{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.19745502413339183,
  "eval_steps": 50,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013163668275559457,
      "grad_norm": 1.478107202034744,
      "learning_rate": 2e-05,
      "loss": 1.7794,
      "step": 1
    },
    {
      "epoch": 0.0026327336551118913,
      "grad_norm": 1.558583193699013,
      "learning_rate": 1.999777729859618e-05,
      "loss": 1.7452,
      "step": 2
    },
    {
      "epoch": 0.003949100482667837,
      "grad_norm": 10.951408075586814,
      "learning_rate": 1.9991110182465032e-05,
      "loss": 2.0813,
      "step": 3
    },
    {
      "epoch": 0.005265467310223783,
      "grad_norm": 8.94156747645014,
      "learning_rate": 1.9980001615408228e-05,
      "loss": 2.0632,
      "step": 4
    },
    {
      "epoch": 0.006581834137779728,
      "grad_norm": 5.090359070474341,
      "learning_rate": 1.9964456535631287e-05,
      "loss": 2.0003,
      "step": 5
    },
    {
      "epoch": 0.007898200965335674,
      "grad_norm": 5.827864512562978,
      "learning_rate": 1.9944481853548335e-05,
      "loss": 1.9603,
      "step": 6
    },
    {
      "epoch": 0.009214567792891619,
      "grad_norm": 2.928916785679452,
      "learning_rate": 1.9920086448710162e-05,
      "loss": 1.9254,
      "step": 7
    },
    {
      "epoch": 0.010530934620447565,
      "grad_norm": 2.1252222216149548,
      "learning_rate": 1.9891281165856876e-05,
      "loss": 1.8795,
      "step": 8
    },
    {
      "epoch": 0.01184730144800351,
      "grad_norm": 1.4989937851932464,
      "learning_rate": 1.9858078810097004e-05,
      "loss": 1.8942,
      "step": 9
    },
    {
      "epoch": 0.013163668275559455,
      "grad_norm": 1.270682601973982,
      "learning_rate": 1.98204941412151e-05,
      "loss": 1.8871,
      "step": 10
    },
    {
      "epoch": 0.014480035103115402,
      "grad_norm": 1.3529821372631183,
      "learning_rate": 1.9778543867110428e-05,
      "loss": 1.8811,
      "step": 11
    },
    {
      "epoch": 0.015796401930671347,
      "grad_norm": 1.0104967698093212,
      "learning_rate": 1.9732246636369605e-05,
      "loss": 1.8791,
      "step": 12
    },
    {
      "epoch": 0.017112768758227294,
      "grad_norm": 1.0569581062124906,
      "learning_rate": 1.968162302997659e-05,
      "loss": 1.8713,
      "step": 13
    },
    {
      "epoch": 0.018429135585783237,
      "grad_norm": 0.9518481663490642,
      "learning_rate": 1.962669555216358e-05,
      "loss": 1.8586,
      "step": 14
    },
    {
      "epoch": 0.019745502413339184,
      "grad_norm": 0.921529722029732,
      "learning_rate": 1.9567488620406984e-05,
      "loss": 1.8868,
      "step": 15
    },
    {
      "epoch": 0.02106186924089513,
      "grad_norm": 0.8738583301224286,
      "learning_rate": 1.9504028554572865e-05,
      "loss": 1.8658,
      "step": 16
    },
    {
      "epoch": 0.022378236068451074,
      "grad_norm": 0.7541343679010786,
      "learning_rate": 1.943634356521671e-05,
      "loss": 1.8799,
      "step": 17
    },
    {
      "epoch": 0.02369460289600702,
      "grad_norm": 0.7400942117317526,
      "learning_rate": 1.9364463741042694e-05,
      "loss": 1.8423,
      "step": 18
    },
    {
      "epoch": 0.025010969723562967,
      "grad_norm": 0.761449374917477,
      "learning_rate": 1.928842103552803e-05,
      "loss": 1.8514,
      "step": 19
    },
    {
      "epoch": 0.02632733655111891,
      "grad_norm": 0.6637380195533215,
      "learning_rate": 1.920824925271838e-05,
      "loss": 1.8703,
      "step": 20
    },
    {
      "epoch": 0.027643703378674857,
      "grad_norm": 0.6510924341068389,
      "learning_rate": 1.9123984032200586e-05,
      "loss": 1.8477,
      "step": 21
    },
    {
      "epoch": 0.028960070206230804,
      "grad_norm": 0.6888677503090413,
      "learning_rate": 1.9035662833259433e-05,
      "loss": 1.8686,
      "step": 22
    },
    {
      "epoch": 0.030276437033786747,
      "grad_norm": 0.690819563000188,
      "learning_rate": 1.8943324918225495e-05,
      "loss": 1.8675,
      "step": 23
    },
    {
      "epoch": 0.031592803861342694,
      "grad_norm": 0.72706300370007,
      "learning_rate": 1.8847011335021447e-05,
      "loss": 1.8546,
      "step": 24
    },
    {
      "epoch": 0.03290917068889864,
      "grad_norm": 0.6717961924431257,
      "learning_rate": 1.874676489891461e-05,
      "loss": 1.8606,
      "step": 25
    },
    {
      "epoch": 0.03422553751645459,
      "grad_norm": 0.7055835327922833,
      "learning_rate": 1.8642630173483832e-05,
      "loss": 1.8624,
      "step": 26
    },
    {
      "epoch": 0.03554190434401053,
      "grad_norm": 0.6700859587473981,
      "learning_rate": 1.85346534508092e-05,
      "loss": 1.8672,
      "step": 27
    },
    {
      "epoch": 0.036858271171566474,
      "grad_norm": 0.6688469199082278,
      "learning_rate": 1.8422882730893323e-05,
      "loss": 1.887,
      "step": 28
    },
    {
      "epoch": 0.03817463799912242,
      "grad_norm": 0.6408372256848319,
      "learning_rate": 1.8307367700323412e-05,
      "loss": 1.8842,
      "step": 29
    },
    {
      "epoch": 0.03949100482667837,
      "grad_norm": 0.6228315694543716,
      "learning_rate": 1.8188159710183595e-05,
      "loss": 1.8553,
      "step": 30
    },
    {
      "epoch": 0.040807371654234315,
      "grad_norm": 0.5919427948000797,
      "learning_rate": 1.8065311753227272e-05,
      "loss": 1.9004,
      "step": 31
    },
    {
      "epoch": 0.04212373848179026,
      "grad_norm": 0.6456371423655457,
      "learning_rate": 1.7938878440319722e-05,
      "loss": 1.8926,
      "step": 32
    },
    {
      "epoch": 0.0434401053093462,
      "grad_norm": 0.6599868030812718,
      "learning_rate": 1.7808915976161364e-05,
      "loss": 1.8968,
      "step": 33
    },
    {
      "epoch": 0.04475647213690215,
      "grad_norm": 0.60352239199404,
      "learning_rate": 1.7675482134302503e-05,
      "loss": 1.8775,
      "step": 34
    },
    {
      "epoch": 0.046072838964458095,
      "grad_norm": 0.6621376662940576,
      "learning_rate": 1.753863623146066e-05,
      "loss": 1.8829,
      "step": 35
    },
    {
      "epoch": 0.04738920579201404,
      "grad_norm": 0.6664933503179751,
      "learning_rate": 1.7398439101151908e-05,
      "loss": 1.8691,
      "step": 36
    },
    {
      "epoch": 0.04870557261956999,
      "grad_norm": 0.6326062739839543,
      "learning_rate": 1.7254953066647915e-05,
      "loss": 1.8746,
      "step": 37
    },
    {
      "epoch": 0.050021939447125935,
      "grad_norm": 0.585003529057695,
      "learning_rate": 1.710824191327075e-05,
      "loss": 1.9078,
      "step": 38
    },
    {
      "epoch": 0.051338306274681875,
      "grad_norm": 0.6172635235914273,
      "learning_rate": 1.695837086003772e-05,
      "loss": 1.8853,
      "step": 39
    },
    {
      "epoch": 0.05265467310223782,
      "grad_norm": 0.534366800769727,
      "learning_rate": 1.680540653066891e-05,
      "loss": 1.8845,
      "step": 40
    },
    {
      "epoch": 0.05397103992979377,
      "grad_norm": 0.5921909430405027,
      "learning_rate": 1.6649416923970248e-05,
      "loss": 1.8858,
      "step": 41
    },
    {
      "epoch": 0.055287406757349715,
      "grad_norm": 0.5490201103081069,
      "learning_rate": 1.649047138360529e-05,
      "loss": 1.8787,
      "step": 42
    },
    {
      "epoch": 0.05660377358490566,
      "grad_norm": 0.579274311316064,
      "learning_rate": 1.632864056726917e-05,
      "loss": 1.881,
      "step": 43
    },
    {
      "epoch": 0.05792014041246161,
      "grad_norm": 0.6456035281783546,
      "learning_rate": 1.6163996415278423e-05,
      "loss": 1.9076,
      "step": 44
    },
    {
      "epoch": 0.05923650724001755,
      "grad_norm": 0.6288660405918178,
      "learning_rate": 1.5996612118590604e-05,
      "loss": 1.8886,
      "step": 45
    },
    {
      "epoch": 0.060552874067573495,
      "grad_norm": 0.6763112822898929,
      "learning_rate": 1.5826562086267956e-05,
      "loss": 1.908,
      "step": 46
    },
    {
      "epoch": 0.06186924089512944,
      "grad_norm": 0.629744205444853,
      "learning_rate": 1.565392191239959e-05,
      "loss": 1.8962,
      "step": 47
    },
    {
      "epoch": 0.06318560772268539,
      "grad_norm": 0.584412537317479,
      "learning_rate": 1.5478768342496872e-05,
      "loss": 1.8869,
      "step": 48
    },
    {
      "epoch": 0.06450197455024133,
      "grad_norm": 0.6241283168793769,
      "learning_rate": 1.5301179239376936e-05,
      "loss": 1.9027,
      "step": 49
    },
    {
      "epoch": 0.06581834137779728,
      "grad_norm": 0.5471717696859852,
      "learning_rate": 1.512123354854955e-05,
      "loss": 1.8828,
      "step": 50
    },
    {
      "epoch": 0.06713470820535322,
      "grad_norm": 0.6688954621412719,
      "learning_rate": 1.4939011263122635e-05,
      "loss": 1.8834,
      "step": 51
    },
    {
      "epoch": 0.06845107503290918,
      "grad_norm": 0.5154479256599076,
      "learning_rate": 1.4754593388242117e-05,
      "loss": 1.8942,
      "step": 52
    },
    {
      "epoch": 0.06976744186046512,
      "grad_norm": 0.6534613473824693,
      "learning_rate": 1.4568061905081874e-05,
      "loss": 1.8858,
      "step": 53
    },
    {
      "epoch": 0.07108380868802106,
      "grad_norm": 0.532231099783857,
      "learning_rate": 1.4379499734399797e-05,
      "loss": 1.899,
      "step": 54
    },
    {
      "epoch": 0.07240017551557701,
      "grad_norm": 0.6653136818055972,
      "learning_rate": 1.4188990699676186e-05,
      "loss": 1.8971,
      "step": 55
    },
    {
      "epoch": 0.07371654234313295,
      "grad_norm": 0.5652108954315214,
      "learning_rate": 1.3996619489850822e-05,
      "loss": 1.9102,
      "step": 56
    },
    {
      "epoch": 0.0750329091706889,
      "grad_norm": 0.6224834604009487,
      "learning_rate": 1.3802471621675337e-05,
      "loss": 1.8948,
      "step": 57
    },
    {
      "epoch": 0.07634927599824484,
      "grad_norm": 0.5658389165959354,
      "learning_rate": 1.3606633401697557e-05,
      "loss": 1.9175,
      "step": 58
    },
    {
      "epoch": 0.0776656428258008,
      "grad_norm": 0.5134836257617336,
      "learning_rate": 1.340919188789477e-05,
      "loss": 1.8994,
      "step": 59
    },
    {
      "epoch": 0.07898200965335674,
      "grad_norm": 0.506386634297645,
      "learning_rate": 1.3210234850972966e-05,
      "loss": 1.9183,
      "step": 60
    },
    {
      "epoch": 0.08029837648091268,
      "grad_norm": 0.48444908783343443,
      "learning_rate": 1.300985073534919e-05,
      "loss": 1.9181,
      "step": 61
    },
    {
      "epoch": 0.08161474330846863,
      "grad_norm": 0.5196550746383986,
      "learning_rate": 1.280812861983446e-05,
      "loss": 1.8983,
      "step": 62
    },
    {
      "epoch": 0.08293111013602457,
      "grad_norm": 0.4635363716296946,
      "learning_rate": 1.2605158178034656e-05,
      "loss": 1.9044,
      "step": 63
    },
    {
      "epoch": 0.08424747696358052,
      "grad_norm": 0.4786482823257171,
      "learning_rate": 1.2401029638486952e-05,
      "loss": 1.8833,
      "step": 64
    },
    {
      "epoch": 0.08556384379113646,
      "grad_norm": 0.4404477886773041,
      "learning_rate": 1.219583374454963e-05,
      "loss": 1.9216,
      "step": 65
    },
    {
      "epoch": 0.0868802106186924,
      "grad_norm": 0.4437992879500064,
      "learning_rate": 1.1989661714063e-05,
      "loss": 1.9386,
      "step": 66
    },
    {
      "epoch": 0.08819657744624836,
      "grad_norm": 0.4633602350670698,
      "learning_rate": 1.1782605198799371e-05,
      "loss": 1.9275,
      "step": 67
    },
    {
      "epoch": 0.0895129442738043,
      "grad_norm": 0.3967538469956234,
      "learning_rate": 1.157475624372018e-05,
      "loss": 1.9003,
      "step": 68
    },
    {
      "epoch": 0.09082931110136025,
      "grad_norm": 0.464659577838311,
      "learning_rate": 1.1366207246058269e-05,
      "loss": 1.9115,
      "step": 69
    },
    {
      "epoch": 0.09214567792891619,
      "grad_norm": 0.39016704222991877,
      "learning_rate": 1.1157050914243614e-05,
      "loss": 1.9259,
      "step": 70
    },
    {
      "epoch": 0.09346204475647214,
      "grad_norm": 0.3983568023063857,
      "learning_rate": 1.0947380226690686e-05,
      "loss": 1.9219,
      "step": 71
    },
    {
      "epoch": 0.09477841158402808,
      "grad_norm": 0.38747854341553906,
      "learning_rate": 1.0737288390465792e-05,
      "loss": 1.9127,
      "step": 72
    },
    {
      "epoch": 0.09609477841158402,
      "grad_norm": 0.3701968617354901,
      "learning_rate": 1.0526868799852797e-05,
      "loss": 1.9362,
      "step": 73
    },
    {
      "epoch": 0.09741114523913998,
      "grad_norm": 0.4203706440448544,
      "learning_rate": 1.031621499483559e-05,
      "loss": 1.93,
      "step": 74
    },
    {
      "epoch": 0.09872751206669592,
      "grad_norm": 0.35781391523432005,
      "learning_rate": 1.0105420619515798e-05,
      "loss": 1.9225,
      "step": 75
    },
    {
      "epoch": 0.10004387889425187,
      "grad_norm": 0.38223891144084254,
      "learning_rate": 9.894579380484206e-06,
      "loss": 1.9388,
      "step": 76
    },
    {
      "epoch": 0.10136024572180781,
      "grad_norm": 0.3685828520037133,
      "learning_rate": 9.683785005164412e-06,
      "loss": 1.9449,
      "step": 77
    },
    {
      "epoch": 0.10267661254936375,
      "grad_norm": 0.3438190438449472,
      "learning_rate": 9.473131200147205e-06,
      "loss": 1.9518,
      "step": 78
    },
    {
      "epoch": 0.1039929793769197,
      "grad_norm": 0.3618774764416367,
      "learning_rate": 9.262711609534211e-06,
      "loss": 1.9377,
      "step": 79
    },
    {
      "epoch": 0.10530934620447564,
      "grad_norm": 0.34010451365490174,
      "learning_rate": 9.052619773309318e-06,
      "loss": 1.9305,
      "step": 80
    },
    {
      "epoch": 0.1066257130320316,
      "grad_norm": 0.3371936981808577,
      "learning_rate": 8.842949085756389e-06,
      "loss": 1.9265,
      "step": 81
    },
    {
      "epoch": 0.10794207985958754,
      "grad_norm": 0.35146588021592196,
      "learning_rate": 8.633792753941733e-06,
      "loss": 1.9377,
      "step": 82
    },
    {
      "epoch": 0.10925844668714349,
      "grad_norm": 0.30860227116860156,
      "learning_rate": 8.425243756279824e-06,
      "loss": 1.917,
      "step": 83
    },
    {
      "epoch": 0.11057481351469943,
      "grad_norm": 0.3279664481204533,
      "learning_rate": 8.217394801200632e-06,
      "loss": 1.9393,
      "step": 84
    },
    {
      "epoch": 0.11189118034225537,
      "grad_norm": 0.3333378347714312,
      "learning_rate": 8.010338285937006e-06,
      "loss": 1.9283,
      "step": 85
    },
    {
      "epoch": 0.11320754716981132,
      "grad_norm": 0.31747932988386196,
      "learning_rate": 7.804166255450372e-06,
      "loss": 1.9339,
      "step": 86
    },
    {
      "epoch": 0.11452391399736726,
      "grad_norm": 0.30534488710182367,
      "learning_rate": 7.598970361513052e-06,
      "loss": 1.9392,
      "step": 87
    },
    {
      "epoch": 0.11584028082492322,
      "grad_norm": 0.3418635851962162,
      "learning_rate": 7.394841821965345e-06,
      "loss": 1.9184,
      "step": 88
    },
    {
      "epoch": 0.11715664765247916,
      "grad_norm": 0.3097262055356444,
      "learning_rate": 7.191871380165538e-06,
      "loss": 1.9436,
      "step": 89
    },
    {
      "epoch": 0.1184730144800351,
      "grad_norm": 0.31332765384816835,
      "learning_rate": 6.990149264650814e-06,
      "loss": 1.936,
      "step": 90
    },
    {
      "epoch": 0.11978938130759105,
      "grad_norm": 0.32677436495185097,
      "learning_rate": 6.789765149027039e-06,
      "loss": 1.943,
      "step": 91
    },
    {
      "epoch": 0.12110574813514699,
      "grad_norm": 0.2742411067382557,
      "learning_rate": 6.590808112105232e-06,
      "loss": 1.9609,
      "step": 92
    },
    {
      "epoch": 0.12242211496270294,
      "grad_norm": 0.3112675679800478,
      "learning_rate": 6.3933665983024465e-06,
      "loss": 1.9462,
      "step": 93
    },
    {
      "epoch": 0.12373848179025888,
      "grad_norm": 0.29802110054243747,
      "learning_rate": 6.197528378324664e-06,
      "loss": 1.9234,
      "step": 94
    },
    {
      "epoch": 0.12505484861781482,
      "grad_norm": 0.30340538870593875,
      "learning_rate": 6.003380510149179e-06,
      "loss": 1.9529,
      "step": 95
    },
    {
      "epoch": 0.12637121544537078,
      "grad_norm": 0.2945912793083695,
      "learning_rate": 5.8110093003238175e-06,
      "loss": 1.9758,
      "step": 96
    },
    {
      "epoch": 0.12768758227292673,
      "grad_norm": 0.2888698196193206,
      "learning_rate": 5.620500265600206e-06,
      "loss": 1.9491,
      "step": 97
    },
    {
      "epoch": 0.12900394910048266,
      "grad_norm": 0.2729389825714769,
      "learning_rate": 5.431938094918132e-06,
      "loss": 1.9611,
      "step": 98
    },
    {
      "epoch": 0.1303203159280386,
      "grad_norm": 0.27012160265285107,
      "learning_rate": 5.245406611757882e-06,
      "loss": 1.9601,
      "step": 99
    },
    {
      "epoch": 0.13163668275559456,
      "grad_norm": 0.2833286401582455,
      "learning_rate": 5.060988736877366e-06,
      "loss": 1.9423,
      "step": 100
    },
    {
      "epoch": 0.13295304958315052,
      "grad_norm": 0.2689819235259652,
      "learning_rate": 4.878766451450451e-06,
      "loss": 1.9494,
      "step": 101
    },
    {
      "epoch": 0.13426941641070644,
      "grad_norm": 0.27038470986999186,
      "learning_rate": 4.698820760623064e-06,
      "loss": 1.9446,
      "step": 102
    },
    {
      "epoch": 0.1355857832382624,
      "grad_norm": 0.2657874660977457,
      "learning_rate": 4.5212316575031325e-06,
      "loss": 1.9583,
      "step": 103
    },
    {
      "epoch": 0.13690215006581835,
      "grad_norm": 0.2548158490455745,
      "learning_rate": 4.346078087600411e-06,
      "loss": 1.9627,
      "step": 104
    },
    {
      "epoch": 0.13821851689337428,
      "grad_norm": 0.2378563164718946,
      "learning_rate": 4.173437913732048e-06,
      "loss": 1.9606,
      "step": 105
    },
    {
      "epoch": 0.13953488372093023,
      "grad_norm": 0.24106091232247806,
      "learning_rate": 4.003387881409397e-06,
      "loss": 1.9652,
      "step": 106
    },
    {
      "epoch": 0.14085125054848618,
      "grad_norm": 0.2534311868268791,
      "learning_rate": 3.836003584721577e-06,
      "loss": 1.9651,
      "step": 107
    },
    {
      "epoch": 0.1421676173760421,
      "grad_norm": 0.24909027318806107,
      "learning_rate": 3.6713594327308343e-06,
      "loss": 1.9506,
      "step": 108
    },
    {
      "epoch": 0.14348398420359806,
      "grad_norm": 0.22450423497361008,
      "learning_rate": 3.509528616394716e-06,
      "loss": 1.9692,
      "step": 109
    },
    {
      "epoch": 0.14480035103115402,
      "grad_norm": 0.22373142436074253,
      "learning_rate": 3.3505830760297543e-06,
      "loss": 1.9655,
      "step": 110
    },
    {
      "epoch": 0.14611671785870997,
      "grad_norm": 0.24436148687662734,
      "learning_rate": 3.1945934693310897e-06,
      "loss": 1.9725,
      "step": 111
    },
    {
      "epoch": 0.1474330846862659,
      "grad_norm": 0.243717242669971,
      "learning_rate": 3.0416291399622834e-06,
      "loss": 1.9983,
      "step": 112
    },
    {
      "epoch": 0.14874945151382185,
      "grad_norm": 0.22693737224142835,
      "learning_rate": 2.891758086729253e-06,
      "loss": 1.9512,
      "step": 113
    },
    {
      "epoch": 0.1500658183413778,
      "grad_norm": 0.2364560703768739,
      "learning_rate": 2.7450469333520856e-06,
      "loss": 1.9573,
      "step": 114
    },
    {
      "epoch": 0.15138218516893373,
      "grad_norm": 0.2083690455132377,
      "learning_rate": 2.6015608988480956e-06,
      "loss": 1.9623,
      "step": 115
    },
    {
      "epoch": 0.15269855199648968,
      "grad_norm": 0.21338449259187517,
      "learning_rate": 2.4613637685393433e-06,
      "loss": 1.9718,
      "step": 116
    },
    {
      "epoch": 0.15401491882404564,
      "grad_norm": 0.20957608579240045,
      "learning_rate": 2.324517865697501e-06,
      "loss": 1.9461,
      "step": 117
    },
    {
      "epoch": 0.1553312856516016,
      "grad_norm": 0.21403383067725684,
      "learning_rate": 2.19108402383864e-06,
      "loss": 1.9597,
      "step": 118
    },
    {
      "epoch": 0.15664765247915752,
      "grad_norm": 0.20648341327283193,
      "learning_rate": 2.06112155968028e-06,
      "loss": 1.9792,
      "step": 119
    },
    {
      "epoch": 0.15796401930671347,
      "grad_norm": 0.21095655453648707,
      "learning_rate": 1.9346882467727323e-06,
      "loss": 1.9845,
      "step": 120
    },
    {
      "epoch": 0.15928038613426942,
      "grad_norm": 0.21390193181094888,
      "learning_rate": 1.811840289816409e-06,
      "loss": 1.9777,
      "step": 121
    },
    {
      "epoch": 0.16059675296182535,
      "grad_norm": 0.20067871192367465,
      "learning_rate": 1.6926322996765899e-06,
      "loss": 1.9792,
      "step": 122
    },
    {
      "epoch": 0.1619131197893813,
      "grad_norm": 0.20159006852736075,
      "learning_rate": 1.5771172691066793e-06,
      "loss": 1.9833,
      "step": 123
    },
    {
      "epoch": 0.16322948661693726,
      "grad_norm": 0.2049490979828284,
      "learning_rate": 1.4653465491908003e-06,
      "loss": 2.0072,
      "step": 124
    },
    {
      "epoch": 0.1645458534444932,
      "grad_norm": 0.2107306041910293,
      "learning_rate": 1.3573698265161683e-06,
      "loss": 1.963,
      "step": 125
    },
    {
      "epoch": 0.16586222027204914,
      "grad_norm": 0.19904765388836215,
      "learning_rate": 1.2532351010853916e-06,
      "loss": 1.9686,
      "step": 126
    },
    {
      "epoch": 0.1671785870996051,
      "grad_norm": 0.20824192677314146,
      "learning_rate": 1.152988664978556e-06,
      "loss": 1.9766,
      "step": 127
    },
    {
      "epoch": 0.16849495392716105,
      "grad_norm": 0.19829139324045503,
      "learning_rate": 1.0566750817745076e-06,
      "loss": 1.9837,
      "step": 128
    },
    {
      "epoch": 0.16981132075471697,
      "grad_norm": 0.1937896866804481,
      "learning_rate": 9.6433716674057e-07,
      "loss": 1.9762,
      "step": 129
    },
    {
      "epoch": 0.17112768758227292,
      "grad_norm": 0.2068527876867058,
      "learning_rate": 8.760159677994174e-07,
      "loss": 1.9808,
      "step": 130
    },
    {
      "epoch": 0.17244405440982888,
      "grad_norm": 0.20276435770975612,
      "learning_rate": 7.91750747281621e-07,
      "loss": 1.9722,
      "step": 131
    },
    {
      "epoch": 0.1737604212373848,
      "grad_norm": 0.19669742261543458,
      "learning_rate": 7.115789644719728e-07,
      "loss": 1.9932,
      "step": 132
    },
    {
      "epoch": 0.17507678806494076,
      "grad_norm": 0.19798421927281493,
      "learning_rate": 6.355362589573078e-07,
      "loss": 1.9745,
      "step": 133
    },
    {
      "epoch": 0.1763931548924967,
      "grad_norm": 0.2012049264044561,
      "learning_rate": 5.636564347832907e-07,
      "loss": 1.9802,
      "step": 134
    },
    {
      "epoch": 0.17770952172005267,
      "grad_norm": 0.1933586784907182,
      "learning_rate": 4.95971445427137e-07,
      "loss": 1.9815,
      "step": 135
    },
    {
      "epoch": 0.1790258885476086,
      "grad_norm": 0.19431666363166195,
      "learning_rate": 4.3251137959302023e-07,
      "loss": 1.9693,
      "step": 136
    },
    {
      "epoch": 0.18034225537516455,
      "grad_norm": 0.19437733809432164,
      "learning_rate": 3.733044478364234e-07,
      "loss": 1.9654,
      "step": 137
    },
    {
      "epoch": 0.1816586222027205,
      "grad_norm": 0.1902377445930519,
      "learning_rate": 3.1837697002341293e-07,
      "loss": 1.9777,
      "step": 138
    },
    {
      "epoch": 0.18297498903027642,
      "grad_norm": 0.18969169318316387,
      "learning_rate": 2.677533636303964e-07,
      "loss": 1.968,
      "step": 139
    },
    {
      "epoch": 0.18429135585783238,
      "grad_norm": 0.19427605561826736,
      "learning_rate": 2.214561328895748e-07,
      "loss": 1.9727,
      "step": 140
    },
    {
      "epoch": 0.18560772268538833,
      "grad_norm": 0.19843810729832032,
      "learning_rate": 1.7950585878489856e-07,
      "loss": 1.9777,
      "step": 141
    },
    {
      "epoch": 0.18692408951294429,
      "grad_norm": 0.20659615140803522,
      "learning_rate": 1.419211899029971e-07,
      "loss": 1.9693,
      "step": 142
    },
    {
      "epoch": 0.1882404563405002,
      "grad_norm": 0.19224056249887123,
      "learning_rate": 1.0871883414312778e-07,
      "loss": 1.9799,
      "step": 143
    },
    {
      "epoch": 0.18955682316805617,
      "grad_norm": 0.18554864852359765,
      "learning_rate": 7.99135512898408e-08,
      "loss": 1.9718,
      "step": 144
    },
    {
      "epoch": 0.19087318999561212,
      "grad_norm": 0.18946113779129892,
      "learning_rate": 5.55181464516652e-08,
      "loss": 1.9736,
      "step": 145
    },
    {
      "epoch": 0.19218955682316805,
      "grad_norm": 0.19283475250568183,
      "learning_rate": 3.554346436871581e-08,
      "loss": 1.9692,
      "step": 146
    },
    {
      "epoch": 0.193505923650724,
      "grad_norm": 0.19209388419499202,
      "learning_rate": 1.9998384591773945e-08,
      "loss": 1.9778,
      "step": 147
    },
    {
      "epoch": 0.19482229047827995,
      "grad_norm": 0.18916243614563277,
      "learning_rate": 8.889817534969425e-09,
      "loss": 1.9855,
      "step": 148
    },
    {
      "epoch": 0.1961386573058359,
      "grad_norm": 0.19279609784225699,
      "learning_rate": 2.222701403818972e-09,
      "loss": 1.9743,
      "step": 149
    },
    {
      "epoch": 0.19745502413339183,
      "grad_norm": 0.20474807316260502,
      "learning_rate": 0.0,
      "loss": 1.9864,
      "step": 150
    },
    {
      "epoch": 0.19745502413339183,
      "step": 150,
      "total_flos": 349677715193856.0,
      "train_loss": 1.9281927061080932,
      "train_runtime": 5210.2137,
      "train_samples_per_second": 58.04,
      "train_steps_per_second": 0.029
    }
  ],
  "logging_steps": 1,
  "max_steps": 150,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 349677715193856.0,
  "train_batch_size": 42,
  "trial_name": null,
  "trial_params": null
}