": 5.175612026156045e-06,
      "loss": 0.4584,
      "step": 188
    },
    {
      "epoch": 1.0856869315461943,
      "grad_norm": 0.3666544258594513,
      "learning_rate": 5.125449799766916e-06,
      "loss": 0.425,
      "step": 189
    },
    {
      "epoch": 1.0914313068453805,
      "grad_norm": 0.3626473844051361,
      "learning_rate": 5.075274935405554e-06,
      "loss": 0.4471,
      "step": 190
    },
    {
      "epoch": 1.0971756821445668,
      "grad_norm": 0.3830001652240753,
      "learning_rate": 5.025092487751552e-06,
      "loss": 0.4522,
      "step": 191
    },
    {
      "epoch": 1.102920057443753,
      "grad_norm": 0.36477676033973694,
      "learning_rate": 4.974907512248451e-06,
      "loss": 0.4169,
      "step": 192
    },
    {
      "epoch": 1.1086644327429391,
      "grad_norm": 0.3420591950416565,
      "learning_rate": 4.924725064594448e-06,
      "loss": 0.4462,
      "step": 193
    },
    {
      "epoch": 1.1144088080421255,
      "grad_norm": 0.35189467668533325,
      "learning_rate": 4.874550200233085e-06,
      "loss": 0.4478,
      "step": 194
    },
    {
      "epoch": 1.1201531833413116,
      "grad_norm": 0.38873544335365295,
      "learning_rate": 4.824387973843957e-06,
      "loss": 0.4704,
      "step": 195
    },
    {
      "epoch": 1.1258975586404978,
      "grad_norm": 0.36383941769599915,
      "learning_rate": 4.7742434388334815e-06,
      "loss": 0.4037,
      "step": 196
    },
    {
      "epoch": 1.1316419339396842,
      "grad_norm": 0.35195353627204895,
      "learning_rate": 4.724121646825838e-06,
      "loss": 0.4587,
      "step": 197
    },
    {
      "epoch": 1.1373863092388703,
      "grad_norm": 0.32282909750938416,
      "learning_rate": 4.674027647154037e-06,
      "loss": 0.4294,
      "step": 198
    },
    {
      "epoch": 1.1431306845380564,
      "grad_norm": 0.3494592010974884,
      "learning_rate": 4.623966486351257e-06,
      "loss": 0.4079,
      "step": 199
    },
    {
      "epoch": 1.1488750598372426,
      "grad_norm": 0.358534574508667,
      "learning_rate": 4.573943207642452e-06,
      "loss": 0.4249,
      "step": 200
    },
    {
      "epoch": 1.154619435136429,
      "grad_norm": 0.4336000978946686,
      "learning_rate": 4.523962850436276e-06,
      "loss": 0.4531,
      "step": 201
    },
    {
      "epoch": 1.160363810435615,
      "grad_norm": 0.40736353397369385,
      "learning_rate": 4.474030449817423e-06,
      "loss": 0.4636,
      "step": 202
    },
    {
      "epoch": 1.1661081857348012,
      "grad_norm": 0.35743412375450134,
      "learning_rate": 4.424151036039381e-06,
      "loss": 0.4292,
      "step": 203
    },
    {
      "epoch": 1.1718525610339876,
      "grad_norm": 0.3722769618034363,
      "learning_rate": 4.3743296340176694e-06,
      "loss": 0.4614,
      "step": 204
    },
    {
      "epoch": 1.1775969363331737,
      "grad_norm": 0.35476091504096985,
      "learning_rate": 4.3245712628236356e-06,
      "loss": 0.4751,
      "step": 205
    },
    {
      "epoch": 1.18334131163236,
      "grad_norm": 0.3447257876396179,
      "learning_rate": 4.274880935178817e-06,
      "loss": 0.4254,
      "step": 206
    },
    {
      "epoch": 1.1890856869315463,
      "grad_norm": 0.36945194005966187,
      "learning_rate": 4.225263656949961e-06,
      "loss": 0.4289,
      "step": 207
    },
    {
      "epoch": 1.1948300622307324,
      "grad_norm": 0.32727038860321045,
      "learning_rate": 4.175724426644724e-06,
      "loss": 0.4423,
      "step": 208
    },
    {
      "epoch": 1.2005744375299185,
      "grad_norm": 0.33243992924690247,
      "learning_rate": 4.12626823490811e-06,
      "loss": 0.4133,
      "step": 209
    },
    {
      "epoch": 1.206318812829105,
      "grad_norm": 0.3578135669231415,
      "learning_rate": 4.076900064019721e-06,
      "loss": 0.4671,
      "step": 210
    },
    {
      "epoch": 1.212063188128291,
      "grad_norm": 0.33452117443084717,
      "learning_rate": 4.02762488739182e-06,
      "loss": 0.4516,
      "step": 211
    },
    {
      "epoch": 1.2178075634274772,
      "grad_norm": 0.35316699743270874,
      "learning_rate": 3.978447669068309e-06,
      "loss": 0.4493,
      "step": 212
    },
    {
      "epoch": 1.2235519387266636,
      "grad_norm": 0.4113364815711975,
      "learning_rate": 3.929373363224654e-06,
      "loss": 0.4208,
      "step": 213
    },
    {
      "epoch": 1.2292963140258497,
      "grad_norm": 0.3599206805229187,
      "learning_rate": 3.8804069136687775e-06,
      "loss": 0.4087,
      "step": 214
    },
    {
      "epoch": 1.2350406893250359,
      "grad_norm": 0.3093276619911194,
      "learning_rate": 3.8315532533430285e-06,
      "loss": 0.4501,
      "step": 215
    },
    {
      "epoch": 1.2407850646242222,
      "grad_norm": 0.32809051871299744,
      "learning_rate": 3.7828173038272266e-06,
      "loss": 0.4572,
      "step": 216
    },
    {
      "epoch": 1.2465294399234084,
      "grad_norm": 0.3610207140445709,
      "learning_rate": 3.7342039748428473e-06,
      "loss": 0.4081,
      "step": 217
    },
    {
      "epoch": 1.2522738152225945,
      "grad_norm": 0.3821544349193573,
      "learning_rate": 3.685718163758427e-06,
      "loss": 0.4373,
      "step": 218
    },
    {
      "epoch": 1.2580181905217809,
      "grad_norm": 0.2960876226425171,
      "learning_rate": 3.6373647550961834e-06,
      "loss": 0.4421,
      "step": 219
    },
    {
      "epoch": 1.263762565820967,
      "grad_norm": 0.34114575386047363,
      "learning_rate": 3.5891486200399413e-06,
      "loss": 0.4779,
      "step": 220
    },
    {
      "epoch": 1.2695069411201532,
      "grad_norm": 0.356026291847229,
      "learning_rate": 3.5410746159444165e-06,
      "loss": 0.4602,
      "step": 221
    },
    {
      "epoch": 1.2752513164193395,
      "grad_norm": 0.3717498183250427,
      "learning_rate": 3.4931475858458634e-06,
      "loss": 0.4321,
      "step": 222
    },
    {
      "epoch": 1.2809956917185257,
      "grad_norm": 0.36155185103416443,
      "learning_rate": 3.445372357974194e-06,
      "loss": 0.4746,
      "step": 223
    },
    {
      "epoch": 1.2867400670177118,
      "grad_norm": 0.3382367193698883,
      "learning_rate": 3.397753745266571e-06,
      "loss": 0.4297,
      "step": 224
    },
    {
      "epoch": 1.292484442316898,
      "grad_norm": 0.32759156823158264,
      "learning_rate": 3.350296544882543e-06,
      "loss": 0.4195,
      "step": 225
    },
    {
      "epoch": 1.2982288176160843,
      "grad_norm": 0.31985870003700256,
      "learning_rate": 3.303005537720778e-06,
      "loss": 0.4246,
      "step": 226
    },
    {
      "epoch": 1.3039731929152705,
      "grad_norm": 0.32732826471328735,
      "learning_rate": 3.255885487937431e-06,
      "loss": 0.4628,
      "step": 227
    },
    {
      "epoch": 1.3097175682144566,
      "grad_norm": 0.3491455316543579,
      "learning_rate": 3.2089411424661864e-06,
      "loss": 0.4993,
      "step": 228
    },
    {
      "epoch": 1.3154619435136428,
      "grad_norm": 0.3540627956390381,
      "learning_rate": 3.1621772305400603e-06,
      "loss": 0.4525,
      "step": 229
    },
    {
      "epoch": 1.3212063188128291,
      "grad_norm": 0.30988261103630066,
      "learning_rate": 3.1155984632149565e-06,
      "loss": 0.4405,
      "step": 230
    },
    {
      "epoch": 1.3269506941120153,
      "grad_norm": 0.31742069125175476,
      "learning_rate": 3.0692095328950843e-06,
      "loss": 0.3974,
      "step": 231
    },
    {
      "epoch": 1.3326950694112014,
      "grad_norm": 0.32312625646591187,
      "learning_rate": 3.023015112860228e-06,
      "loss": 0.4913,
      "step": 232
    },
    {
      "epoch": 1.3384394447103878,
      "grad_norm": 0.30435702204704285,
      "learning_rate": 2.977019856794955e-06,
      "loss": 0.4387,
      "step": 233
    },
    {
      "epoch": 1.344183820009574,
      "grad_norm": 0.30270928144454956,
      "learning_rate": 2.93122839831981e-06,
      "loss": 0.4523,
      "step": 234
    },
    {
      "epoch": 1.34992819530876,
      "grad_norm": 0.3011631965637207,
      "learning_rate": 2.8856453505245018e-06,
      "loss": 0.4306,
      "step": 235
    },
    {
      "epoch": 1.3556725706079464,
      "grad_norm": 0.3339727520942688,
      "learning_rate": 2.840275305503186e-06,
      "loss": 0.488,
      "step": 236
    },
    {
      "epoch": 1.3614169459071326,
      "grad_norm": 0.2971702516078949,
      "learning_rate": 2.7951228338918506e-06,
      "loss": 0.4138,
      "step": 237
    },
    {
      "epoch": 1.3671613212063187,
      "grad_norm": 0.3056834042072296,
      "learning_rate": 2.7501924844078538e-06,
      "loss": 0.4418,
      "step": 238
    },
    {
      "epoch": 1.372905696505505,
      "grad_norm": 0.4925230145454407,
      "learning_rate": 2.7054887833916933e-06,
      "loss": 0.4637,
      "step": 239
    },
    {
      "epoch": 1.3786500718046912,
      "grad_norm": 0.30578091740608215,
      "learning_rate": 2.6610162343510183e-06,
      "loss": 0.4602,
      "step": 240
    },
    {
      "epoch": 1.3843944471038774,
      "grad_norm": 0.2952449321746826,
      "learning_rate": 2.616779317506921e-06,
      "loss": 0.4247,
      "step": 241
    },
    {
      "epoch": 1.3901388224030637,
      "grad_norm": 0.2833816409111023,
      "learning_rate": 2.572782489342617e-06,
      "loss": 0.3988,
      "step": 242
    },
    {
      "epoch": 1.39588319770225,
      "grad_norm": 0.3349515199661255,
      "learning_rate": 2.5290301821544826e-06,
      "loss": 0.4604,
      "step": 243
    },
    {
      "epoch": 1.401627573001436,
      "grad_norm": 0.2861647307872772,
      "learning_rate": 2.4855268036055346e-06,
      "loss": 0.4553,
      "step": 244
    },
    {
      "epoch": 1.4073719483006224,
      "grad_norm": 0.30491623282432556,
      "learning_rate": 2.4422767362814045e-06,
      "loss": 0.4616,
      "step": 245
    },
    {
      "epoch": 1.4131163235998085,
      "grad_norm": 0.29965779185295105,
      "learning_rate": 2.3992843372488357e-06,
      "loss": 0.4335,
      "step": 246
    },
    {
      "epoch": 1.4188606988989947,
      "grad_norm": 0.30574852228164673,
      "learning_rate": 2.3565539376167295e-06,
      "loss": 0.4634,
      "step": 247
    },
    {
      "epoch": 1.424605074198181,
      "grad_norm": 0.30110111832618713,
      "learning_rate": 2.3140898420998425e-06,
      "loss": 0.4267,
      "step": 248
    },
    {
      "epoch": 1.4303494494973672,
      "grad_norm": 0.3108730614185333,
      "learning_rate": 2.271896328585114e-06,
      "loss": 0.4417,
      "step": 249
    },
    {
      "epoch": 1.4360938247965533,
      "grad_norm": 0.3105560541152954,
      "learning_rate": 2.2299776477007073e-06,
      "loss": 0.4324,
      "step": 250
    },
    {
      "epoch": 1.4418382000957397,
      "grad_norm": 0.3196023106575012,
      "learning_rate": 2.1883380223878004e-06,
      "loss": 0.4613,
      "step": 251
    },
    {
      "epoch": 1.4475825753949259,
      "grad_norm": 0.32986778020858765,
      "learning_rate": 2.1469816474751566e-06,
      "loss": 0.4786,
      "step": 252
    },
    {
      "epoch": 1.453326950694112,
      "grad_norm": 0.3233254551887512,
      "learning_rate": 2.105912689256533e-06,
      "loss": 0.4174,
      "step": 253
    },
    {
      "epoch": 1.4590713259932984,
      "grad_norm": 0.32636070251464844,
      "learning_rate": 2.0651352850709656e-06,
      "loss": 0.4419,
      "step": 254
    },
    {
      "epoch": 1.4648157012924845,
      "grad_norm": 0.477860689163208,
      "learning_rate": 2.0246535428859652e-06,
      "loss": 0.4522,
      "step": 255
    },
    {
      "epoch": 1.4705600765916707,
      "grad_norm": 0.2859998643398285,
      "learning_rate": 1.984471540883679e-06,
      "loss": 0.4436,
      "step": 256
    },
    {
      "epoch": 1.4763044518908568,
      "grad_norm": 0.3192572295665741,
      "learning_rate": 1.9445933270500444e-06,
      "loss": 0.4702,
      "step": 257
    },
    {
      "epoch": 1.4820488271900432,
      "grad_norm": 0.29136887192726135,
      "learning_rate": 1.905022918766995e-06,
      "loss": 0.4227,
      "step": 258
    },
    {
      "epoch": 1.4877932024892293,
      "grad_norm": 0.3080131709575653,
      "learning_rate": 1.8657643024077431e-06,
      "loss": 0.4756,
      "step": 259
    },
    {
      "epoch": 1.4935375777884154,
      "grad_norm": 0.3019583523273468,
      "learning_rate": 1.8268214329351797e-06,
      "loss": 0.4422,
      "step": 260
    },
    {
      "epoch": 1.4992819530876016,
      "grad_norm": 0.27546194195747375,
      "learning_rate": 1.7881982335034625e-06,
      "loss": 0.3934,
      "step": 261
    },
    {
      "epoch": 1.505026328386788,
      "grad_norm": 0.2833578288555145,
      "learning_rate": 1.7498985950627794e-06,
      "loss": 0.4139,
      "step": 262
    },
    {
      "epoch": 1.510770703685974,
      "grad_norm": 0.3095320761203766,
      "learning_rate": 1.7119263759673677e-06,
      "loss": 0.4288,
      "step": 263
    },
    {
      "epoch": 1.5165150789851602,
      "grad_norm": 0.3324244022369385,
      "learning_rate": 1.6742854015868349e-06,
      "loss": 0.468,
      "step": 264
    },
    {
      "epoch": 1.5222594542843466,
      "grad_norm": 0.5882402658462524,
      "learning_rate": 1.6369794639207626e-06,
      "loss": 0.4201,
      "step": 265
    },
    {
      "epoch": 1.5280038295835328,
      "grad_norm": 0.3166416585445404,
      "learning_rate": 1.6000123212167158e-06,
      "loss": 0.4298,
      "step": 266
    },
    {
      "epoch": 1.533748204882719,
      "grad_norm": 0.2847976088523865,
      "learning_rate": 1.5633876975916261e-06,
      "loss": 0.4389,
      "step": 267
    },
    {
      "epoch": 1.5394925801819053,
      "grad_norm": 0.3158165514469147,
      "learning_rate": 1.5271092826566108e-06,
      "loss": 0.4485,
      "step": 268
    },
    {
      "epoch": 1.5452369554810914,
      "grad_norm": 0.307921826839447,
      "learning_rate": 1.4911807311452874e-06,
      "loss": 0.445,
      "step": 269
    },
    {
      "epoch": 1.5509813307802776,
      "grad_norm": 0.3212144672870636,
      "learning_rate": 1.4556056625455922e-06,
      "loss": 0.4302,
      "step": 270
    },
    {
      "epoch": 1.556725706079464,
      "grad_norm": 0.30903419852256775,
      "learning_rate": 1.4203876607351347e-06,
      "loss": 0.4981,
      "step": 271
    },
    {
      "epoch": 1.56247008137865,
      "grad_norm": 0.2960156500339508,
      "learning_rate": 1.3855302736201686e-06,
      "loss": 0.4278,
      "step": 272
    },
    {
      "epoch": 1.5682144566778362,
      "grad_norm": 0.3086850643157959,
      "learning_rate": 1.3510370127781635e-06,
      "loss": 0.4574,
      "step": 273
    },
    {
      "epoch": 1.5739588319770226,
      "grad_norm": 0.3112443685531616,
      "learning_rate": 1.3169113531040462e-06,
      "loss": 0.4648,
      "step": 274
    },
    {
      "epoch": 1.5797032072762087,
      "grad_norm": 0.3102812170982361,
      "learning_rate": 1.2831567324601325e-06,
      "loss": 0.4525,
      "step": 275
    },
    {
      "epoch": 1.5854475825753949,
      "grad_norm": 0.2903468906879425,
      "learning_rate": 1.2497765513297976e-06,
      "loss": 0.4499,
      "step": 276
    },
    {
      "epoch": 1.5911919578745812,
      "grad_norm": 0.31451210379600525,
      "learning_rate": 1.2167741724749026e-06,
      "loss": 0.4458,
      "step": 277
    },
    {
      "epoch": 1.5969363331737674,
      "grad_norm": 0.24717064201831818,
      "learning_rate": 1.1841529205970281e-06,
      "loss": 0.3864,
      "step": 278
    },
    {
      "epoch": 1.6026807084729535,
      "grad_norm": 0.3028728663921356,
      "learning_rate": 1.1519160820025382e-06,
      "loss": 0.4503,
      "step": 279
    },
    {
      "epoch": 1.6084250837721399,
      "grad_norm": 0.2956719398498535,
      "learning_rate": 1.1200669042715163e-06,
      "loss": 0.4482,
      "step": 280
    },
    {
      "epoch": 1.614169459071326,
      "grad_norm": 0.27994483709335327,
      "learning_rate": 1.0886085959305915e-06,
      "loss": 0.4636,
      "step": 281
    },
    {
      "epoch": 1.6199138343705122,
      "grad_norm": 0.32012465596199036,
      "learning_rate": 1.057544326129723e-06,
      "loss": 0.4454,
      "step": 282
    },
    {
      "epoch": 1.6256582096696985,
      "grad_norm": 0.28910741209983826,
      "learning_rate": 1.026877224322923e-06,
      "loss": 0.4586,
      "step": 283
    },
    {
      "epoch": 1.6314025849688847,
      "grad_norm": 0.2875552475452423,
      "learning_rate": 9.966103799529891e-07,
      "loss": 0.4808,
      "step": 284
    },
    {
      "epoch": 1.6371469602680708,
      "grad_norm": 0.29085999727249146,
      "learning_rate": 9.66746842140287e-07,
      "loss": 0.4212,
      "step": 285
    },
    {
      "epoch": 1.6428913355672572,
      "grad_norm": 0.28769615292549133,
      "learning_rate": 9.372896193755621e-07,
      "loss": 0.4386,
      "step": 286
    },
    {
      "epoch": 1.6486357108664431,
      "grad_norm": 0.2984837293624878,
      "learning_rate": 9.082416792168608e-07,
      "loss": 0.4733,
      "step": 287
    },
    {
      "epoch": 1.6543800861656295,
      "grad_norm": 0.2679706811904907,
      "learning_rate": 8.7960594799059e-07,
      "loss": 0.4095,
      "step": 288
    },
    {
      "epoch": 1.6601244614648158,
      "grad_norm": 0.2708052396774292,
      "learning_rate": 8.513853104966951e-07,
      "loss": 0.4599,
      "step": 289
    },
    {
      "epoch": 1.6658688367640018,
      "grad_norm": 0.3161284327507019,
      "learning_rate": 8.235826097180566e-07,
      "loss": 0.4477,
      "step": 290
    },
    {
      "epoch": 1.6716132120631881,
      "grad_norm": 0.2906995415687561,
      "learning_rate": 7.962006465340821e-07,
      "loss": 0.4149,
      "step": 291
    },
    {
      "epoch": 1.6773575873623745,
      "grad_norm": 0.26941245794296265,
      "learning_rate": 7.692421794385313e-07,
      "loss": 0.4313,
      "step": 292
    },
    {
      "epoch": 1.6831019626615604,
      "grad_norm": 0.303433358669281,
      "learning_rate": 7.427099242616348e-07,
      "loss": 0.4704,
      "step": 293
    },
    {
      "epoch": 1.6888463379607468,
      "grad_norm": 0.3116205036640167,
      "learning_rate": 7.166065538964955e-07,
      "loss": 0.4917,
      "step": 294
    },
    {
      "epoch": 1.694590713259933,
      "grad_norm": 0.2791763246059418,
      "learning_rate": 6.909346980298093e-07,
      "loss": 0.4372,
      "step": 295
    },
    {
      "epoch": 1.700335088559119,
      "grad_norm": 0.2720767855644226,
      "learning_rate": 6.656969428769567e-07,
      "loss": 0.4113,
      "step": 296
    },
    {
      "epoch": 1.7060794638583054,
      "grad_norm": 0.28897154331207275,
      "learning_rate": 6.408958309214597e-07,
      "loss": 0.4363,
      "step": 297
    },
    {
      "epoch": 1.7118238391574916,
      "grad_norm": 0.35382387042045593,
      "learning_rate": 6.165338606588517e-07,
      "loss": 0.4666,
      "step": 298
    },
    {
      "epoch": 1.7175682144566777,
      "grad_norm": 0.2790905237197876,
      "learning_rate": 5.926134863449712e-07,
      "loss": 0.4397,
      "step": 299
    },
    {
      "epoch": 1.723312589755864,
      "grad_norm": 0.2667299807071686,
      "learning_rate": 5.691371177487215e-07,
      "loss": 0.4103,
      "step": 300
    },
    {
      "epoch": 1.7290569650550502,
      "grad_norm": 0.31199657917022705,
      "learning_rate": 5.461071199093048e-07,
      "loss": 0.4793,
      "step": 301
    },
    {
      "epoch": 1.7348013403542364,
      "grad_norm": 0.26627758145332336,
      "learning_rate": 5.235258128979676e-07,
      "loss": 0.4194,
      "step": 302
    },
    {
      "epoch": 1.7405457156534228,
      "grad_norm": 0.28980743885040283,
      "learning_rate": 5.0139547158427e-07,
      "loss": 0.4207,
      "step": 303
    },
    {
      "epoch": 1.746290090952609,
      "grad_norm": 0.2944933772087097,
      "learning_rate": 4.797183254069176e-07,
      "loss": 0.4546,
      "step": 304
    },
    {
      "epoch": 1.752034466251795,
      "grad_norm": 0.2938905358314514,
      "learning_rate": 4.5849655814915683e-07,
      "loss": 0.4705,
      "step": 305
    },
    {
      "epoch": 1.7577788415509814,
      "grad_norm": 0.27462682127952576,
      "learning_rate": 4.3773230771879004e-07,
      "loss": 0.4507,
      "step": 306
    },
    {
      "epoch": 1.7635232168501676,
      "grad_norm": 0.2787008583545685,
      "learning_rate": 4.1742766593278974e-07,
      "loss": 0.4194,
      "step": 307
    },
    {
      "epoch": 1.7692675921493537,
      "grad_norm": 0.29002130031585693,
      "learning_rate": 3.9758467830656623e-07,
      "loss": 0.4573,
      "step": 308
    },
    {
      "epoch": 1.77501196744854,
      "grad_norm": 0.28262341022491455,
      "learning_rate": 3.782053438479094e-07,
      "loss": 0.4291,
      "step": 309
    },
    {
      "epoch": 1.7807563427477262,
      "grad_norm": 0.26790839433670044,
      "learning_rate": 3.5929161485559694e-07,
      "loss": 0.4161,
      "step": 310
    },
    {
      "epoch": 1.7865007180469124,
      "grad_norm": 0.32133734226226807,
      "learning_rate": 3.4084539672271764e-07,
      "loss": 0.4695,
      "step": 311
    },
    {
      "epoch": 1.7922450933460987,
      "grad_norm": 0.3312699496746063,
      "learning_rate": 3.228685477447291e-07,
      "loss": 0.4346,
      "step": 312
    },
    {
      "epoch": 1.7979894686452849,
      "grad_norm": 0.3207743167877197,
      "learning_rate": 3.0536287893223603e-07,
      "loss": 0.4238,
      "step": 313
    },
    {
      "epoch": 1.803733843944471,
      "grad_norm": 0.2662029564380646,
      "learning_rate": 2.883301538285582e-07,
      "loss": 0.4106,
      "step": 314
    },
    {
      "epoch": 1.8094782192436574,
      "grad_norm": 0.28297159075737,
      "learning_rate": 2.717720883320685e-07,
      "loss": 0.443,
      "step": 315
    },
    {
      "epoch": 1.8152225945428435,
      "grad_norm": 0.29099133610725403,
      "learning_rate": 2.556903505233216e-07,
      "loss": 0.4834,
      "step": 316
    },
    {
      "epoch": 1.8209669698420297,
      "grad_norm": 0.27968353033065796,
      "learning_rate": 2.4008656049701875e-07,
      "loss": 0.4441,
      "step": 317
    },
    {
      "epoch": 1.826711345141216,
      "grad_norm": 0.2966521680355072,
      "learning_rate": 2.2496229019879635e-07,
      "loss": 0.43,
      "step": 318
    },
    {
      "epoch": 1.832455720440402,
      "grad_norm": 0.31343600153923035,
      "learning_rate": 2.1031906326685946e-07,
      "loss": 0.4286,
      "step": 319
    },
    {
      "epoch": 1.8382000957395883,
      "grad_norm": 0.3268885314464569,
      "learning_rate": 1.9615835487849677e-07,
      "loss": 0.4353,
      "step": 320
    },
    {
      "epoch": 1.8439444710387747,
      "grad_norm": 0.28709664940834045,
      "learning_rate": 1.824815916014644e-07,
      "loss": 0.4607,
      "step": 321
    },
    {
      "epoch": 1.8496888463379606,
      "grad_norm": 0.24914994835853577,
      "learning_rate": 1.6929015125027314e-07,
      "loss": 0.4053,
      "step": 322
    },
    {
      "epoch": 1.855433221637147,
      "grad_norm": 0.3060120642185211,
      "learning_rate": 1.5658536274738623e-07,
      "loss": 0.46,
      "step": 323
    },
    {
      "epoch": 1.8611775969363333,
      "grad_norm": 0.27555692195892334,
      "learning_rate": 1.443685059893396e-07,
      "loss": 0.4623,
      "step": 324
    },
    {
      "epoch": 1.8669219722355193,
      "grad_norm": 0.29192641377449036,
      "learning_rate": 1.3264081171780797e-07,
      "loss": 0.4451,
      "step": 325
    },
    {
      "epoch": 1.8726663475347056,
      "grad_norm": 0.276792049407959,
      "learning_rate": 1.2140346139561277e-07,
      "loss": 0.4271,
      "step": 326
    },
    {
      "epoch": 1.8784107228338918,
      "grad_norm": 0.27528291940689087,
      "learning_rate": 1.1065758708770468e-07,
      "loss": 0.4431,
      "step": 327
    },
    {
      "epoch": 1.884155098133078,
      "grad_norm": 0.2996566891670227,
      "learning_rate": 1.004042713471165e-07,
      "loss": 0.4568,
      "step": 328
    },
    {
      "epoch": 1.8898994734322643,
      "grad_norm": 0.2685716450214386,
      "learning_rate": 9.064454710590253e-08,
      "loss": 0.4199,
      "step": 329
    },
    {
      "epoch": 1.8956438487314504,
      "grad_norm": 0.2753286063671112,
      "learning_rate": 8.137939757108526e-08,
      "loss": 0.4178,
      "step": 330
    },
    {
      "epoch": 1.9013882240306366,
      "grad_norm": 0.34809115529060364,
      "learning_rate": 7.260975612560173e-08,
      "loss": 0.4383,
      "step": 331
    },
    {
      "epoch": 1.907132599329823,
      "grad_norm": 0.28402262926101685,
      "learning_rate": 6.433650623427379e-08,
      "loss": 0.3941,
      "step": 332
    },
    {
      "epoch": 1.912876974629009,
      "grad_norm": 0.30577704310417175,
      "learning_rate": 5.6560481354807625e-08,
      "loss": 0.4199,
      "step": 333
    },
    {
      "epoch": 1.9186213499281952,
      "grad_norm": 0.27827057242393494,
      "learning_rate": 4.928246485383148e-08,
      "loss": 0.461,
      "step": 334
    },
    {
      "epoch": 1.9243657252273816,
      "grad_norm": 0.2772233188152313,
      "learning_rate": 4.250318992797375e-08,
      "loss": 0.4617,
      "step": 335
    },
    {
      "epoch": 1.9301101005265677,
      "grad_norm": 0.2958879768848419,
      "learning_rate": 3.622333953000601e-08,
      "loss": 0.409,
      "step": 336
    },
    {
      "epoch": 1.9358544758257539,
      "grad_norm": 0.28397998213768005,
      "learning_rate": 3.0443546300035764e-08,
      "loss": 0.3996,
      "step": 337
    },
    {
      "epoch": 1.9415988511249402,
      "grad_norm": 0.2538606524467468,
      "learning_rate": 2.516439250177749e-08,
      "loss": 0.4192,
      "step": 338
    },
    {
      "epoch": 1.9473432264241264,
      "grad_norm": 0.28398269414901733,
      "learning_rate": 2.038640996389285e-08,
      "loss": 0.4791,
      "step": 339
    },
    {
      "epoch": 1.9530876017233125,
      "grad_norm": 0.2674563527107239,
      "learning_rate": 1.6110080026414123e-08,
      "loss": 0.4467,
      "step": 340
    },
    {
      "epoch": 1.958831977022499,
      "grad_norm": 0.28281354904174805,
      "learning_rate": 1.2335833492252425e-08,
      "loss": 0.4558,
      "step": 341
    },
    {
      "epoch": 1.964576352321685,
      "grad_norm": 0.27911072969436646,
      "learning_rate": 9.06405058380022e-09,
      "loss": 0.4585,
      "step": 342
    },
    {
      "epoch": 1.9703207276208712,
      "grad_norm": 0.3073008060455322,
      "learning_rate": 6.295060904623618e-09,
      "loss": 0.4179,
      "step": 343
    },
    {
      "epoch": 1.9760651029200575,
      "grad_norm": 0.2823576033115387,
      "learning_rate": 4.02914340626226e-09,
      "loss": 0.4345,
      "step": 344
    },
    {
      "epoch": 1.9818094782192437,
      "grad_norm": 0.261004775762558,
      "learning_rate": 2.2665263601240328e-09,
      "loss": 0.4481,
      "step": 345
    },
    {
      "epoch": 1.9875538535184298,
      "grad_norm": 0.280528724193573,
      "learning_rate": 1.0073873344895735e-09,
      "loss": 0.4627,
      "step": 346
    },
    {
      "epoch": 1.9932982288176162,
      "grad_norm": 0.2972052991390228,
      "learning_rate": 2.5185317662490547e-10,
      "loss": 0.466,
      "step": 347
    },
    {
      "epoch": 1.9990426041168023,
      "grad_norm": 0.26770320534706116,
      "learning_rate": 0.0,
      "loss": 0.3974,
      "step": 348
    },
    {
      "epoch": 1.9990426041168023,
      "step": 348,
      "total_flos": 904766317002752.0,
      "train_loss": 0.4906231573429601,
      "train_runtime": 70045.1505,
      "train_samples_per_second": 0.477,
      "train_steps_per_second": 0.005
    }
  ],
  "logging_steps": 1,
  "max_steps": 348,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 904766317002752.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}