{
  "best_metric": 10.70785140991211,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.07426661715558856,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003713330857779428,
      "grad_norm": 0.47366201877593994,
      "learning_rate": 1.009e-05,
      "loss": 10.835,
      "step": 1
    },
    {
      "epoch": 0.0003713330857779428,
      "eval_loss": 10.832501411437988,
      "eval_runtime": 8.632,
      "eval_samples_per_second": 131.372,
      "eval_steps_per_second": 32.901,
      "step": 1
    },
    {
      "epoch": 0.0007426661715558856,
      "grad_norm": 0.39668479561805725,
      "learning_rate": 2.018e-05,
      "loss": 10.8341,
      "step": 2
    },
    {
      "epoch": 0.0011139992573338284,
      "grad_norm": 0.36069685220718384,
      "learning_rate": 3.027e-05,
      "loss": 10.8294,
      "step": 3
    },
    {
      "epoch": 0.0014853323431117712,
      "grad_norm": 0.3754219710826874,
      "learning_rate": 4.036e-05,
      "loss": 10.8319,
      "step": 4
    },
    {
      "epoch": 0.001856665428889714,
      "grad_norm": 0.4531574547290802,
      "learning_rate": 5.045e-05,
      "loss": 10.8283,
      "step": 5
    },
    {
      "epoch": 0.0022279985146676567,
      "grad_norm": 0.46290841698646545,
      "learning_rate": 6.054e-05,
      "loss": 10.8327,
      "step": 6
    },
    {
      "epoch": 0.0025993316004456,
      "grad_norm": 0.3586803674697876,
      "learning_rate": 7.062999999999999e-05,
      "loss": 10.8329,
      "step": 7
    },
    {
      "epoch": 0.0029706646862235424,
      "grad_norm": 0.4396965205669403,
      "learning_rate": 8.072e-05,
      "loss": 10.8277,
      "step": 8
    },
    {
      "epoch": 0.0033419977720014855,
      "grad_norm": 0.5143634676933289,
      "learning_rate": 9.081e-05,
      "loss": 10.8298,
      "step": 9
    },
    {
      "epoch": 0.003713330857779428,
      "grad_norm": 0.41205930709838867,
      "learning_rate": 0.0001009,
      "loss": 10.8375,
      "step": 10
    },
    {
      "epoch": 0.004084663943557371,
      "grad_norm": 0.4410165548324585,
      "learning_rate": 0.00010036894736842106,
      "loss": 10.8263,
      "step": 11
    },
    {
      "epoch": 0.004455997029335313,
      "grad_norm": 0.47727352380752563,
      "learning_rate": 9.98378947368421e-05,
      "loss": 10.8128,
      "step": 12
    },
    {
      "epoch": 0.004827330115113257,
      "grad_norm": 0.48599427938461304,
      "learning_rate": 9.930684210526315e-05,
      "loss": 10.8048,
      "step": 13
    },
    {
      "epoch": 0.0051986632008912,
      "grad_norm": 0.5060449838638306,
      "learning_rate": 9.877578947368421e-05,
      "loss": 10.8201,
      "step": 14
    },
    {
      "epoch": 0.005569996286669142,
      "grad_norm": 0.4900880455970764,
      "learning_rate": 9.824473684210527e-05,
      "loss": 10.8079,
      "step": 15
    },
    {
      "epoch": 0.005941329372447085,
      "grad_norm": 0.5277690291404724,
      "learning_rate": 9.771368421052632e-05,
      "loss": 10.8078,
      "step": 16
    },
    {
      "epoch": 0.0063126624582250275,
      "grad_norm": 0.5447446703910828,
      "learning_rate": 9.718263157894736e-05,
      "loss": 10.8192,
      "step": 17
    },
    {
      "epoch": 0.006683995544002971,
      "grad_norm": 0.5573148131370544,
      "learning_rate": 9.665157894736842e-05,
      "loss": 10.798,
      "step": 18
    },
    {
      "epoch": 0.007055328629780914,
      "grad_norm": 0.5032119750976562,
      "learning_rate": 9.612052631578948e-05,
      "loss": 10.7994,
      "step": 19
    },
    {
      "epoch": 0.007426661715558856,
      "grad_norm": 0.46621081233024597,
      "learning_rate": 9.558947368421052e-05,
      "loss": 10.8053,
      "step": 20
    },
    {
      "epoch": 0.007797994801336799,
      "grad_norm": 0.4873751103878021,
      "learning_rate": 9.505842105263159e-05,
      "loss": 10.8143,
      "step": 21
    },
    {
      "epoch": 0.008169327887114742,
      "grad_norm": 0.618602454662323,
      "learning_rate": 9.452736842105263e-05,
      "loss": 10.7913,
      "step": 22
    },
    {
      "epoch": 0.008540660972892685,
      "grad_norm": 0.5478153228759766,
      "learning_rate": 9.399631578947368e-05,
      "loss": 10.7964,
      "step": 23
    },
    {
      "epoch": 0.008911994058670627,
      "grad_norm": 0.5051780939102173,
      "learning_rate": 9.346526315789474e-05,
      "loss": 10.7684,
      "step": 24
    },
    {
      "epoch": 0.00928332714444857,
      "grad_norm": 0.5710480213165283,
      "learning_rate": 9.293421052631578e-05,
      "loss": 10.7952,
      "step": 25
    },
    {
      "epoch": 0.009654660230226514,
      "grad_norm": 0.5759515166282654,
      "learning_rate": 9.240315789473684e-05,
      "loss": 10.7925,
      "step": 26
    },
    {
      "epoch": 0.010025993316004456,
      "grad_norm": 0.5633918642997742,
      "learning_rate": 9.18721052631579e-05,
      "loss": 10.7648,
      "step": 27
    },
    {
      "epoch": 0.0103973264017824,
      "grad_norm": 0.5759943723678589,
      "learning_rate": 9.134105263157895e-05,
      "loss": 10.756,
      "step": 28
    },
    {
      "epoch": 0.010768659487560341,
      "grad_norm": 0.516233503818512,
      "learning_rate": 9.081e-05,
      "loss": 10.7807,
      "step": 29
    },
    {
      "epoch": 0.011139992573338284,
      "grad_norm": 0.5473969578742981,
      "learning_rate": 9.027894736842105e-05,
      "loss": 10.7649,
      "step": 30
    },
    {
      "epoch": 0.011511325659116228,
      "grad_norm": 0.46038320660591125,
      "learning_rate": 8.97478947368421e-05,
      "loss": 10.7741,
      "step": 31
    },
    {
      "epoch": 0.01188265874489417,
      "grad_norm": 0.5598097443580627,
      "learning_rate": 8.921684210526316e-05,
      "loss": 10.7898,
      "step": 32
    },
    {
      "epoch": 0.012253991830672113,
      "grad_norm": 0.5376091003417969,
      "learning_rate": 8.86857894736842e-05,
      "loss": 10.7795,
      "step": 33
    },
    {
      "epoch": 0.012625324916450055,
      "grad_norm": 0.4991442859172821,
      "learning_rate": 8.815473684210527e-05,
      "loss": 10.7416,
      "step": 34
    },
    {
      "epoch": 0.012996658002227999,
      "grad_norm": 0.5336849093437195,
      "learning_rate": 8.762368421052631e-05,
      "loss": 10.7558,
      "step": 35
    },
    {
      "epoch": 0.013367991088005942,
      "grad_norm": 0.55255526304245,
      "learning_rate": 8.709263157894737e-05,
      "loss": 10.743,
      "step": 36
    },
    {
      "epoch": 0.013739324173783884,
      "grad_norm": 0.5434712767601013,
      "learning_rate": 8.656157894736843e-05,
      "loss": 10.7494,
      "step": 37
    },
    {
      "epoch": 0.014110657259561827,
      "grad_norm": 0.5447563529014587,
      "learning_rate": 8.603052631578947e-05,
      "loss": 10.7471,
      "step": 38
    },
    {
      "epoch": 0.014481990345339769,
      "grad_norm": 0.5778723955154419,
      "learning_rate": 8.549947368421052e-05,
      "loss": 10.732,
      "step": 39
    },
    {
      "epoch": 0.014853323431117713,
      "grad_norm": 0.5451592206954956,
      "learning_rate": 8.496842105263158e-05,
      "loss": 10.7317,
      "step": 40
    },
    {
      "epoch": 0.015224656516895656,
      "grad_norm": 0.509575605392456,
      "learning_rate": 8.443736842105264e-05,
      "loss": 10.76,
      "step": 41
    },
    {
      "epoch": 0.015595989602673598,
      "grad_norm": 0.5755866765975952,
      "learning_rate": 8.390631578947369e-05,
      "loss": 10.7328,
      "step": 42
    },
    {
      "epoch": 0.01596732268845154,
      "grad_norm": 0.5568187832832336,
      "learning_rate": 8.337526315789473e-05,
      "loss": 10.7154,
      "step": 43
    },
    {
      "epoch": 0.016338655774229483,
      "grad_norm": 0.5118264555931091,
      "learning_rate": 8.284421052631579e-05,
      "loss": 10.7452,
      "step": 44
    },
    {
      "epoch": 0.01670998886000743,
      "grad_norm": 0.559354841709137,
      "learning_rate": 8.231315789473685e-05,
      "loss": 10.731,
      "step": 45
    },
    {
      "epoch": 0.01708132194578537,
      "grad_norm": 0.5593016147613525,
      "learning_rate": 8.178210526315789e-05,
      "loss": 10.7448,
      "step": 46
    },
    {
      "epoch": 0.017452655031563312,
      "grad_norm": 0.6811694502830505,
      "learning_rate": 8.125105263157894e-05,
      "loss": 10.7603,
      "step": 47
    },
    {
      "epoch": 0.017823988117341254,
      "grad_norm": 0.4952441155910492,
      "learning_rate": 8.072e-05,
      "loss": 10.7203,
      "step": 48
    },
    {
      "epoch": 0.0181953212031192,
      "grad_norm": 0.5311501026153564,
      "learning_rate": 8.018894736842106e-05,
      "loss": 10.7215,
      "step": 49
    },
    {
      "epoch": 0.01856665428889714,
      "grad_norm": 0.5852133631706238,
      "learning_rate": 7.965789473684211e-05,
      "loss": 10.728,
      "step": 50
    },
    {
      "epoch": 0.01856665428889714,
      "eval_loss": 10.737244606018066,
      "eval_runtime": 8.5322,
      "eval_samples_per_second": 132.909,
      "eval_steps_per_second": 33.286,
      "step": 50
    },
    {
      "epoch": 0.018937987374675083,
      "grad_norm": 0.42146560549736023,
      "learning_rate": 7.912684210526315e-05,
      "loss": 10.7457,
      "step": 51
    },
    {
      "epoch": 0.019309320460453028,
      "grad_norm": 0.30984774231910706,
      "learning_rate": 7.859578947368421e-05,
      "loss": 10.7561,
      "step": 52
    },
    {
      "epoch": 0.01968065354623097,
      "grad_norm": 0.3268715441226959,
      "learning_rate": 7.806473684210527e-05,
      "loss": 10.745,
      "step": 53
    },
    {
      "epoch": 0.02005198663200891,
      "grad_norm": 0.33481699228286743,
      "learning_rate": 7.753368421052631e-05,
      "loss": 10.7458,
      "step": 54
    },
    {
      "epoch": 0.020423319717786857,
      "grad_norm": 0.33349162340164185,
      "learning_rate": 7.700263157894738e-05,
      "loss": 10.7573,
      "step": 55
    },
    {
      "epoch": 0.0207946528035648,
      "grad_norm": 0.340987890958786,
      "learning_rate": 7.647157894736842e-05,
      "loss": 10.7356,
      "step": 56
    },
    {
      "epoch": 0.02116598588934274,
      "grad_norm": 0.38484814763069153,
      "learning_rate": 7.594052631578948e-05,
      "loss": 10.7508,
      "step": 57
    },
    {
      "epoch": 0.021537318975120682,
      "grad_norm": 0.37901782989501953,
      "learning_rate": 7.540947368421053e-05,
      "loss": 10.7437,
      "step": 58
    },
    {
      "epoch": 0.021908652060898627,
      "grad_norm": 0.3089980483055115,
      "learning_rate": 7.487842105263157e-05,
      "loss": 10.7464,
      "step": 59
    },
    {
      "epoch": 0.02227998514667657,
      "grad_norm": 0.3815767168998718,
      "learning_rate": 7.434736842105263e-05,
      "loss": 10.7555,
      "step": 60
    },
    {
      "epoch": 0.02265131823245451,
      "grad_norm": 0.312149316072464,
      "learning_rate": 7.381631578947368e-05,
      "loss": 10.7363,
      "step": 61
    },
    {
      "epoch": 0.023022651318232456,
      "grad_norm": 0.31698325276374817,
      "learning_rate": 7.328526315789474e-05,
      "loss": 10.7385,
      "step": 62
    },
    {
      "epoch": 0.023393984404010398,
      "grad_norm": 0.3198413848876953,
      "learning_rate": 7.27542105263158e-05,
      "loss": 10.7304,
      "step": 63
    },
    {
      "epoch": 0.02376531748978834,
      "grad_norm": 0.32194074988365173,
      "learning_rate": 7.222315789473684e-05,
      "loss": 10.7269,
      "step": 64
    },
    {
      "epoch": 0.024136650575566285,
      "grad_norm": 0.31859320402145386,
      "learning_rate": 7.16921052631579e-05,
      "loss": 10.7275,
      "step": 65
    },
    {
      "epoch": 0.024507983661344226,
      "grad_norm": 0.3795001804828644,
      "learning_rate": 7.116105263157895e-05,
      "loss": 10.757,
      "step": 66
    },
    {
      "epoch": 0.024879316747122168,
      "grad_norm": 0.35307377576828003,
      "learning_rate": 7.062999999999999e-05,
      "loss": 10.7512,
      "step": 67
    },
    {
      "epoch": 0.02525064983290011,
      "grad_norm": 0.3632878065109253,
      "learning_rate": 7.009894736842106e-05,
      "loss": 10.7457,
      "step": 68
    },
    {
      "epoch": 0.025621982918678055,
      "grad_norm": 0.29969894886016846,
      "learning_rate": 6.95678947368421e-05,
      "loss": 10.7376,
      "step": 69
    },
    {
      "epoch": 0.025993316004455997,
      "grad_norm": 0.32437172532081604,
      "learning_rate": 6.903684210526316e-05,
      "loss": 10.7413,
      "step": 70
    },
    {
      "epoch": 0.02636464909023394,
      "grad_norm": 0.297307550907135,
      "learning_rate": 6.850578947368422e-05,
      "loss": 10.7514,
      "step": 71
    },
    {
      "epoch": 0.026735982176011884,
      "grad_norm": 0.3031882047653198,
      "learning_rate": 6.797473684210526e-05,
      "loss": 10.7323,
      "step": 72
    },
    {
      "epoch": 0.027107315261789826,
      "grad_norm": 0.31023070216178894,
      "learning_rate": 6.744368421052631e-05,
      "loss": 10.7279,
      "step": 73
    },
    {
      "epoch": 0.027478648347567768,
      "grad_norm": 0.3623669445514679,
      "learning_rate": 6.691263157894736e-05,
      "loss": 10.7336,
      "step": 74
    },
    {
      "epoch": 0.027849981433345713,
      "grad_norm": 0.3755355477333069,
      "learning_rate": 6.638157894736843e-05,
      "loss": 10.7147,
      "step": 75
    },
    {
      "epoch": 0.028221314519123655,
      "grad_norm": 0.317091703414917,
      "learning_rate": 6.585052631578948e-05,
      "loss": 10.738,
      "step": 76
    },
    {
      "epoch": 0.028592647604901596,
      "grad_norm": 0.3252423107624054,
      "learning_rate": 6.531947368421052e-05,
      "loss": 10.736,
      "step": 77
    },
    {
      "epoch": 0.028963980690679538,
      "grad_norm": 0.4029505252838135,
      "learning_rate": 6.478842105263158e-05,
      "loss": 10.7214,
      "step": 78
    },
    {
      "epoch": 0.029335313776457483,
      "grad_norm": 0.37360379099845886,
      "learning_rate": 6.425736842105264e-05,
      "loss": 10.7328,
      "step": 79
    },
    {
      "epoch": 0.029706646862235425,
      "grad_norm": 0.35306406021118164,
      "learning_rate": 6.372631578947368e-05,
      "loss": 10.7361,
      "step": 80
    },
    {
      "epoch": 0.030077979948013367,
      "grad_norm": 0.35350292921066284,
      "learning_rate": 6.319526315789473e-05,
      "loss": 10.7224,
      "step": 81
    },
    {
      "epoch": 0.030449313033791312,
      "grad_norm": 0.3728892207145691,
      "learning_rate": 6.266421052631579e-05,
      "loss": 10.7082,
      "step": 82
    },
    {
      "epoch": 0.030820646119569254,
      "grad_norm": 0.38188886642456055,
      "learning_rate": 6.213315789473685e-05,
      "loss": 10.7431,
      "step": 83
    },
    {
      "epoch": 0.031191979205347196,
      "grad_norm": 0.41717568039894104,
      "learning_rate": 6.16021052631579e-05,
      "loss": 10.7524,
      "step": 84
    },
    {
      "epoch": 0.03156331229112514,
      "grad_norm": 0.39692479372024536,
      "learning_rate": 6.107105263157894e-05,
      "loss": 10.7133,
      "step": 85
    },
    {
      "epoch": 0.03193464537690308,
      "grad_norm": 0.4064257740974426,
      "learning_rate": 6.054e-05,
      "loss": 10.7193,
      "step": 86
    },
    {
      "epoch": 0.032305978462681025,
      "grad_norm": 0.41969412565231323,
      "learning_rate": 6.000894736842105e-05,
      "loss": 10.7052,
      "step": 87
    },
    {
      "epoch": 0.032677311548458966,
      "grad_norm": 0.4684526026248932,
      "learning_rate": 5.94778947368421e-05,
      "loss": 10.7227,
      "step": 88
    },
    {
      "epoch": 0.03304864463423691,
      "grad_norm": 0.4255547821521759,
      "learning_rate": 5.894684210526316e-05,
      "loss": 10.7029,
      "step": 89
    },
    {
      "epoch": 0.03341997772001486,
      "grad_norm": 0.47345179319381714,
      "learning_rate": 5.841578947368421e-05,
      "loss": 10.7109,
      "step": 90
    },
    {
      "epoch": 0.0337913108057928,
      "grad_norm": 0.47380679845809937,
      "learning_rate": 5.7884736842105265e-05,
      "loss": 10.7027,
      "step": 91
    },
    {
      "epoch": 0.03416264389157074,
      "grad_norm": 0.45350202918052673,
      "learning_rate": 5.7353684210526314e-05,
      "loss": 10.6865,
      "step": 92
    },
    {
      "epoch": 0.03453397697734868,
      "grad_norm": 0.5256607532501221,
      "learning_rate": 5.6822631578947364e-05,
      "loss": 10.7193,
      "step": 93
    },
    {
      "epoch": 0.034905310063126624,
      "grad_norm": 0.540717363357544,
      "learning_rate": 5.629157894736842e-05,
      "loss": 10.7297,
      "step": 94
    },
    {
      "epoch": 0.035276643148904566,
      "grad_norm": 0.4825849235057831,
      "learning_rate": 5.576052631578948e-05,
      "loss": 10.7247,
      "step": 95
    },
    {
      "epoch": 0.03564797623468251,
      "grad_norm": 0.5021780729293823,
      "learning_rate": 5.522947368421053e-05,
      "loss": 10.6791,
      "step": 96
    },
    {
      "epoch": 0.036019309320460456,
      "grad_norm": 0.4982372224330902,
      "learning_rate": 5.469842105263158e-05,
      "loss": 10.714,
      "step": 97
    },
    {
      "epoch": 0.0363906424062384,
      "grad_norm": 0.5147339701652527,
      "learning_rate": 5.416736842105263e-05,
      "loss": 10.7015,
      "step": 98
    },
    {
      "epoch": 0.03676197549201634,
      "grad_norm": 0.6090283989906311,
      "learning_rate": 5.3636315789473685e-05,
      "loss": 10.712,
      "step": 99
    },
    {
      "epoch": 0.03713330857779428,
      "grad_norm": 0.757976233959198,
      "learning_rate": 5.3105263157894734e-05,
      "loss": 10.697,
      "step": 100
    },
    {
      "epoch": 0.03713330857779428,
      "eval_loss": 10.718417167663574,
      "eval_runtime": 8.5984,
      "eval_samples_per_second": 131.886,
      "eval_steps_per_second": 33.03,
      "step": 100
    },
    {
      "epoch": 0.03750464166357222,
      "grad_norm": 0.296906054019928,
      "learning_rate": 5.257421052631578e-05,
      "loss": 10.7377,
      "step": 101
    },
    {
      "epoch": 0.037875974749350165,
      "grad_norm": 0.3151768445968628,
      "learning_rate": 5.2043157894736846e-05,
      "loss": 10.7402,
      "step": 102
    },
    {
      "epoch": 0.03824730783512811,
      "grad_norm": 0.3774714171886444,
      "learning_rate": 5.1512105263157895e-05,
      "loss": 10.7228,
      "step": 103
    },
    {
      "epoch": 0.038618640920906055,
      "grad_norm": 0.37634870409965515,
      "learning_rate": 5.098105263157895e-05,
      "loss": 10.7249,
      "step": 104
    },
    {
      "epoch": 0.038989974006684,
      "grad_norm": 0.2864985764026642,
      "learning_rate": 5.045e-05,
      "loss": 10.7256,
      "step": 105
    },
    {
      "epoch": 0.03936130709246194,
      "grad_norm": 0.3519158363342285,
      "learning_rate": 4.991894736842105e-05,
      "loss": 10.7307,
      "step": 106
    },
    {
      "epoch": 0.03973264017823988,
      "grad_norm": 0.39623257517814636,
      "learning_rate": 4.9387894736842105e-05,
      "loss": 10.723,
      "step": 107
    },
    {
      "epoch": 0.04010397326401782,
      "grad_norm": 0.39145541191101074,
      "learning_rate": 4.885684210526316e-05,
      "loss": 10.7214,
      "step": 108
    },
    {
      "epoch": 0.040475306349795764,
      "grad_norm": 0.3571566939353943,
      "learning_rate": 4.832578947368421e-05,
      "loss": 10.7348,
      "step": 109
    },
    {
      "epoch": 0.04084663943557371,
      "grad_norm": 0.3960326015949249,
      "learning_rate": 4.779473684210526e-05,
      "loss": 10.7309,
      "step": 110
    },
    {
      "epoch": 0.041217972521351655,
      "grad_norm": 0.4168102443218231,
      "learning_rate": 4.7263684210526315e-05,
      "loss": 10.7314,
      "step": 111
    },
    {
      "epoch": 0.0415893056071296,
      "grad_norm": 0.3861686885356903,
      "learning_rate": 4.673263157894737e-05,
      "loss": 10.7539,
      "step": 112
    },
    {
      "epoch": 0.04196063869290754,
      "grad_norm": 0.34592297673225403,
      "learning_rate": 4.620157894736842e-05,
      "loss": 10.7263,
      "step": 113
    },
    {
      "epoch": 0.04233197177868548,
      "grad_norm": 0.3721361458301544,
      "learning_rate": 4.5670526315789475e-05,
      "loss": 10.7193,
      "step": 114
    },
    {
      "epoch": 0.04270330486446342,
      "grad_norm": 0.3191013038158417,
      "learning_rate": 4.5139473684210524e-05,
      "loss": 10.7208,
      "step": 115
    },
    {
      "epoch": 0.043074637950241364,
      "grad_norm": 0.32601889967918396,
      "learning_rate": 4.460842105263158e-05,
      "loss": 10.7358,
      "step": 116
    },
    {
      "epoch": 0.04344597103601931,
      "grad_norm": 0.3087843954563141,
      "learning_rate": 4.4077368421052636e-05,
      "loss": 10.7348,
      "step": 117
    },
    {
      "epoch": 0.043817304121797254,
      "grad_norm": 0.33753201365470886,
      "learning_rate": 4.3546315789473685e-05,
      "loss": 10.7203,
      "step": 118
    },
    {
      "epoch": 0.044188637207575196,
      "grad_norm": 0.3254026174545288,
      "learning_rate": 4.3015263157894734e-05,
      "loss": 10.7333,
      "step": 119
    },
    {
      "epoch": 0.04455997029335314,
      "grad_norm": 0.3384432792663574,
      "learning_rate": 4.248421052631579e-05,
      "loss": 10.7171,
      "step": 120
    },
    {
      "epoch": 0.04493130337913108,
      "grad_norm": 0.3901914954185486,
      "learning_rate": 4.1953157894736846e-05,
      "loss": 10.7343,
      "step": 121
    },
    {
      "epoch": 0.04530263646490902,
      "grad_norm": 0.31527823209762573,
      "learning_rate": 4.1422105263157895e-05,
      "loss": 10.7208,
      "step": 122
    },
    {
      "epoch": 0.04567396955068696,
      "grad_norm": 0.38066819310188293,
      "learning_rate": 4.0891052631578944e-05,
      "loss": 10.7062,
      "step": 123
    },
    {
      "epoch": 0.04604530263646491,
      "grad_norm": 0.3694753646850586,
      "learning_rate": 4.036e-05,
      "loss": 10.7212,
      "step": 124
    },
    {
      "epoch": 0.046416635722242854,
      "grad_norm": 0.358316570520401,
      "learning_rate": 3.9828947368421056e-05,
      "loss": 10.7281,
      "step": 125
    },
    {
      "epoch": 0.046787968808020795,
      "grad_norm": 0.32940492033958435,
      "learning_rate": 3.9297894736842105e-05,
      "loss": 10.7151,
      "step": 126
    },
    {
      "epoch": 0.04715930189379874,
      "grad_norm": 0.3700310289859772,
      "learning_rate": 3.8766842105263154e-05,
      "loss": 10.7148,
      "step": 127
    },
    {
      "epoch": 0.04753063497957668,
      "grad_norm": 0.3548131585121155,
      "learning_rate": 3.823578947368421e-05,
      "loss": 10.7201,
      "step": 128
    },
    {
      "epoch": 0.04790196806535462,
      "grad_norm": 0.3971237242221832,
      "learning_rate": 3.7704736842105265e-05,
      "loss": 10.7027,
      "step": 129
    },
    {
      "epoch": 0.04827330115113257,
      "grad_norm": 0.41516757011413574,
      "learning_rate": 3.7173684210526315e-05,
      "loss": 10.7175,
      "step": 130
    },
    {
      "epoch": 0.04864463423691051,
      "grad_norm": 0.39661961793899536,
      "learning_rate": 3.664263157894737e-05,
      "loss": 10.6913,
      "step": 131
    },
    {
      "epoch": 0.04901596732268845,
      "grad_norm": 0.34752222895622253,
      "learning_rate": 3.611157894736842e-05,
      "loss": 10.707,
      "step": 132
    },
    {
      "epoch": 0.049387300408466395,
      "grad_norm": 0.3332189917564392,
      "learning_rate": 3.5580526315789475e-05,
      "loss": 10.717,
      "step": 133
    },
    {
      "epoch": 0.049758633494244336,
      "grad_norm": 0.31573349237442017,
      "learning_rate": 3.504947368421053e-05,
      "loss": 10.695,
      "step": 134
    },
    {
      "epoch": 0.05012996658002228,
      "grad_norm": 0.40504634380340576,
      "learning_rate": 3.451842105263158e-05,
      "loss": 10.6974,
      "step": 135
    },
    {
      "epoch": 0.05050129966580022,
      "grad_norm": 0.36559829115867615,
      "learning_rate": 3.398736842105263e-05,
      "loss": 10.7029,
      "step": 136
    },
    {
      "epoch": 0.05087263275157817,
      "grad_norm": 0.34160315990448,
      "learning_rate": 3.345631578947368e-05,
      "loss": 10.7138,
      "step": 137
    },
    {
      "epoch": 0.05124396583735611,
      "grad_norm": 0.49344658851623535,
      "learning_rate": 3.292526315789474e-05,
      "loss": 10.6774,
      "step": 138
    },
    {
      "epoch": 0.05161529892313405,
      "grad_norm": 0.41159558296203613,
      "learning_rate": 3.239421052631579e-05,
      "loss": 10.698,
      "step": 139
    },
    {
      "epoch": 0.051986632008911994,
      "grad_norm": 0.4433419406414032,
      "learning_rate": 3.186315789473684e-05,
      "loss": 10.7159,
      "step": 140
    },
    {
      "epoch": 0.052357965094689936,
      "grad_norm": 0.38202357292175293,
      "learning_rate": 3.1332105263157895e-05,
      "loss": 10.6968,
      "step": 141
    },
    {
      "epoch": 0.05272929818046788,
      "grad_norm": 0.3602319061756134,
      "learning_rate": 3.080105263157895e-05,
      "loss": 10.6922,
      "step": 142
    },
    {
      "epoch": 0.05310063126624582,
      "grad_norm": 0.45938384532928467,
      "learning_rate": 3.027e-05,
      "loss": 10.6852,
      "step": 143
    },
    {
      "epoch": 0.05347196435202377,
      "grad_norm": 0.4686320126056671,
      "learning_rate": 2.973894736842105e-05,
      "loss": 10.7081,
      "step": 144
    },
    {
      "epoch": 0.05384329743780171,
      "grad_norm": 0.36970701813697815,
      "learning_rate": 2.9207894736842105e-05,
      "loss": 10.6916,
      "step": 145
    },
    {
      "epoch": 0.05421463052357965,
      "grad_norm": 0.545552670955658,
      "learning_rate": 2.8676842105263157e-05,
      "loss": 10.7078,
      "step": 146
    },
    {
      "epoch": 0.05458596360935759,
      "grad_norm": 0.43362295627593994,
      "learning_rate": 2.814578947368421e-05,
      "loss": 10.6728,
      "step": 147
    },
    {
      "epoch": 0.054957296695135535,
      "grad_norm": 0.5415939688682556,
      "learning_rate": 2.7614736842105266e-05,
      "loss": 10.6628,
      "step": 148
    },
    {
      "epoch": 0.05532862978091348,
      "grad_norm": 0.5985761284828186,
      "learning_rate": 2.7083684210526315e-05,
      "loss": 10.6833,
      "step": 149
    },
    {
      "epoch": 0.055699962866691426,
      "grad_norm": 0.6192039847373962,
      "learning_rate": 2.6552631578947367e-05,
      "loss": 10.6364,
      "step": 150
    },
    {
      "epoch": 0.055699962866691426,
      "eval_loss": 10.709182739257812,
      "eval_runtime": 8.5261,
      "eval_samples_per_second": 133.003,
      "eval_steps_per_second": 33.309,
      "step": 150
    },
    {
      "epoch": 0.05607129595246937,
      "grad_norm": 0.2916700541973114,
      "learning_rate": 2.6021578947368423e-05,
      "loss": 10.7382,
      "step": 151
    },
    {
      "epoch": 0.05644262903824731,
      "grad_norm": 0.29082828760147095,
      "learning_rate": 2.5490526315789475e-05,
      "loss": 10.7209,
      "step": 152
    },
    {
      "epoch": 0.05681396212402525,
      "grad_norm": 0.25077447295188904,
      "learning_rate": 2.4959473684210524e-05,
      "loss": 10.7209,
      "step": 153
    },
    {
      "epoch": 0.05718529520980319,
      "grad_norm": 0.2654978632926941,
      "learning_rate": 2.442842105263158e-05,
      "loss": 10.7401,
      "step": 154
    },
    {
      "epoch": 0.057556628295581135,
      "grad_norm": 0.2980552315711975,
      "learning_rate": 2.389736842105263e-05,
      "loss": 10.7409,
      "step": 155
    },
    {
      "epoch": 0.057927961381359076,
      "grad_norm": 0.28160735964775085,
      "learning_rate": 2.3366315789473685e-05,
      "loss": 10.7515,
      "step": 156
    },
    {
      "epoch": 0.058299294467137025,
      "grad_norm": 0.32138633728027344,
      "learning_rate": 2.2835263157894738e-05,
      "loss": 10.733,
      "step": 157
    },
    {
      "epoch": 0.05867062755291497,
      "grad_norm": 0.24646006524562836,
      "learning_rate": 2.230421052631579e-05,
      "loss": 10.7129,
      "step": 158
    },
    {
      "epoch": 0.05904196063869291,
      "grad_norm": 0.2982631027698517,
      "learning_rate": 2.1773157894736843e-05,
      "loss": 10.7029,
      "step": 159
    },
    {
      "epoch": 0.05941329372447085,
      "grad_norm": 0.23823799192905426,
      "learning_rate": 2.1242105263157895e-05,
      "loss": 10.7194,
      "step": 160
    },
    {
      "epoch": 0.05978462681024879,
      "grad_norm": 0.30195873975753784,
      "learning_rate": 2.0711052631578947e-05,
      "loss": 10.7269,
      "step": 161
    },
    {
      "epoch": 0.060155959896026734,
      "grad_norm": 0.2737445533275604,
      "learning_rate": 2.018e-05,
      "loss": 10.7119,
      "step": 162
    },
    {
      "epoch": 0.060527292981804676,
      "grad_norm": 0.29084184765815735,
      "learning_rate": 1.9648947368421052e-05,
      "loss": 10.739,
      "step": 163
    },
    {
      "epoch": 0.060898626067582624,
      "grad_norm": 0.365295946598053,
      "learning_rate": 1.9117894736842105e-05,
      "loss": 10.7013,
      "step": 164
    },
    {
      "epoch": 0.061269959153360566,
      "grad_norm": 0.31969207525253296,
      "learning_rate": 1.8586842105263157e-05,
      "loss": 10.7008,
      "step": 165
    },
    {
      "epoch": 0.06164129223913851,
      "grad_norm": 0.2848905026912689,
      "learning_rate": 1.805578947368421e-05,
      "loss": 10.7239,
      "step": 166
    },
    {
      "epoch": 0.06201262532491645,
      "grad_norm": 0.3124282658100128,
      "learning_rate": 1.7524736842105266e-05,
      "loss": 10.7078,
      "step": 167
    },
    {
      "epoch": 0.06238395841069439,
      "grad_norm": 0.3112058639526367,
      "learning_rate": 1.6993684210526315e-05,
      "loss": 10.7357,
      "step": 168
    },
    {
      "epoch": 0.06275529149647234,
      "grad_norm": 0.35105863213539124,
      "learning_rate": 1.646263157894737e-05,
      "loss": 10.7035,
      "step": 169
    },
    {
      "epoch": 0.06312662458225028,
      "grad_norm": 0.36087125539779663,
      "learning_rate": 1.593157894736842e-05,
      "loss": 10.7081,
      "step": 170
    },
    {
      "epoch": 0.06349795766802822,
      "grad_norm": 0.3364165127277374,
      "learning_rate": 1.5400526315789475e-05,
      "loss": 10.7028,
      "step": 171
    },
    {
      "epoch": 0.06386929075380617,
      "grad_norm": 0.3719669282436371,
      "learning_rate": 1.4869473684210524e-05,
      "loss": 10.7114,
      "step": 172
    },
    {
      "epoch": 0.06424062383958411,
      "grad_norm": 0.3243717849254608,
      "learning_rate": 1.4338421052631579e-05,
      "loss": 10.7056,
      "step": 173
    },
    {
      "epoch": 0.06461195692536205,
      "grad_norm": 0.34301552176475525,
      "learning_rate": 1.3807368421052633e-05,
      "loss": 10.7298,
      "step": 174
    },
    {
      "epoch": 0.06498329001113999,
      "grad_norm": 0.33205053210258484,
      "learning_rate": 1.3276315789473684e-05,
      "loss": 10.7258,
      "step": 175
    },
    {
      "epoch": 0.06535462309691793,
      "grad_norm": 0.32598814368247986,
      "learning_rate": 1.2745263157894738e-05,
      "loss": 10.7148,
      "step": 176
    },
    {
      "epoch": 0.06572595618269587,
      "grad_norm": 0.3364621698856354,
      "learning_rate": 1.221421052631579e-05,
      "loss": 10.7189,
      "step": 177
    },
    {
      "epoch": 0.06609728926847382,
      "grad_norm": 0.30798718333244324,
      "learning_rate": 1.1683157894736843e-05,
      "loss": 10.7077,
      "step": 178
    },
    {
      "epoch": 0.06646862235425176,
      "grad_norm": 0.45846202969551086,
      "learning_rate": 1.1152105263157895e-05,
      "loss": 10.689,
      "step": 179
    },
    {
      "epoch": 0.06683995544002971,
      "grad_norm": 0.32936862111091614,
      "learning_rate": 1.0621052631578948e-05,
      "loss": 10.7056,
      "step": 180
    },
    {
      "epoch": 0.06721128852580766,
      "grad_norm": 0.4312770664691925,
      "learning_rate": 1.009e-05,
      "loss": 10.7056,
      "step": 181
    },
    {
      "epoch": 0.0675826216115856,
      "grad_norm": 0.33121156692504883,
      "learning_rate": 9.558947368421052e-06,
      "loss": 10.7258,
      "step": 182
    },
    {
      "epoch": 0.06795395469736354,
      "grad_norm": 0.3501925468444824,
      "learning_rate": 9.027894736842105e-06,
      "loss": 10.6905,
      "step": 183
    },
    {
      "epoch": 0.06832528778314148,
      "grad_norm": 0.3568207025527954,
      "learning_rate": 8.496842105263157e-06,
      "loss": 10.7212,
      "step": 184
    },
    {
      "epoch": 0.06869662086891942,
      "grad_norm": 0.334271103143692,
      "learning_rate": 7.96578947368421e-06,
      "loss": 10.7061,
      "step": 185
    },
    {
      "epoch": 0.06906795395469736,
      "grad_norm": 0.37924298644065857,
      "learning_rate": 7.434736842105262e-06,
      "loss": 10.7275,
      "step": 186
    },
    {
      "epoch": 0.0694392870404753,
      "grad_norm": 0.375325471162796,
      "learning_rate": 6.903684210526316e-06,
      "loss": 10.701,
      "step": 187
    },
    {
      "epoch": 0.06981062012625325,
      "grad_norm": 0.3595587909221649,
      "learning_rate": 6.372631578947369e-06,
      "loss": 10.7056,
      "step": 188
    },
    {
      "epoch": 0.07018195321203119,
      "grad_norm": 0.43467050790786743,
      "learning_rate": 5.841578947368421e-06,
      "loss": 10.7071,
      "step": 189
    },
    {
      "epoch": 0.07055328629780913,
      "grad_norm": 0.3964037597179413,
      "learning_rate": 5.310526315789474e-06,
      "loss": 10.7053,
      "step": 190
    },
    {
      "epoch": 0.07092461938358707,
      "grad_norm": 0.4399015009403229,
      "learning_rate": 4.779473684210526e-06,
      "loss": 10.6696,
      "step": 191
    },
    {
      "epoch": 0.07129595246936501,
      "grad_norm": 0.3865761160850525,
      "learning_rate": 4.248421052631579e-06,
      "loss": 10.7024,
      "step": 192
    },
    {
      "epoch": 0.07166728555514296,
      "grad_norm": 0.4131179451942444,
      "learning_rate": 3.717368421052631e-06,
      "loss": 10.6978,
      "step": 193
    },
    {
      "epoch": 0.07203861864092091,
      "grad_norm": 0.4591720700263977,
      "learning_rate": 3.1863157894736844e-06,
      "loss": 10.6869,
      "step": 194
    },
    {
      "epoch": 0.07240995172669885,
      "grad_norm": 0.4387078881263733,
      "learning_rate": 2.655263157894737e-06,
      "loss": 10.6734,
      "step": 195
    },
    {
      "epoch": 0.0727812848124768,
      "grad_norm": 0.5036574006080627,
      "learning_rate": 2.1242105263157893e-06,
      "loss": 10.7041,
      "step": 196
    },
    {
      "epoch": 0.07315261789825474,
      "grad_norm": 0.5212882161140442,
      "learning_rate": 1.5931578947368422e-06,
      "loss": 10.7093,
      "step": 197
    },
    {
      "epoch": 0.07352395098403268,
      "grad_norm": 0.5136047005653381,
      "learning_rate": 1.0621052631578947e-06,
      "loss": 10.6584,
      "step": 198
    },
    {
      "epoch": 0.07389528406981062,
      "grad_norm": 0.6663255095481873,
      "learning_rate": 5.310526315789473e-07,
      "loss": 10.6604,
      "step": 199
    },
    {
      "epoch": 0.07426661715558856,
      "grad_norm": 0.7245345711708069,
      "learning_rate": 0.0,
      "loss": 10.6654,
      "step": 200
    },
    {
      "epoch": 0.07426661715558856,
      "eval_loss": 10.70785140991211,
      "eval_runtime": 8.53,
      "eval_samples_per_second": 132.942,
      "eval_steps_per_second": 33.294,
      "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": 34155056332800.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}