{
  "best_metric": 0.8892158269882202,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.13065490772497143,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006532745386248571,
      "grad_norm": 3.071519613265991,
      "learning_rate": 1e-05,
      "loss": 4.8197,
      "step": 1
    },
    {
      "epoch": 0.0006532745386248571,
      "eval_loss": 1.2523456811904907,
      "eval_runtime": 39.7707,
      "eval_samples_per_second": 64.822,
      "eval_steps_per_second": 16.218,
      "step": 1
    },
    {
      "epoch": 0.0013065490772497142,
      "grad_norm": 3.2814316749572754,
      "learning_rate": 2e-05,
      "loss": 4.5823,
      "step": 2
    },
    {
      "epoch": 0.0019598236158745713,
      "grad_norm": 2.486225128173828,
      "learning_rate": 3e-05,
      "loss": 4.6482,
      "step": 3
    },
    {
      "epoch": 0.0026130981544994283,
      "grad_norm": 4.542166709899902,
      "learning_rate": 4e-05,
      "loss": 4.5401,
      "step": 4
    },
    {
      "epoch": 0.0032663726931242854,
      "grad_norm": 5.1017303466796875,
      "learning_rate": 5e-05,
      "loss": 4.5357,
      "step": 5
    },
    {
      "epoch": 0.0039196472317491425,
      "grad_norm": 4.591429233551025,
      "learning_rate": 6e-05,
      "loss": 4.4991,
      "step": 6
    },
    {
      "epoch": 0.004572921770374,
      "grad_norm": 5.83657169342041,
      "learning_rate": 7e-05,
      "loss": 4.5506,
      "step": 7
    },
    {
      "epoch": 0.005226196308998857,
      "grad_norm": 4.699705123901367,
      "learning_rate": 8e-05,
      "loss": 4.502,
      "step": 8
    },
    {
      "epoch": 0.005879470847623714,
      "grad_norm": 4.715086460113525,
      "learning_rate": 9e-05,
      "loss": 4.3762,
      "step": 9
    },
    {
      "epoch": 0.006532745386248571,
      "grad_norm": 8.546138763427734,
      "learning_rate": 0.0001,
      "loss": 4.5987,
      "step": 10
    },
    {
      "epoch": 0.007186019924873428,
      "grad_norm": 5.174031734466553,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.5173,
      "step": 11
    },
    {
      "epoch": 0.007839294463498285,
      "grad_norm": 8.635031700134277,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.6462,
      "step": 12
    },
    {
      "epoch": 0.008492569002123142,
      "grad_norm": 3.8267672061920166,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.3896,
      "step": 13
    },
    {
      "epoch": 0.009145843540748,
      "grad_norm": 10.137066841125488,
      "learning_rate": 9.989068136093873e-05,
      "loss": 4.4271,
      "step": 14
    },
    {
      "epoch": 0.009799118079372856,
      "grad_norm": 6.932178497314453,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.3313,
      "step": 15
    },
    {
      "epoch": 0.010452392617997713,
      "grad_norm": 7.159353733062744,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.4989,
      "step": 16
    },
    {
      "epoch": 0.01110566715662257,
      "grad_norm": 7.863455772399902,
      "learning_rate": 9.966546331768191e-05,
      "loss": 4.4278,
      "step": 17
    },
    {
      "epoch": 0.011758941695247428,
      "grad_norm": 3.303664207458496,
      "learning_rate": 9.956320346634876e-05,
      "loss": 4.3716,
      "step": 18
    },
    {
      "epoch": 0.012412216233872285,
      "grad_norm": 4.627152442932129,
      "learning_rate": 9.944739353007344e-05,
      "loss": 4.3446,
      "step": 19
    },
    {
      "epoch": 0.013065490772497142,
      "grad_norm": 3.165895938873291,
      "learning_rate": 9.931806517013612e-05,
      "loss": 4.336,
      "step": 20
    },
    {
      "epoch": 0.013718765311121999,
      "grad_norm": 3.2391879558563232,
      "learning_rate": 9.917525374361912e-05,
      "loss": 4.3012,
      "step": 21
    },
    {
      "epoch": 0.014372039849746856,
      "grad_norm": 5.369521617889404,
      "learning_rate": 9.901899829374047e-05,
      "loss": 4.2431,
      "step": 22
    },
    {
      "epoch": 0.015025314388371713,
      "grad_norm": 4.120507717132568,
      "learning_rate": 9.884934153917997e-05,
      "loss": 4.2157,
      "step": 23
    },
    {
      "epoch": 0.01567858892699657,
      "grad_norm": 4.48763370513916,
      "learning_rate": 9.86663298624003e-05,
      "loss": 4.3074,
      "step": 24
    },
    {
      "epoch": 0.01633186346562143,
      "grad_norm": 5.116293907165527,
      "learning_rate": 9.847001329696653e-05,
      "loss": 4.3278,
      "step": 25
    },
    {
      "epoch": 0.016985138004246284,
      "grad_norm": 3.4045839309692383,
      "learning_rate": 9.826044551386744e-05,
      "loss": 4.3476,
      "step": 26
    },
    {
      "epoch": 0.017638412542871143,
      "grad_norm": 3.72531795501709,
      "learning_rate": 9.803768380684242e-05,
      "loss": 4.7405,
      "step": 27
    },
    {
      "epoch": 0.018291687081496,
      "grad_norm": 2.6001501083374023,
      "learning_rate": 9.780178907671789e-05,
      "loss": 4.3331,
      "step": 28
    },
    {
      "epoch": 0.018944961620120857,
      "grad_norm": 3.126338481903076,
      "learning_rate": 9.755282581475769e-05,
      "loss": 4.4433,
      "step": 29
    },
    {
      "epoch": 0.019598236158745713,
      "grad_norm": 3.749473810195923,
      "learning_rate": 9.729086208503174e-05,
      "loss": 4.8921,
      "step": 30
    },
    {
      "epoch": 0.02025151069737057,
      "grad_norm": 2.6032521724700928,
      "learning_rate": 9.701596950580806e-05,
      "loss": 4.4764,
      "step": 31
    },
    {
      "epoch": 0.020904785235995427,
      "grad_norm": 4.05047607421875,
      "learning_rate": 9.672822322997305e-05,
      "loss": 4.8734,
      "step": 32
    },
    {
      "epoch": 0.021558059774620286,
      "grad_norm": 2.990489959716797,
      "learning_rate": 9.642770192448536e-05,
      "loss": 4.6791,
      "step": 33
    },
    {
      "epoch": 0.02221133431324514,
      "grad_norm": 5.1107587814331055,
      "learning_rate": 9.611448774886924e-05,
      "loss": 4.62,
      "step": 34
    },
    {
      "epoch": 0.02286460885187,
      "grad_norm": 4.496232032775879,
      "learning_rate": 9.578866633275288e-05,
      "loss": 4.8254,
      "step": 35
    },
    {
      "epoch": 0.023517883390494855,
      "grad_norm": 4.5912370681762695,
      "learning_rate": 9.545032675245813e-05,
      "loss": 4.5929,
      "step": 36
    },
    {
      "epoch": 0.024171157929119714,
      "grad_norm": 4.44483757019043,
      "learning_rate": 9.509956150664796e-05,
      "loss": 4.8056,
      "step": 37
    },
    {
      "epoch": 0.02482443246774457,
      "grad_norm": 4.195216178894043,
      "learning_rate": 9.473646649103818e-05,
      "loss": 4.6178,
      "step": 38
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 4.176390647888184,
      "learning_rate": 9.43611409721806e-05,
      "loss": 4.4497,
      "step": 39
    },
    {
      "epoch": 0.026130981544994283,
      "grad_norm": 3.8274049758911133,
      "learning_rate": 9.397368756032445e-05,
      "loss": 4.6279,
      "step": 40
    },
    {
      "epoch": 0.026784256083619142,
      "grad_norm": 5.833320140838623,
      "learning_rate": 9.357421218136386e-05,
      "loss": 4.7703,
      "step": 41
    },
    {
      "epoch": 0.027437530622243998,
      "grad_norm": 4.18890905380249,
      "learning_rate": 9.316282404787871e-05,
      "loss": 4.8122,
      "step": 42
    },
    {
      "epoch": 0.028090805160868856,
      "grad_norm": 6.048251628875732,
      "learning_rate": 9.273963562927695e-05,
      "loss": 4.8024,
      "step": 43
    },
    {
      "epoch": 0.02874407969949371,
      "grad_norm": 5.917573928833008,
      "learning_rate": 9.230476262104677e-05,
      "loss": 5.0333,
      "step": 44
    },
    {
      "epoch": 0.02939735423811857,
      "grad_norm": 6.865419387817383,
      "learning_rate": 9.185832391312644e-05,
      "loss": 4.9252,
      "step": 45
    },
    {
      "epoch": 0.030050628776743426,
      "grad_norm": 7.134896755218506,
      "learning_rate": 9.140044155740101e-05,
      "loss": 4.8173,
      "step": 46
    },
    {
      "epoch": 0.030703903315368285,
      "grad_norm": 6.421178340911865,
      "learning_rate": 9.093124073433463e-05,
      "loss": 4.7522,
      "step": 47
    },
    {
      "epoch": 0.03135717785399314,
      "grad_norm": 7.813950538635254,
      "learning_rate": 9.045084971874738e-05,
      "loss": 4.2685,
      "step": 48
    },
    {
      "epoch": 0.032010452392617995,
      "grad_norm": 6.996130466461182,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.3189,
      "step": 49
    },
    {
      "epoch": 0.03266372693124286,
      "grad_norm": 12.550222396850586,
      "learning_rate": 8.945702546981969e-05,
      "loss": 5.9807,
      "step": 50
    },
    {
      "epoch": 0.03266372693124286,
      "eval_loss": 1.1339781284332275,
      "eval_runtime": 39.6592,
      "eval_samples_per_second": 65.004,
      "eval_steps_per_second": 16.264,
      "step": 50
    },
    {
      "epoch": 0.03331700146986771,
      "grad_norm": 14.085443496704102,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.4881,
      "step": 51
    },
    {
      "epoch": 0.03397027600849257,
      "grad_norm": 9.75980281829834,
      "learning_rate": 8.842005554284296e-05,
      "loss": 4.2517,
      "step": 52
    },
    {
      "epoch": 0.034623550547117424,
      "grad_norm": 6.037260055541992,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.1351,
      "step": 53
    },
    {
      "epoch": 0.035276825085742286,
      "grad_norm": 4.675490379333496,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.9956,
      "step": 54
    },
    {
      "epoch": 0.03593009962436714,
      "grad_norm": 3.2766640186309814,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.9198,
      "step": 55
    },
    {
      "epoch": 0.036583374162992,
      "grad_norm": 4.602175235748291,
      "learning_rate": 8.622126023955446e-05,
      "loss": 4.0836,
      "step": 56
    },
    {
      "epoch": 0.03723664870161685,
      "grad_norm": 3.565039873123169,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.9597,
      "step": 57
    },
    {
      "epoch": 0.037889923240241714,
      "grad_norm": 4.522263050079346,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.7911,
      "step": 58
    },
    {
      "epoch": 0.03854319777886657,
      "grad_norm": 2.9280788898468018,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.8086,
      "step": 59
    },
    {
      "epoch": 0.039196472317491425,
      "grad_norm": 4.336091041564941,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.8195,
      "step": 60
    },
    {
      "epoch": 0.03984974685611628,
      "grad_norm": 4.253453731536865,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.993,
      "step": 61
    },
    {
      "epoch": 0.04050302139474114,
      "grad_norm": 3.1280858516693115,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.8574,
      "step": 62
    },
    {
      "epoch": 0.041156295933366,
      "grad_norm": 3.3707027435302734,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.7375,
      "step": 63
    },
    {
      "epoch": 0.04180957047199085,
      "grad_norm": 2.288654327392578,
      "learning_rate": 8.135881792367686e-05,
      "loss": 3.8654,
      "step": 64
    },
    {
      "epoch": 0.04246284501061571,
      "grad_norm": 2.9625179767608643,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.6704,
      "step": 65
    },
    {
      "epoch": 0.04311611954924057,
      "grad_norm": 2.08366322517395,
      "learning_rate": 8.005405736415126e-05,
      "loss": 3.843,
      "step": 66
    },
    {
      "epoch": 0.043769394087865426,
      "grad_norm": 2.7566492557525635,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.669,
      "step": 67
    },
    {
      "epoch": 0.04442266862649028,
      "grad_norm": 2.4501678943634033,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.8105,
      "step": 68
    },
    {
      "epoch": 0.04507594316511514,
      "grad_norm": 2.2400918006896973,
      "learning_rate": 7.803575286758364e-05,
      "loss": 3.7917,
      "step": 69
    },
    {
      "epoch": 0.04572921770374,
      "grad_norm": 2.490662097930908,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.86,
      "step": 70
    },
    {
      "epoch": 0.046382492242364855,
      "grad_norm": 1.7450947761535645,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.9378,
      "step": 71
    },
    {
      "epoch": 0.04703576678098971,
      "grad_norm": 2.1534664630889893,
      "learning_rate": 7.594847868906076e-05,
      "loss": 3.8169,
      "step": 72
    },
    {
      "epoch": 0.047689041319614565,
      "grad_norm": 2.0554611682891846,
      "learning_rate": 7.52382768867422e-05,
      "loss": 3.8316,
      "step": 73
    },
    {
      "epoch": 0.04834231585823943,
      "grad_norm": 2.428230047225952,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.6406,
      "step": 74
    },
    {
      "epoch": 0.04899559039686428,
      "grad_norm": 2.7007052898406982,
      "learning_rate": 7.379736965185368e-05,
      "loss": 3.8438,
      "step": 75
    },
    {
      "epoch": 0.04964886493548914,
      "grad_norm": 2.6560676097869873,
      "learning_rate": 7.30670581489344e-05,
      "loss": 3.9475,
      "step": 76
    },
    {
      "epoch": 0.050302139474113994,
      "grad_norm": 2.52561616897583,
      "learning_rate": 7.233044034264034e-05,
      "loss": 4.0081,
      "step": 77
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 2.915048599243164,
      "learning_rate": 7.158771761692464e-05,
      "loss": 3.9402,
      "step": 78
    },
    {
      "epoch": 0.05160868855136371,
      "grad_norm": 3.199303150177002,
      "learning_rate": 7.083909302476453e-05,
      "loss": 4.28,
      "step": 79
    },
    {
      "epoch": 0.05226196308998857,
      "grad_norm": 2.74607253074646,
      "learning_rate": 7.008477123264848e-05,
      "loss": 4.0714,
      "step": 80
    },
    {
      "epoch": 0.05291523762861342,
      "grad_norm": 3.2236008644104004,
      "learning_rate": 6.932495846462261e-05,
      "loss": 4.0507,
      "step": 81
    },
    {
      "epoch": 0.053568512167238284,
      "grad_norm": 2.6463396549224854,
      "learning_rate": 6.855986244591104e-05,
      "loss": 3.8816,
      "step": 82
    },
    {
      "epoch": 0.05422178670586314,
      "grad_norm": 2.703219175338745,
      "learning_rate": 6.778969234612584e-05,
      "loss": 4.4263,
      "step": 83
    },
    {
      "epoch": 0.054875061244487995,
      "grad_norm": 3.069406270980835,
      "learning_rate": 6.701465872208216e-05,
      "loss": 4.3897,
      "step": 84
    },
    {
      "epoch": 0.05552833578311285,
      "grad_norm": 3.0235583782196045,
      "learning_rate": 6.623497346023418e-05,
      "loss": 4.1988,
      "step": 85
    },
    {
      "epoch": 0.05618161032173771,
      "grad_norm": 3.773071765899658,
      "learning_rate": 6.545084971874738e-05,
      "loss": 4.7398,
      "step": 86
    },
    {
      "epoch": 0.05683488486036257,
      "grad_norm": 4.131563186645508,
      "learning_rate": 6.466250186922325e-05,
      "loss": 4.2208,
      "step": 87
    },
    {
      "epoch": 0.05748815939898742,
      "grad_norm": 3.3429348468780518,
      "learning_rate": 6.387014543809223e-05,
      "loss": 3.909,
      "step": 88
    },
    {
      "epoch": 0.05814143393761228,
      "grad_norm": 4.636299133300781,
      "learning_rate": 6.307399704769099e-05,
      "loss": 4.5743,
      "step": 89
    },
    {
      "epoch": 0.05879470847623714,
      "grad_norm": 4.548759937286377,
      "learning_rate": 6.227427435703997e-05,
      "loss": 4.3087,
      "step": 90
    },
    {
      "epoch": 0.059447983014861996,
      "grad_norm": 4.737814903259277,
      "learning_rate": 6.147119600233758e-05,
      "loss": 4.362,
      "step": 91
    },
    {
      "epoch": 0.06010125755348685,
      "grad_norm": 5.241085529327393,
      "learning_rate": 6.066498153718735e-05,
      "loss": 4.2418,
      "step": 92
    },
    {
      "epoch": 0.06075453209211171,
      "grad_norm": 5.200568675994873,
      "learning_rate": 5.985585137257401e-05,
      "loss": 4.811,
      "step": 93
    },
    {
      "epoch": 0.06140780663073657,
      "grad_norm": 4.870446681976318,
      "learning_rate": 5.90440267166055e-05,
      "loss": 4.0559,
      "step": 94
    },
    {
      "epoch": 0.062061081169361425,
      "grad_norm": 6.289549827575684,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 4.3993,
      "step": 95
    },
    {
      "epoch": 0.06271435570798628,
      "grad_norm": 6.104208469390869,
      "learning_rate": 5.74131823855921e-05,
      "loss": 4.1192,
      "step": 96
    },
    {
      "epoch": 0.06336763024661114,
      "grad_norm": 7.346236228942871,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 4.7962,
      "step": 97
    },
    {
      "epoch": 0.06402090478523599,
      "grad_norm": 7.959908962249756,
      "learning_rate": 5.577423184847932e-05,
      "loss": 5.281,
      "step": 98
    },
    {
      "epoch": 0.06467417932386085,
      "grad_norm": 7.723431587219238,
      "learning_rate": 5.495227651252315e-05,
      "loss": 5.0611,
      "step": 99
    },
    {
      "epoch": 0.06532745386248572,
      "grad_norm": 11.250986099243164,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 5.7178,
      "step": 100
    },
    {
      "epoch": 0.06532745386248572,
      "eval_loss": 0.9772670865058899,
      "eval_runtime": 39.5601,
      "eval_samples_per_second": 65.167,
      "eval_steps_per_second": 16.304,
      "step": 100
    },
    {
      "epoch": 0.06598072840111056,
      "grad_norm": 5.774088382720947,
      "learning_rate": 5.330452921628497e-05,
      "loss": 3.6978,
      "step": 101
    },
    {
      "epoch": 0.06663400293973543,
      "grad_norm": 4.752793312072754,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.6759,
      "step": 102
    },
    {
      "epoch": 0.06728727747836027,
      "grad_norm": 3.714458465576172,
      "learning_rate": 5.165316846586541e-05,
      "loss": 3.6406,
      "step": 103
    },
    {
      "epoch": 0.06794055201698514,
      "grad_norm": 3.4480419158935547,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.6109,
      "step": 104
    },
    {
      "epoch": 0.06859382655561,
      "grad_norm": 3.1864945888519287,
      "learning_rate": 5e-05,
      "loss": 3.6656,
      "step": 105
    },
    {
      "epoch": 0.06924710109423485,
      "grad_norm": 2.5918140411376953,
      "learning_rate": 4.917330276168208e-05,
      "loss": 3.5345,
      "step": 106
    },
    {
      "epoch": 0.06990037563285971,
      "grad_norm": 2.317542791366577,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.5209,
      "step": 107
    },
    {
      "epoch": 0.07055365017148457,
      "grad_norm": 3.0625545978546143,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.5249,
      "step": 108
    },
    {
      "epoch": 0.07120692471010942,
      "grad_norm": 2.7957537174224854,
      "learning_rate": 4.669547078371504e-05,
      "loss": 3.4871,
      "step": 109
    },
    {
      "epoch": 0.07186019924873428,
      "grad_norm": 2.5409247875213623,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 3.5908,
      "step": 110
    },
    {
      "epoch": 0.07251347378735913,
      "grad_norm": 2.3540966510772705,
      "learning_rate": 4.504772348747687e-05,
      "loss": 3.4871,
      "step": 111
    },
    {
      "epoch": 0.073166748325984,
      "grad_norm": 2.186793088912964,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.4821,
      "step": 112
    },
    {
      "epoch": 0.07382002286460886,
      "grad_norm": 2.11293888092041,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 3.5103,
      "step": 113
    },
    {
      "epoch": 0.0744732974032337,
      "grad_norm": 1.9093185663223267,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.3416,
      "step": 114
    },
    {
      "epoch": 0.07512657194185857,
      "grad_norm": 1.9486185312271118,
      "learning_rate": 4.17702704859633e-05,
      "loss": 3.6038,
      "step": 115
    },
    {
      "epoch": 0.07577984648048343,
      "grad_norm": 2.2110652923583984,
      "learning_rate": 4.095597328339452e-05,
      "loss": 3.4521,
      "step": 116
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 2.0127384662628174,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 3.3426,
      "step": 117
    },
    {
      "epoch": 0.07708639555773314,
      "grad_norm": 1.8601311445236206,
      "learning_rate": 3.933501846281267e-05,
      "loss": 3.4721,
      "step": 118
    },
    {
      "epoch": 0.07773967009635799,
      "grad_norm": 2.1661839485168457,
      "learning_rate": 3.852880399766243e-05,
      "loss": 3.6352,
      "step": 119
    },
    {
      "epoch": 0.07839294463498285,
      "grad_norm": 2.313729763031006,
      "learning_rate": 3.772572564296005e-05,
      "loss": 3.4735,
      "step": 120
    },
    {
      "epoch": 0.07904621917360771,
      "grad_norm": 1.9038708209991455,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 3.5132,
      "step": 121
    },
    {
      "epoch": 0.07969949371223256,
      "grad_norm": 2.19059157371521,
      "learning_rate": 3.612985456190778e-05,
      "loss": 3.4762,
      "step": 122
    },
    {
      "epoch": 0.08035276825085742,
      "grad_norm": 2.2744879722595215,
      "learning_rate": 3.533749813077677e-05,
      "loss": 3.6078,
      "step": 123
    },
    {
      "epoch": 0.08100604278948229,
      "grad_norm": 2.3661816120147705,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.4587,
      "step": 124
    },
    {
      "epoch": 0.08165931732810713,
      "grad_norm": 2.1660680770874023,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 3.6221,
      "step": 125
    },
    {
      "epoch": 0.082312591866732,
      "grad_norm": 1.7373437881469727,
      "learning_rate": 3.298534127791785e-05,
      "loss": 3.7606,
      "step": 126
    },
    {
      "epoch": 0.08296586640535684,
      "grad_norm": 2.5906012058258057,
      "learning_rate": 3.221030765387417e-05,
      "loss": 3.6921,
      "step": 127
    },
    {
      "epoch": 0.0836191409439817,
      "grad_norm": 2.2709522247314453,
      "learning_rate": 3.144013755408895e-05,
      "loss": 3.8475,
      "step": 128
    },
    {
      "epoch": 0.08427241548260657,
      "grad_norm": 2.2256815433502197,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 3.7146,
      "step": 129
    },
    {
      "epoch": 0.08492569002123142,
      "grad_norm": 2.4315743446350098,
      "learning_rate": 2.991522876735154e-05,
      "loss": 3.5815,
      "step": 130
    },
    {
      "epoch": 0.08557896455985628,
      "grad_norm": 2.5622146129608154,
      "learning_rate": 2.916090697523549e-05,
      "loss": 3.7962,
      "step": 131
    },
    {
      "epoch": 0.08623223909848114,
      "grad_norm": 2.752903699874878,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 4.0775,
      "step": 132
    },
    {
      "epoch": 0.08688551363710599,
      "grad_norm": 2.7100331783294678,
      "learning_rate": 2.766955965735968e-05,
      "loss": 3.9692,
      "step": 133
    },
    {
      "epoch": 0.08753878817573085,
      "grad_norm": 3.270238161087036,
      "learning_rate": 2.693294185106562e-05,
      "loss": 3.8109,
      "step": 134
    },
    {
      "epoch": 0.0881920627143557,
      "grad_norm": 3.353745222091675,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 4.3294,
      "step": 135
    },
    {
      "epoch": 0.08884533725298056,
      "grad_norm": 3.5236260890960693,
      "learning_rate": 2.547882480847461e-05,
      "loss": 4.0811,
      "step": 136
    },
    {
      "epoch": 0.08949861179160543,
      "grad_norm": 3.774054527282715,
      "learning_rate": 2.476172311325783e-05,
      "loss": 4.3374,
      "step": 137
    },
    {
      "epoch": 0.09015188633023027,
      "grad_norm": 4.036478519439697,
      "learning_rate": 2.405152131093926e-05,
      "loss": 4.0773,
      "step": 138
    },
    {
      "epoch": 0.09080516086885514,
      "grad_norm": 4.228728771209717,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 4.0459,
      "step": 139
    },
    {
      "epoch": 0.09145843540748,
      "grad_norm": 4.082521915435791,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 4.0264,
      "step": 140
    },
    {
      "epoch": 0.09211170994610485,
      "grad_norm": 4.148725986480713,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.9779,
      "step": 141
    },
    {
      "epoch": 0.09276498448472971,
      "grad_norm": 4.940046787261963,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.9118,
      "step": 142
    },
    {
      "epoch": 0.09341825902335456,
      "grad_norm": 5.009074687957764,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.1053,
      "step": 143
    },
    {
      "epoch": 0.09407153356197942,
      "grad_norm": 6.002546310424805,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 4.6179,
      "step": 144
    },
    {
      "epoch": 0.09472480810060428,
      "grad_norm": 5.81488561630249,
      "learning_rate": 1.928936436551661e-05,
      "loss": 4.1827,
      "step": 145
    },
    {
      "epoch": 0.09537808263922913,
      "grad_norm": 7.2876434326171875,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 4.2981,
      "step": 146
    },
    {
      "epoch": 0.096031357177854,
      "grad_norm": 6.615135669708252,
      "learning_rate": 1.800157297483417e-05,
      "loss": 4.2244,
      "step": 147
    },
    {
      "epoch": 0.09668463171647886,
      "grad_norm": 6.6335320472717285,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 4.6781,
      "step": 148
    },
    {
      "epoch": 0.0973379062551037,
      "grad_norm": 8.589600563049316,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 5.009,
      "step": 149
    },
    {
      "epoch": 0.09799118079372857,
      "grad_norm": 11.927630424499512,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 6.1241,
      "step": 150
    },
    {
      "epoch": 0.09799118079372857,
      "eval_loss": 0.906474232673645,
      "eval_runtime": 39.7169,
      "eval_samples_per_second": 64.909,
      "eval_steps_per_second": 16.24,
      "step": 150
    },
    {
      "epoch": 0.09864445533235343,
      "grad_norm": 3.6621720790863037,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.3566,
      "step": 151
    },
    {
      "epoch": 0.09929772987097828,
      "grad_norm": 3.6745262145996094,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 3.4512,
      "step": 152
    },
    {
      "epoch": 0.09995100440960314,
      "grad_norm": 3.3215622901916504,
      "learning_rate": 1.435357758543015e-05,
      "loss": 3.4355,
      "step": 153
    },
    {
      "epoch": 0.10060427894822799,
      "grad_norm": 2.7204957008361816,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 3.4343,
      "step": 154
    },
    {
      "epoch": 0.10125755348685285,
      "grad_norm": 2.647876739501953,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 3.3816,
      "step": 155
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 2.959148645401001,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 3.3274,
      "step": 156
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 2.6461031436920166,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 3.3055,
      "step": 157
    },
    {
      "epoch": 0.10321737710272742,
      "grad_norm": 2.347519636154175,
      "learning_rate": 1.157994445715706e-05,
      "loss": 3.416,
      "step": 158
    },
    {
      "epoch": 0.10387065164135229,
      "grad_norm": 2.1401801109313965,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.3025,
      "step": 159
    },
    {
      "epoch": 0.10452392617997713,
      "grad_norm": 2.0405924320220947,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 3.3942,
      "step": 160
    },
    {
      "epoch": 0.105177200718602,
      "grad_norm": 1.8742287158966064,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 3.2757,
      "step": 161
    },
    {
      "epoch": 0.10583047525722684,
      "grad_norm": 1.7792880535125732,
      "learning_rate": 9.549150281252633e-06,
      "loss": 3.4388,
      "step": 162
    },
    {
      "epoch": 0.1064837497958517,
      "grad_norm": 1.6856492757797241,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.3436,
      "step": 163
    },
    {
      "epoch": 0.10713702433447657,
      "grad_norm": 1.7957298755645752,
      "learning_rate": 8.599558442598998e-06,
      "loss": 3.2971,
      "step": 164
    },
    {
      "epoch": 0.10779029887310142,
      "grad_norm": 2.001119375228882,
      "learning_rate": 8.141676086873572e-06,
      "loss": 3.4581,
      "step": 165
    },
    {
      "epoch": 0.10844357341172628,
      "grad_norm": 1.8569761514663696,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.2454,
      "step": 166
    },
    {
      "epoch": 0.10909684795035114,
      "grad_norm": 1.9288561344146729,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.4321,
      "step": 167
    },
    {
      "epoch": 0.10975012248897599,
      "grad_norm": 1.916345477104187,
      "learning_rate": 6.837175952121306e-06,
      "loss": 3.547,
      "step": 168
    },
    {
      "epoch": 0.11040339702760085,
      "grad_norm": 1.643003225326538,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.339,
      "step": 169
    },
    {
      "epoch": 0.1110566715662257,
      "grad_norm": 1.814770221710205,
      "learning_rate": 6.026312439675552e-06,
      "loss": 3.4491,
      "step": 170
    },
    {
      "epoch": 0.11170994610485056,
      "grad_norm": 1.870467185974121,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 3.4128,
      "step": 171
    },
    {
      "epoch": 0.11236322064347543,
      "grad_norm": 1.6289000511169434,
      "learning_rate": 5.263533508961827e-06,
      "loss": 3.514,
      "step": 172
    },
    {
      "epoch": 0.11301649518210027,
      "grad_norm": 1.882980227470398,
      "learning_rate": 4.900438493352055e-06,
      "loss": 3.5481,
      "step": 173
    },
    {
      "epoch": 0.11366976972072514,
      "grad_norm": 1.7779327630996704,
      "learning_rate": 4.549673247541875e-06,
      "loss": 3.458,
      "step": 174
    },
    {
      "epoch": 0.11432304425935,
      "grad_norm": 1.7439157962799072,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 3.3835,
      "step": 175
    },
    {
      "epoch": 0.11497631879797485,
      "grad_norm": 1.8990323543548584,
      "learning_rate": 3.885512251130763e-06,
      "loss": 3.637,
      "step": 176
    },
    {
      "epoch": 0.11562959333659971,
      "grad_norm": 2.0242409706115723,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 3.7646,
      "step": 177
    },
    {
      "epoch": 0.11628286787522456,
      "grad_norm": 2.1451168060302734,
      "learning_rate": 3.271776770026963e-06,
      "loss": 3.805,
      "step": 178
    },
    {
      "epoch": 0.11693614241384942,
      "grad_norm": 2.539607286453247,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 3.6092,
      "step": 179
    },
    {
      "epoch": 0.11758941695247428,
      "grad_norm": 2.3244380950927734,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 3.6768,
      "step": 180
    },
    {
      "epoch": 0.11824269149109913,
      "grad_norm": 2.5504250526428223,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 3.8788,
      "step": 181
    },
    {
      "epoch": 0.11889596602972399,
      "grad_norm": 2.5190818309783936,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 3.9667,
      "step": 182
    },
    {
      "epoch": 0.11954924056834886,
      "grad_norm": 2.6391968727111816,
      "learning_rate": 1.962316193157593e-06,
      "loss": 3.6875,
      "step": 183
    },
    {
      "epoch": 0.1202025151069737,
      "grad_norm": 2.807856559753418,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 4.0797,
      "step": 184
    },
    {
      "epoch": 0.12085578964559857,
      "grad_norm": 2.7421610355377197,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 4.6649,
      "step": 185
    },
    {
      "epoch": 0.12150906418422341,
      "grad_norm": 3.2052602767944336,
      "learning_rate": 1.333670137599713e-06,
      "loss": 4.1278,
      "step": 186
    },
    {
      "epoch": 0.12216233872284828,
      "grad_norm": 3.4585554599761963,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 4.0451,
      "step": 187
    },
    {
      "epoch": 0.12281561326147314,
      "grad_norm": 3.651754140853882,
      "learning_rate": 9.810017062595322e-07,
      "loss": 4.3103,
      "step": 188
    },
    {
      "epoch": 0.12346888780009799,
      "grad_norm": 3.901911973953247,
      "learning_rate": 8.247462563808817e-07,
      "loss": 4.1074,
      "step": 189
    },
    {
      "epoch": 0.12412216233872285,
      "grad_norm": 3.8312618732452393,
      "learning_rate": 6.819348298638839e-07,
      "loss": 4.0429,
      "step": 190
    },
    {
      "epoch": 0.12477543687734771,
      "grad_norm": 4.298215866088867,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.997,
      "step": 191
    },
    {
      "epoch": 0.12542871141597256,
      "grad_norm": 4.657850742340088,
      "learning_rate": 4.367965336512403e-07,
      "loss": 4.0727,
      "step": 192
    },
    {
      "epoch": 0.1260819859545974,
      "grad_norm": 5.246180534362793,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 4.5917,
      "step": 193
    },
    {
      "epoch": 0.12673526049322228,
      "grad_norm": 5.862904071807861,
      "learning_rate": 2.458548727494292e-07,
      "loss": 4.0001,
      "step": 194
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 5.669924736022949,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 4.6843,
      "step": 195
    },
    {
      "epoch": 0.12804180957047198,
      "grad_norm": 5.923580646514893,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 4.073,
      "step": 196
    },
    {
      "epoch": 0.12869508410909686,
      "grad_norm": 6.806802749633789,
      "learning_rate": 6.150154258476315e-08,
      "loss": 4.4933,
      "step": 197
    },
    {
      "epoch": 0.1293483586477217,
      "grad_norm": 6.4826860427856445,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 4.6264,
      "step": 198
    },
    {
      "epoch": 0.13000163318634655,
      "grad_norm": 8.152565002441406,
      "learning_rate": 6.834750376549792e-09,
      "loss": 4.4419,
      "step": 199
    },
    {
      "epoch": 0.13065490772497143,
      "grad_norm": 9.753582000732422,
      "learning_rate": 0.0,
      "loss": 4.7468,
      "step": 200
    },
    {
      "epoch": 0.13065490772497143,
      "eval_loss": 0.8892158269882202,
      "eval_runtime": 39.7759,
      "eval_samples_per_second": 64.813,
      "eval_steps_per_second": 16.216,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.9394829049921536e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}