{
  "best_metric": 2.022968292236328,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.008986138880776403,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.493069440388201e-05,
      "grad_norm": 0.9749875664710999,
      "learning_rate": 1.001e-05,
      "loss": 1.9179,
      "step": 1
    },
    {
      "epoch": 4.493069440388201e-05,
      "eval_loss": 2.865260362625122,
      "eval_runtime": 1318.9517,
      "eval_samples_per_second": 7.106,
      "eval_steps_per_second": 1.776,
      "step": 1
    },
    {
      "epoch": 8.986138880776402e-05,
      "grad_norm": 1.453717827796936,
      "learning_rate": 2.002e-05,
      "loss": 2.5071,
      "step": 2
    },
    {
      "epoch": 0.00013479208321164605,
      "grad_norm": 1.8178796768188477,
      "learning_rate": 3.0029999999999995e-05,
      "loss": 3.4818,
      "step": 3
    },
    {
      "epoch": 0.00017972277761552804,
      "grad_norm": 1.5889410972595215,
      "learning_rate": 4.004e-05,
      "loss": 2.6923,
      "step": 4
    },
    {
      "epoch": 0.00022465347201941005,
      "grad_norm": 1.5432227849960327,
      "learning_rate": 5.005e-05,
      "loss": 2.3713,
      "step": 5
    },
    {
      "epoch": 0.0002695841664232921,
      "grad_norm": 1.9612302780151367,
      "learning_rate": 6.005999999999999e-05,
      "loss": 2.5215,
      "step": 6
    },
    {
      "epoch": 0.0003145148608271741,
      "grad_norm": 2.0271384716033936,
      "learning_rate": 7.006999999999998e-05,
      "loss": 2.5916,
      "step": 7
    },
    {
      "epoch": 0.0003594455552310561,
      "grad_norm": 1.8546284437179565,
      "learning_rate": 8.008e-05,
      "loss": 2.799,
      "step": 8
    },
    {
      "epoch": 0.0004043762496349381,
      "grad_norm": 1.9049793481826782,
      "learning_rate": 9.009e-05,
      "loss": 2.7529,
      "step": 9
    },
    {
      "epoch": 0.0004493069440388201,
      "grad_norm": 2.092958688735962,
      "learning_rate": 0.0001001,
      "loss": 2.4417,
      "step": 10
    },
    {
      "epoch": 0.0004942376384427022,
      "grad_norm": 2.2183825969696045,
      "learning_rate": 9.957315789473684e-05,
      "loss": 2.9033,
      "step": 11
    },
    {
      "epoch": 0.0005391683328465842,
      "grad_norm": 2.383800983428955,
      "learning_rate": 9.904631578947367e-05,
      "loss": 2.8646,
      "step": 12
    },
    {
      "epoch": 0.0005840990272504661,
      "grad_norm": 2.405416250228882,
      "learning_rate": 9.851947368421052e-05,
      "loss": 2.1852,
      "step": 13
    },
    {
      "epoch": 0.0006290297216543482,
      "grad_norm": 2.645289182662964,
      "learning_rate": 9.799263157894736e-05,
      "loss": 2.5829,
      "step": 14
    },
    {
      "epoch": 0.0006739604160582302,
      "grad_norm": 2.2245421409606934,
      "learning_rate": 9.746578947368421e-05,
      "loss": 2.2014,
      "step": 15
    },
    {
      "epoch": 0.0007188911104621122,
      "grad_norm": 2.6380414962768555,
      "learning_rate": 9.693894736842104e-05,
      "loss": 2.8572,
      "step": 16
    },
    {
      "epoch": 0.0007638218048659942,
      "grad_norm": 2.561542510986328,
      "learning_rate": 9.641210526315789e-05,
      "loss": 2.503,
      "step": 17
    },
    {
      "epoch": 0.0008087524992698762,
      "grad_norm": 2.8893611431121826,
      "learning_rate": 9.588526315789473e-05,
      "loss": 2.7106,
      "step": 18
    },
    {
      "epoch": 0.0008536831936737582,
      "grad_norm": 2.734529495239258,
      "learning_rate": 9.535842105263157e-05,
      "loss": 2.6329,
      "step": 19
    },
    {
      "epoch": 0.0008986138880776402,
      "grad_norm": 2.347940683364868,
      "learning_rate": 9.483157894736841e-05,
      "loss": 2.2582,
      "step": 20
    },
    {
      "epoch": 0.0009435445824815223,
      "grad_norm": 2.2515718936920166,
      "learning_rate": 9.430473684210526e-05,
      "loss": 2.3672,
      "step": 21
    },
    {
      "epoch": 0.0009884752768854043,
      "grad_norm": 2.001131534576416,
      "learning_rate": 9.37778947368421e-05,
      "loss": 2.0849,
      "step": 22
    },
    {
      "epoch": 0.0010334059712892862,
      "grad_norm": 2.1524012088775635,
      "learning_rate": 9.325105263157894e-05,
      "loss": 1.9876,
      "step": 23
    },
    {
      "epoch": 0.0010783366656931684,
      "grad_norm": 2.1994237899780273,
      "learning_rate": 9.272421052631578e-05,
      "loss": 2.1119,
      "step": 24
    },
    {
      "epoch": 0.0011232673600970503,
      "grad_norm": 2.1171867847442627,
      "learning_rate": 9.219736842105263e-05,
      "loss": 2.1807,
      "step": 25
    },
    {
      "epoch": 0.0011681980545009323,
      "grad_norm": 2.242772102355957,
      "learning_rate": 9.167052631578946e-05,
      "loss": 2.6407,
      "step": 26
    },
    {
      "epoch": 0.0012131287489048144,
      "grad_norm": 2.039992570877075,
      "learning_rate": 9.114368421052632e-05,
      "loss": 2.214,
      "step": 27
    },
    {
      "epoch": 0.0012580594433086963,
      "grad_norm": 1.9102648496627808,
      "learning_rate": 9.061684210526315e-05,
      "loss": 1.8997,
      "step": 28
    },
    {
      "epoch": 0.0013029901377125783,
      "grad_norm": 1.8196017742156982,
      "learning_rate": 9.009e-05,
      "loss": 2.0056,
      "step": 29
    },
    {
      "epoch": 0.0013479208321164604,
      "grad_norm": 2.0965042114257812,
      "learning_rate": 8.956315789473683e-05,
      "loss": 2.0932,
      "step": 30
    },
    {
      "epoch": 0.0013928515265203424,
      "grad_norm": 1.8538213968276978,
      "learning_rate": 8.903631578947368e-05,
      "loss": 2.0703,
      "step": 31
    },
    {
      "epoch": 0.0014377822209242243,
      "grad_norm": 1.6811614036560059,
      "learning_rate": 8.850947368421052e-05,
      "loss": 2.0772,
      "step": 32
    },
    {
      "epoch": 0.0014827129153281065,
      "grad_norm": 2.006547212600708,
      "learning_rate": 8.798263157894736e-05,
      "loss": 2.0949,
      "step": 33
    },
    {
      "epoch": 0.0015276436097319884,
      "grad_norm": 2.2675957679748535,
      "learning_rate": 8.745578947368422e-05,
      "loss": 2.1614,
      "step": 34
    },
    {
      "epoch": 0.0015725743041358703,
      "grad_norm": 2.1679954528808594,
      "learning_rate": 8.692894736842105e-05,
      "loss": 2.4816,
      "step": 35
    },
    {
      "epoch": 0.0016175049985397525,
      "grad_norm": 2.089261531829834,
      "learning_rate": 8.64021052631579e-05,
      "loss": 2.1532,
      "step": 36
    },
    {
      "epoch": 0.0016624356929436344,
      "grad_norm": 2.0510172843933105,
      "learning_rate": 8.587526315789473e-05,
      "loss": 2.2617,
      "step": 37
    },
    {
      "epoch": 0.0017073663873475164,
      "grad_norm": 2.3127708435058594,
      "learning_rate": 8.534842105263157e-05,
      "loss": 2.6826,
      "step": 38
    },
    {
      "epoch": 0.0017522970817513985,
      "grad_norm": 2.004573345184326,
      "learning_rate": 8.482157894736842e-05,
      "loss": 2.2024,
      "step": 39
    },
    {
      "epoch": 0.0017972277761552804,
      "grad_norm": 1.9959832429885864,
      "learning_rate": 8.429473684210525e-05,
      "loss": 2.3131,
      "step": 40
    },
    {
      "epoch": 0.0018421584705591626,
      "grad_norm": 2.010679244995117,
      "learning_rate": 8.376789473684211e-05,
      "loss": 2.6558,
      "step": 41
    },
    {
      "epoch": 0.0018870891649630445,
      "grad_norm": 1.8978559970855713,
      "learning_rate": 8.324105263157894e-05,
      "loss": 1.7133,
      "step": 42
    },
    {
      "epoch": 0.0019320198593669265,
      "grad_norm": 1.9281487464904785,
      "learning_rate": 8.271421052631579e-05,
      "loss": 1.6634,
      "step": 43
    },
    {
      "epoch": 0.0019769505537708086,
      "grad_norm": 2.3254244327545166,
      "learning_rate": 8.218736842105262e-05,
      "loss": 2.209,
      "step": 44
    },
    {
      "epoch": 0.0020218812481746905,
      "grad_norm": 2.354820728302002,
      "learning_rate": 8.166052631578947e-05,
      "loss": 2.0182,
      "step": 45
    },
    {
      "epoch": 0.0020668119425785725,
      "grad_norm": 2.071244239807129,
      "learning_rate": 8.113368421052631e-05,
      "loss": 1.9602,
      "step": 46
    },
    {
      "epoch": 0.0021117426369824544,
      "grad_norm": 1.8261961936950684,
      "learning_rate": 8.060684210526315e-05,
      "loss": 1.7813,
      "step": 47
    },
    {
      "epoch": 0.002156673331386337,
      "grad_norm": 2.7406086921691895,
      "learning_rate": 8.008e-05,
      "loss": 2.3802,
      "step": 48
    },
    {
      "epoch": 0.0022016040257902187,
      "grad_norm": 2.408278703689575,
      "learning_rate": 7.955315789473684e-05,
      "loss": 2.3663,
      "step": 49
    },
    {
      "epoch": 0.0022465347201941007,
      "grad_norm": 2.055316925048828,
      "learning_rate": 7.902631578947368e-05,
      "loss": 2.3711,
      "step": 50
    },
    {
      "epoch": 0.0022465347201941007,
      "eval_loss": 2.1835074424743652,
      "eval_runtime": 1324.1769,
      "eval_samples_per_second": 7.078,
      "eval_steps_per_second": 1.769,
      "step": 50
    },
    {
      "epoch": 0.0022914654145979826,
      "grad_norm": 1.5457913875579834,
      "learning_rate": 7.849947368421052e-05,
      "loss": 2.3197,
      "step": 51
    },
    {
      "epoch": 0.0023363961090018645,
      "grad_norm": 1.7222747802734375,
      "learning_rate": 7.797263157894736e-05,
      "loss": 2.3836,
      "step": 52
    },
    {
      "epoch": 0.0023813268034057465,
      "grad_norm": 1.8542462587356567,
      "learning_rate": 7.744578947368421e-05,
      "loss": 2.2357,
      "step": 53
    },
    {
      "epoch": 0.002426257497809629,
      "grad_norm": 1.869113802909851,
      "learning_rate": 7.691894736842104e-05,
      "loss": 2.6175,
      "step": 54
    },
    {
      "epoch": 0.0024711881922135108,
      "grad_norm": 1.865387201309204,
      "learning_rate": 7.63921052631579e-05,
      "loss": 2.1006,
      "step": 55
    },
    {
      "epoch": 0.0025161188866173927,
      "grad_norm": 1.756447196006775,
      "learning_rate": 7.586526315789473e-05,
      "loss": 2.1646,
      "step": 56
    },
    {
      "epoch": 0.0025610495810212746,
      "grad_norm": 1.8337773084640503,
      "learning_rate": 7.533842105263158e-05,
      "loss": 2.1912,
      "step": 57
    },
    {
      "epoch": 0.0026059802754251566,
      "grad_norm": 2.07173228263855,
      "learning_rate": 7.481157894736841e-05,
      "loss": 2.5571,
      "step": 58
    },
    {
      "epoch": 0.0026509109698290385,
      "grad_norm": 1.9832768440246582,
      "learning_rate": 7.428473684210526e-05,
      "loss": 2.3895,
      "step": 59
    },
    {
      "epoch": 0.002695841664232921,
      "grad_norm": 1.8779370784759521,
      "learning_rate": 7.375789473684209e-05,
      "loss": 1.9904,
      "step": 60
    },
    {
      "epoch": 0.002740772358636803,
      "grad_norm": 2.2189080715179443,
      "learning_rate": 7.323105263157895e-05,
      "loss": 2.479,
      "step": 61
    },
    {
      "epoch": 0.0027857030530406847,
      "grad_norm": 2.1359362602233887,
      "learning_rate": 7.270421052631578e-05,
      "loss": 2.239,
      "step": 62
    },
    {
      "epoch": 0.0028306337474445667,
      "grad_norm": 2.0250144004821777,
      "learning_rate": 7.217736842105263e-05,
      "loss": 2.3436,
      "step": 63
    },
    {
      "epoch": 0.0028755644418484486,
      "grad_norm": 2.2144904136657715,
      "learning_rate": 7.165052631578947e-05,
      "loss": 2.0083,
      "step": 64
    },
    {
      "epoch": 0.002920495136252331,
      "grad_norm": 2.109282970428467,
      "learning_rate": 7.11236842105263e-05,
      "loss": 2.2343,
      "step": 65
    },
    {
      "epoch": 0.002965425830656213,
      "grad_norm": 1.87382173538208,
      "learning_rate": 7.059684210526315e-05,
      "loss": 2.4058,
      "step": 66
    },
    {
      "epoch": 0.003010356525060095,
      "grad_norm": 2.178577184677124,
      "learning_rate": 7.006999999999998e-05,
      "loss": 1.5616,
      "step": 67
    },
    {
      "epoch": 0.003055287219463977,
      "grad_norm": 2.0847856998443604,
      "learning_rate": 6.954315789473684e-05,
      "loss": 2.1887,
      "step": 68
    },
    {
      "epoch": 0.0031002179138678587,
      "grad_norm": 2.1739280223846436,
      "learning_rate": 6.901631578947368e-05,
      "loss": 2.2799,
      "step": 69
    },
    {
      "epoch": 0.0031451486082717407,
      "grad_norm": 1.968297004699707,
      "learning_rate": 6.848947368421052e-05,
      "loss": 2.2577,
      "step": 70
    },
    {
      "epoch": 0.003190079302675623,
      "grad_norm": 2.0107929706573486,
      "learning_rate": 6.796263157894737e-05,
      "loss": 2.202,
      "step": 71
    },
    {
      "epoch": 0.003235009997079505,
      "grad_norm": 1.9575681686401367,
      "learning_rate": 6.74357894736842e-05,
      "loss": 2.0874,
      "step": 72
    },
    {
      "epoch": 0.003279940691483387,
      "grad_norm": 1.8866405487060547,
      "learning_rate": 6.690894736842105e-05,
      "loss": 1.9911,
      "step": 73
    },
    {
      "epoch": 0.003324871385887269,
      "grad_norm": 1.875723123550415,
      "learning_rate": 6.638210526315788e-05,
      "loss": 1.9437,
      "step": 74
    },
    {
      "epoch": 0.0033698020802911508,
      "grad_norm": 1.9436726570129395,
      "learning_rate": 6.585526315789474e-05,
      "loss": 2.4976,
      "step": 75
    },
    {
      "epoch": 0.0034147327746950327,
      "grad_norm": 1.8745689392089844,
      "learning_rate": 6.532842105263157e-05,
      "loss": 1.9435,
      "step": 76
    },
    {
      "epoch": 0.003459663469098915,
      "grad_norm": 2.6744322776794434,
      "learning_rate": 6.480157894736842e-05,
      "loss": 1.9938,
      "step": 77
    },
    {
      "epoch": 0.003504594163502797,
      "grad_norm": 2.346561908721924,
      "learning_rate": 6.427473684210526e-05,
      "loss": 2.2839,
      "step": 78
    },
    {
      "epoch": 0.003549524857906679,
      "grad_norm": 1.8465696573257446,
      "learning_rate": 6.37478947368421e-05,
      "loss": 2.0044,
      "step": 79
    },
    {
      "epoch": 0.003594455552310561,
      "grad_norm": 2.5786619186401367,
      "learning_rate": 6.322105263157894e-05,
      "loss": 1.8605,
      "step": 80
    },
    {
      "epoch": 0.003639386246714443,
      "grad_norm": 2.4463517665863037,
      "learning_rate": 6.269421052631577e-05,
      "loss": 2.2545,
      "step": 81
    },
    {
      "epoch": 0.003684316941118325,
      "grad_norm": 2.071136474609375,
      "learning_rate": 6.216736842105263e-05,
      "loss": 2.1263,
      "step": 82
    },
    {
      "epoch": 0.003729247635522207,
      "grad_norm": 1.967814326286316,
      "learning_rate": 6.164052631578947e-05,
      "loss": 1.865,
      "step": 83
    },
    {
      "epoch": 0.003774178329926089,
      "grad_norm": 2.2158045768737793,
      "learning_rate": 6.111368421052631e-05,
      "loss": 2.2745,
      "step": 84
    },
    {
      "epoch": 0.003819109024329971,
      "grad_norm": 2.4370172023773193,
      "learning_rate": 6.058684210526315e-05,
      "loss": 2.4499,
      "step": 85
    },
    {
      "epoch": 0.003864039718733853,
      "grad_norm": 2.521904945373535,
      "learning_rate": 6.005999999999999e-05,
      "loss": 1.9882,
      "step": 86
    },
    {
      "epoch": 0.003908970413137735,
      "grad_norm": 2.680880546569824,
      "learning_rate": 5.953315789473684e-05,
      "loss": 2.1553,
      "step": 87
    },
    {
      "epoch": 0.003953901107541617,
      "grad_norm": 2.1294827461242676,
      "learning_rate": 5.9006315789473676e-05,
      "loss": 2.0562,
      "step": 88
    },
    {
      "epoch": 0.003998831801945499,
      "grad_norm": 2.2988648414611816,
      "learning_rate": 5.847947368421053e-05,
      "loss": 2.3066,
      "step": 89
    },
    {
      "epoch": 0.004043762496349381,
      "grad_norm": 2.1352288722991943,
      "learning_rate": 5.795263157894737e-05,
      "loss": 2.004,
      "step": 90
    },
    {
      "epoch": 0.004088693190753263,
      "grad_norm": 2.0327036380767822,
      "learning_rate": 5.742578947368421e-05,
      "loss": 2.0546,
      "step": 91
    },
    {
      "epoch": 0.004133623885157145,
      "grad_norm": 2.385735511779785,
      "learning_rate": 5.6898947368421046e-05,
      "loss": 1.8421,
      "step": 92
    },
    {
      "epoch": 0.004178554579561027,
      "grad_norm": 2.20995831489563,
      "learning_rate": 5.6372105263157886e-05,
      "loss": 2.0368,
      "step": 93
    },
    {
      "epoch": 0.004223485273964909,
      "grad_norm": 2.3523311614990234,
      "learning_rate": 5.584526315789473e-05,
      "loss": 2.2827,
      "step": 94
    },
    {
      "epoch": 0.004268415968368791,
      "grad_norm": 2.2304766178131104,
      "learning_rate": 5.531842105263158e-05,
      "loss": 1.9297,
      "step": 95
    },
    {
      "epoch": 0.004313346662772674,
      "grad_norm": 2.3118209838867188,
      "learning_rate": 5.4791578947368424e-05,
      "loss": 2.0775,
      "step": 96
    },
    {
      "epoch": 0.0043582773571765555,
      "grad_norm": 2.2064402103424072,
      "learning_rate": 5.426473684210526e-05,
      "loss": 2.0467,
      "step": 97
    },
    {
      "epoch": 0.0044032080515804374,
      "grad_norm": 2.449071168899536,
      "learning_rate": 5.37378947368421e-05,
      "loss": 2.2515,
      "step": 98
    },
    {
      "epoch": 0.004448138745984319,
      "grad_norm": 2.0371315479278564,
      "learning_rate": 5.321105263157894e-05,
      "loss": 1.7399,
      "step": 99
    },
    {
      "epoch": 0.004493069440388201,
      "grad_norm": 2.121981143951416,
      "learning_rate": 5.268421052631578e-05,
      "loss": 2.5136,
      "step": 100
    },
    {
      "epoch": 0.004493069440388201,
      "eval_loss": 2.089129686355591,
      "eval_runtime": 1321.9647,
      "eval_samples_per_second": 7.089,
      "eval_steps_per_second": 1.772,
      "step": 100
    },
    {
      "epoch": 0.004538000134792083,
      "grad_norm": 1.5324777364730835,
      "learning_rate": 5.2157368421052626e-05,
      "loss": 1.8549,
      "step": 101
    },
    {
      "epoch": 0.004582930829195965,
      "grad_norm": 1.585769534111023,
      "learning_rate": 5.163052631578947e-05,
      "loss": 2.0778,
      "step": 102
    },
    {
      "epoch": 0.004627861523599847,
      "grad_norm": 1.5087424516677856,
      "learning_rate": 5.110368421052632e-05,
      "loss": 1.7318,
      "step": 103
    },
    {
      "epoch": 0.004672792218003729,
      "grad_norm": 1.8027831315994263,
      "learning_rate": 5.057684210526316e-05,
      "loss": 2.5847,
      "step": 104
    },
    {
      "epoch": 0.004717722912407611,
      "grad_norm": 1.7983766794204712,
      "learning_rate": 5.005e-05,
      "loss": 2.3703,
      "step": 105
    },
    {
      "epoch": 0.004762653606811493,
      "grad_norm": 2.110668659210205,
      "learning_rate": 4.9523157894736836e-05,
      "loss": 2.284,
      "step": 106
    },
    {
      "epoch": 0.004807584301215375,
      "grad_norm": 1.8841075897216797,
      "learning_rate": 4.899631578947368e-05,
      "loss": 2.0461,
      "step": 107
    },
    {
      "epoch": 0.004852514995619258,
      "grad_norm": 1.998826265335083,
      "learning_rate": 4.846947368421052e-05,
      "loss": 2.2872,
      "step": 108
    },
    {
      "epoch": 0.00489744569002314,
      "grad_norm": 2.051478862762451,
      "learning_rate": 4.794263157894737e-05,
      "loss": 2.3491,
      "step": 109
    },
    {
      "epoch": 0.0049423763844270215,
      "grad_norm": 2.1531167030334473,
      "learning_rate": 4.7415789473684206e-05,
      "loss": 2.0238,
      "step": 110
    },
    {
      "epoch": 0.0049873070788309035,
      "grad_norm": 2.3136911392211914,
      "learning_rate": 4.688894736842105e-05,
      "loss": 2.0826,
      "step": 111
    },
    {
      "epoch": 0.005032237773234785,
      "grad_norm": 2.1972248554229736,
      "learning_rate": 4.636210526315789e-05,
      "loss": 2.394,
      "step": 112
    },
    {
      "epoch": 0.005077168467638667,
      "grad_norm": 2.2175815105438232,
      "learning_rate": 4.583526315789473e-05,
      "loss": 1.8484,
      "step": 113
    },
    {
      "epoch": 0.005122099162042549,
      "grad_norm": 2.4054577350616455,
      "learning_rate": 4.530842105263158e-05,
      "loss": 2.3123,
      "step": 114
    },
    {
      "epoch": 0.005167029856446431,
      "grad_norm": 2.4042704105377197,
      "learning_rate": 4.4781578947368416e-05,
      "loss": 1.8743,
      "step": 115
    },
    {
      "epoch": 0.005211960550850313,
      "grad_norm": 2.772090435028076,
      "learning_rate": 4.425473684210526e-05,
      "loss": 2.1385,
      "step": 116
    },
    {
      "epoch": 0.005256891245254195,
      "grad_norm": 2.160134792327881,
      "learning_rate": 4.372789473684211e-05,
      "loss": 1.9255,
      "step": 117
    },
    {
      "epoch": 0.005301821939658077,
      "grad_norm": 2.093456745147705,
      "learning_rate": 4.320105263157895e-05,
      "loss": 2.3033,
      "step": 118
    },
    {
      "epoch": 0.00534675263406196,
      "grad_norm": 1.9034905433654785,
      "learning_rate": 4.2674210526315786e-05,
      "loss": 2.052,
      "step": 119
    },
    {
      "epoch": 0.005391683328465842,
      "grad_norm": 2.014765739440918,
      "learning_rate": 4.2147368421052626e-05,
      "loss": 1.991,
      "step": 120
    },
    {
      "epoch": 0.005436614022869724,
      "grad_norm": 2.068211555480957,
      "learning_rate": 4.162052631578947e-05,
      "loss": 2.2685,
      "step": 121
    },
    {
      "epoch": 0.005481544717273606,
      "grad_norm": 2.1787354946136475,
      "learning_rate": 4.109368421052631e-05,
      "loss": 2.2642,
      "step": 122
    },
    {
      "epoch": 0.0055264754116774875,
      "grad_norm": 1.9796912670135498,
      "learning_rate": 4.056684210526316e-05,
      "loss": 1.7448,
      "step": 123
    },
    {
      "epoch": 0.0055714061060813695,
      "grad_norm": 2.124720573425293,
      "learning_rate": 4.004e-05,
      "loss": 1.854,
      "step": 124
    },
    {
      "epoch": 0.005616336800485251,
      "grad_norm": 2.0373129844665527,
      "learning_rate": 3.951315789473684e-05,
      "loss": 1.8925,
      "step": 125
    },
    {
      "epoch": 0.005661267494889133,
      "grad_norm": 1.8155055046081543,
      "learning_rate": 3.898631578947368e-05,
      "loss": 1.8812,
      "step": 126
    },
    {
      "epoch": 0.005706198189293015,
      "grad_norm": 1.8750555515289307,
      "learning_rate": 3.845947368421052e-05,
      "loss": 2.3017,
      "step": 127
    },
    {
      "epoch": 0.005751128883696897,
      "grad_norm": 1.9103080034255981,
      "learning_rate": 3.7932631578947367e-05,
      "loss": 1.6344,
      "step": 128
    },
    {
      "epoch": 0.005796059578100779,
      "grad_norm": 2.2333719730377197,
      "learning_rate": 3.7405789473684206e-05,
      "loss": 2.137,
      "step": 129
    },
    {
      "epoch": 0.005840990272504662,
      "grad_norm": 1.7498968839645386,
      "learning_rate": 3.6878947368421045e-05,
      "loss": 1.7955,
      "step": 130
    },
    {
      "epoch": 0.005885920966908544,
      "grad_norm": 2.310913562774658,
      "learning_rate": 3.635210526315789e-05,
      "loss": 2.0951,
      "step": 131
    },
    {
      "epoch": 0.005930851661312426,
      "grad_norm": 1.861606478691101,
      "learning_rate": 3.582526315789474e-05,
      "loss": 1.9386,
      "step": 132
    },
    {
      "epoch": 0.005975782355716308,
      "grad_norm": 2.2353315353393555,
      "learning_rate": 3.5298421052631576e-05,
      "loss": 2.0958,
      "step": 133
    },
    {
      "epoch": 0.00602071305012019,
      "grad_norm": 2.313389778137207,
      "learning_rate": 3.477157894736842e-05,
      "loss": 1.9005,
      "step": 134
    },
    {
      "epoch": 0.006065643744524072,
      "grad_norm": 1.824509859085083,
      "learning_rate": 3.424473684210526e-05,
      "loss": 2.0138,
      "step": 135
    },
    {
      "epoch": 0.006110574438927954,
      "grad_norm": 1.9519466161727905,
      "learning_rate": 3.37178947368421e-05,
      "loss": 1.8819,
      "step": 136
    },
    {
      "epoch": 0.0061555051333318355,
      "grad_norm": 1.961268424987793,
      "learning_rate": 3.319105263157894e-05,
      "loss": 1.7123,
      "step": 137
    },
    {
      "epoch": 0.0062004358277357174,
      "grad_norm": 2.225808620452881,
      "learning_rate": 3.2664210526315786e-05,
      "loss": 2.0749,
      "step": 138
    },
    {
      "epoch": 0.006245366522139599,
      "grad_norm": 2.0264053344726562,
      "learning_rate": 3.213736842105263e-05,
      "loss": 1.9238,
      "step": 139
    },
    {
      "epoch": 0.006290297216543481,
      "grad_norm": 1.8889691829681396,
      "learning_rate": 3.161052631578947e-05,
      "loss": 1.7536,
      "step": 140
    },
    {
      "epoch": 0.006335227910947364,
      "grad_norm": 2.0300066471099854,
      "learning_rate": 3.108368421052632e-05,
      "loss": 1.744,
      "step": 141
    },
    {
      "epoch": 0.006380158605351246,
      "grad_norm": 1.9760944843292236,
      "learning_rate": 3.0556842105263156e-05,
      "loss": 2.233,
      "step": 142
    },
    {
      "epoch": 0.006425089299755128,
      "grad_norm": 2.7234182357788086,
      "learning_rate": 3.0029999999999995e-05,
      "loss": 2.2925,
      "step": 143
    },
    {
      "epoch": 0.00647001999415901,
      "grad_norm": 2.251617670059204,
      "learning_rate": 2.9503157894736838e-05,
      "loss": 1.4938,
      "step": 144
    },
    {
      "epoch": 0.006514950688562892,
      "grad_norm": 2.431299924850464,
      "learning_rate": 2.8976315789473684e-05,
      "loss": 1.942,
      "step": 145
    },
    {
      "epoch": 0.006559881382966774,
      "grad_norm": 2.0769991874694824,
      "learning_rate": 2.8449473684210523e-05,
      "loss": 1.6405,
      "step": 146
    },
    {
      "epoch": 0.006604812077370656,
      "grad_norm": 2.384251594543457,
      "learning_rate": 2.7922631578947366e-05,
      "loss": 2.2737,
      "step": 147
    },
    {
      "epoch": 0.006649742771774538,
      "grad_norm": 2.013692617416382,
      "learning_rate": 2.7395789473684212e-05,
      "loss": 1.97,
      "step": 148
    },
    {
      "epoch": 0.00669467346617842,
      "grad_norm": 2.7628982067108154,
      "learning_rate": 2.686894736842105e-05,
      "loss": 2.3824,
      "step": 149
    },
    {
      "epoch": 0.0067396041605823015,
      "grad_norm": 2.5335357189178467,
      "learning_rate": 2.634210526315789e-05,
      "loss": 2.0715,
      "step": 150
    },
    {
      "epoch": 0.0067396041605823015,
      "eval_loss": 2.0427417755126953,
      "eval_runtime": 1323.5526,
      "eval_samples_per_second": 7.081,
      "eval_steps_per_second": 1.77,
      "step": 150
    },
    {
      "epoch": 0.0067845348549861835,
      "grad_norm": 1.2939319610595703,
      "learning_rate": 2.5815263157894736e-05,
      "loss": 1.6482,
      "step": 151
    },
    {
      "epoch": 0.006829465549390065,
      "grad_norm": 1.5967990159988403,
      "learning_rate": 2.528842105263158e-05,
      "loss": 2.0326,
      "step": 152
    },
    {
      "epoch": 0.006874396243793948,
      "grad_norm": 1.8107272386550903,
      "learning_rate": 2.4761578947368418e-05,
      "loss": 2.2003,
      "step": 153
    },
    {
      "epoch": 0.00691932693819783,
      "grad_norm": 1.8108916282653809,
      "learning_rate": 2.423473684210526e-05,
      "loss": 1.9096,
      "step": 154
    },
    {
      "epoch": 0.006964257632601712,
      "grad_norm": 1.7029316425323486,
      "learning_rate": 2.3707894736842103e-05,
      "loss": 2.1272,
      "step": 155
    },
    {
      "epoch": 0.007009188327005594,
      "grad_norm": 2.031301259994507,
      "learning_rate": 2.3181052631578946e-05,
      "loss": 2.2461,
      "step": 156
    },
    {
      "epoch": 0.007054119021409476,
      "grad_norm": 1.6262093782424927,
      "learning_rate": 2.265421052631579e-05,
      "loss": 1.7345,
      "step": 157
    },
    {
      "epoch": 0.007099049715813358,
      "grad_norm": 2.0770440101623535,
      "learning_rate": 2.212736842105263e-05,
      "loss": 2.322,
      "step": 158
    },
    {
      "epoch": 0.00714398041021724,
      "grad_norm": 1.9273806810379028,
      "learning_rate": 2.1600526315789474e-05,
      "loss": 1.8112,
      "step": 159
    },
    {
      "epoch": 0.007188911104621122,
      "grad_norm": 2.2276177406311035,
      "learning_rate": 2.1073684210526313e-05,
      "loss": 2.0591,
      "step": 160
    },
    {
      "epoch": 0.007233841799025004,
      "grad_norm": 2.3354694843292236,
      "learning_rate": 2.0546842105263155e-05,
      "loss": 2.1254,
      "step": 161
    },
    {
      "epoch": 0.007278772493428886,
      "grad_norm": 2.806878089904785,
      "learning_rate": 2.002e-05,
      "loss": 2.4075,
      "step": 162
    },
    {
      "epoch": 0.0073237031878327676,
      "grad_norm": 2.16422176361084,
      "learning_rate": 1.949315789473684e-05,
      "loss": 2.1182,
      "step": 163
    },
    {
      "epoch": 0.00736863388223665,
      "grad_norm": 2.8581697940826416,
      "learning_rate": 1.8966315789473683e-05,
      "loss": 2.3394,
      "step": 164
    },
    {
      "epoch": 0.007413564576640532,
      "grad_norm": 2.2438290119171143,
      "learning_rate": 1.8439473684210522e-05,
      "loss": 1.9132,
      "step": 165
    },
    {
      "epoch": 0.007458495271044414,
      "grad_norm": 2.672802448272705,
      "learning_rate": 1.791263157894737e-05,
      "loss": 2.6549,
      "step": 166
    },
    {
      "epoch": 0.007503425965448296,
      "grad_norm": 2.280754804611206,
      "learning_rate": 1.738578947368421e-05,
      "loss": 2.3295,
      "step": 167
    },
    {
      "epoch": 0.007548356659852178,
      "grad_norm": 1.8638712167739868,
      "learning_rate": 1.685894736842105e-05,
      "loss": 1.9558,
      "step": 168
    },
    {
      "epoch": 0.00759328735425606,
      "grad_norm": 1.9110428094863892,
      "learning_rate": 1.6332105263157893e-05,
      "loss": 2.0123,
      "step": 169
    },
    {
      "epoch": 0.007638218048659942,
      "grad_norm": 2.131638765335083,
      "learning_rate": 1.5805263157894735e-05,
      "loss": 1.4546,
      "step": 170
    },
    {
      "epoch": 0.007683148743063824,
      "grad_norm": 2.094499349594116,
      "learning_rate": 1.5278421052631578e-05,
      "loss": 1.9308,
      "step": 171
    },
    {
      "epoch": 0.007728079437467706,
      "grad_norm": 1.8845446109771729,
      "learning_rate": 1.4751578947368419e-05,
      "loss": 1.6387,
      "step": 172
    },
    {
      "epoch": 0.007773010131871588,
      "grad_norm": 1.6916247606277466,
      "learning_rate": 1.4224736842105262e-05,
      "loss": 1.7056,
      "step": 173
    },
    {
      "epoch": 0.00781794082627547,
      "grad_norm": 1.8374581336975098,
      "learning_rate": 1.3697894736842106e-05,
      "loss": 2.1303,
      "step": 174
    },
    {
      "epoch": 0.007862871520679353,
      "grad_norm": 2.183013916015625,
      "learning_rate": 1.3171052631578945e-05,
      "loss": 1.9973,
      "step": 175
    },
    {
      "epoch": 0.007907802215083234,
      "grad_norm": 2.2911391258239746,
      "learning_rate": 1.264421052631579e-05,
      "loss": 2.202,
      "step": 176
    },
    {
      "epoch": 0.007952732909487116,
      "grad_norm": 2.632253885269165,
      "learning_rate": 1.211736842105263e-05,
      "loss": 2.3454,
      "step": 177
    },
    {
      "epoch": 0.007997663603890998,
      "grad_norm": 2.447774887084961,
      "learning_rate": 1.1590526315789473e-05,
      "loss": 1.9924,
      "step": 178
    },
    {
      "epoch": 0.00804259429829488,
      "grad_norm": 2.108670473098755,
      "learning_rate": 1.1063684210526316e-05,
      "loss": 2.2341,
      "step": 179
    },
    {
      "epoch": 0.008087524992698762,
      "grad_norm": 2.132281541824341,
      "learning_rate": 1.0536842105263156e-05,
      "loss": 1.9931,
      "step": 180
    },
    {
      "epoch": 0.008132455687102644,
      "grad_norm": 1.7290847301483154,
      "learning_rate": 1.001e-05,
      "loss": 1.7781,
      "step": 181
    },
    {
      "epoch": 0.008177386381506526,
      "grad_norm": 2.1073453426361084,
      "learning_rate": 9.483157894736842e-06,
      "loss": 1.6803,
      "step": 182
    },
    {
      "epoch": 0.008222317075910408,
      "grad_norm": 2.2671337127685547,
      "learning_rate": 8.956315789473684e-06,
      "loss": 2.4534,
      "step": 183
    },
    {
      "epoch": 0.00826724777031429,
      "grad_norm": 2.219877243041992,
      "learning_rate": 8.429473684210525e-06,
      "loss": 1.7814,
      "step": 184
    },
    {
      "epoch": 0.008312178464718172,
      "grad_norm": 1.8467822074890137,
      "learning_rate": 7.902631578947368e-06,
      "loss": 1.6241,
      "step": 185
    },
    {
      "epoch": 0.008357109159122054,
      "grad_norm": 2.22536563873291,
      "learning_rate": 7.3757894736842095e-06,
      "loss": 1.6642,
      "step": 186
    },
    {
      "epoch": 0.008402039853525936,
      "grad_norm": 2.0499343872070312,
      "learning_rate": 6.848947368421053e-06,
      "loss": 1.8032,
      "step": 187
    },
    {
      "epoch": 0.008446970547929818,
      "grad_norm": 1.8307327032089233,
      "learning_rate": 6.322105263157895e-06,
      "loss": 1.6625,
      "step": 188
    },
    {
      "epoch": 0.0084919012423337,
      "grad_norm": 2.4157087802886963,
      "learning_rate": 5.7952631578947365e-06,
      "loss": 1.8355,
      "step": 189
    },
    {
      "epoch": 0.008536831936737582,
      "grad_norm": 2.4110615253448486,
      "learning_rate": 5.268421052631578e-06,
      "loss": 1.6817,
      "step": 190
    },
    {
      "epoch": 0.008581762631141463,
      "grad_norm": 2.389373540878296,
      "learning_rate": 4.741578947368421e-06,
      "loss": 1.7818,
      "step": 191
    },
    {
      "epoch": 0.008626693325545347,
      "grad_norm": 1.9679090976715088,
      "learning_rate": 4.2147368421052626e-06,
      "loss": 1.9776,
      "step": 192
    },
    {
      "epoch": 0.008671624019949229,
      "grad_norm": 3.9061481952667236,
      "learning_rate": 3.6878947368421047e-06,
      "loss": 1.8022,
      "step": 193
    },
    {
      "epoch": 0.008716554714353111,
      "grad_norm": 2.2966439723968506,
      "learning_rate": 3.1610526315789474e-06,
      "loss": 2.017,
      "step": 194
    },
    {
      "epoch": 0.008761485408756993,
      "grad_norm": 2.285691738128662,
      "learning_rate": 2.634210526315789e-06,
      "loss": 2.2703,
      "step": 195
    },
    {
      "epoch": 0.008806416103160875,
      "grad_norm": 2.352679967880249,
      "learning_rate": 2.1073684210526313e-06,
      "loss": 1.7985,
      "step": 196
    },
    {
      "epoch": 0.008851346797564757,
      "grad_norm": 2.3461921215057373,
      "learning_rate": 1.5805263157894737e-06,
      "loss": 1.7361,
      "step": 197
    },
    {
      "epoch": 0.008896277491968639,
      "grad_norm": 2.690016508102417,
      "learning_rate": 1.0536842105263156e-06,
      "loss": 2.1687,
      "step": 198
    },
    {
      "epoch": 0.00894120818637252,
      "grad_norm": 2.5040085315704346,
      "learning_rate": 5.268421052631578e-07,
      "loss": 2.4819,
      "step": 199
    },
    {
      "epoch": 0.008986138880776403,
      "grad_norm": 2.3786873817443848,
      "learning_rate": 0.0,
      "loss": 2.7339,
      "step": 200
    },
    {
      "epoch": 0.008986138880776403,
      "eval_loss": 2.022968292236328,
      "eval_runtime": 1321.9224,
      "eval_samples_per_second": 7.09,
      "eval_steps_per_second": 1.772,
      "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": 9.562201408929792e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}