{
  "best_metric": 1.1319868564605713,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.2259248799774075,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0011296243998870376,
      "grad_norm": 0.8550522327423096,
      "learning_rate": 1e-05,
      "loss": 1.2735,
      "step": 1
    },
    {
      "epoch": 0.0011296243998870376,
      "eval_loss": 1.40424382686615,
      "eval_runtime": 127.3742,
      "eval_samples_per_second": 11.706,
      "eval_steps_per_second": 2.928,
      "step": 1
    },
    {
      "epoch": 0.0022592487997740753,
      "grad_norm": 0.9225540161132812,
      "learning_rate": 2e-05,
      "loss": 1.3856,
      "step": 2
    },
    {
      "epoch": 0.003388873199661113,
      "grad_norm": 0.8702269196510315,
      "learning_rate": 3e-05,
      "loss": 1.3373,
      "step": 3
    },
    {
      "epoch": 0.0045184975995481505,
      "grad_norm": 0.8701692223548889,
      "learning_rate": 4e-05,
      "loss": 1.4469,
      "step": 4
    },
    {
      "epoch": 0.005648121999435188,
      "grad_norm": 0.6605684161186218,
      "learning_rate": 5e-05,
      "loss": 1.4126,
      "step": 5
    },
    {
      "epoch": 0.006777746399322226,
      "grad_norm": 0.49637076258659363,
      "learning_rate": 6e-05,
      "loss": 1.1769,
      "step": 6
    },
    {
      "epoch": 0.007907370799209263,
      "grad_norm": 0.9770664572715759,
      "learning_rate": 7e-05,
      "loss": 1.0853,
      "step": 7
    },
    {
      "epoch": 0.009036995199096301,
      "grad_norm": 0.981417715549469,
      "learning_rate": 8e-05,
      "loss": 1.1423,
      "step": 8
    },
    {
      "epoch": 0.010166619598983339,
      "grad_norm": 0.7733097672462463,
      "learning_rate": 9e-05,
      "loss": 1.2347,
      "step": 9
    },
    {
      "epoch": 0.011296243998870376,
      "grad_norm": 0.47179657220840454,
      "learning_rate": 0.0001,
      "loss": 1.0974,
      "step": 10
    },
    {
      "epoch": 0.012425868398757414,
      "grad_norm": 0.397799015045166,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.0695,
      "step": 11
    },
    {
      "epoch": 0.013555492798644452,
      "grad_norm": 0.39556679129600525,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.0873,
      "step": 12
    },
    {
      "epoch": 0.014685117198531489,
      "grad_norm": 0.45747268199920654,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.9936,
      "step": 13
    },
    {
      "epoch": 0.015814741598418527,
      "grad_norm": 0.4742383658885956,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.0999,
      "step": 14
    },
    {
      "epoch": 0.016944365998305563,
      "grad_norm": 0.4812389314174652,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.2281,
      "step": 15
    },
    {
      "epoch": 0.018073990398192602,
      "grad_norm": 0.4086254835128784,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.0491,
      "step": 16
    },
    {
      "epoch": 0.019203614798079638,
      "grad_norm": 0.46562886238098145,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.1335,
      "step": 17
    },
    {
      "epoch": 0.020333239197966677,
      "grad_norm": 0.4353958070278168,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.1902,
      "step": 18
    },
    {
      "epoch": 0.021462863597853713,
      "grad_norm": 0.4419581890106201,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.0544,
      "step": 19
    },
    {
      "epoch": 0.022592487997740753,
      "grad_norm": 0.4766595661640167,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.1409,
      "step": 20
    },
    {
      "epoch": 0.02372211239762779,
      "grad_norm": 0.4358370304107666,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1142,
      "step": 21
    },
    {
      "epoch": 0.024851736797514828,
      "grad_norm": 0.41619443893432617,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.1425,
      "step": 22
    },
    {
      "epoch": 0.025981361197401864,
      "grad_norm": 0.4168614447116852,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.1312,
      "step": 23
    },
    {
      "epoch": 0.027110985597288903,
      "grad_norm": 0.4369964897632599,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.1372,
      "step": 24
    },
    {
      "epoch": 0.02824060999717594,
      "grad_norm": 0.4619179666042328,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.0682,
      "step": 25
    },
    {
      "epoch": 0.029370234397062978,
      "grad_norm": 0.46231937408447266,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.2379,
      "step": 26
    },
    {
      "epoch": 0.030499858796950014,
      "grad_norm": 0.41597914695739746,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.9266,
      "step": 27
    },
    {
      "epoch": 0.031629483196837054,
      "grad_norm": 0.5123385190963745,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.1509,
      "step": 28
    },
    {
      "epoch": 0.03275910759672409,
      "grad_norm": 0.5512084364891052,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.1412,
      "step": 29
    },
    {
      "epoch": 0.033888731996611125,
      "grad_norm": 0.429770290851593,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.1877,
      "step": 30
    },
    {
      "epoch": 0.03501835639649816,
      "grad_norm": 0.47011324763298035,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.1351,
      "step": 31
    },
    {
      "epoch": 0.036147980796385204,
      "grad_norm": 0.4633503258228302,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.0962,
      "step": 32
    },
    {
      "epoch": 0.03727760519627224,
      "grad_norm": 0.47156020998954773,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.1748,
      "step": 33
    },
    {
      "epoch": 0.038407229596159276,
      "grad_norm": 0.46058446168899536,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.2682,
      "step": 34
    },
    {
      "epoch": 0.03953685399604631,
      "grad_norm": 0.4544873833656311,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.1279,
      "step": 35
    },
    {
      "epoch": 0.040666478395933355,
      "grad_norm": 0.500059962272644,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.3243,
      "step": 36
    },
    {
      "epoch": 0.04179610279582039,
      "grad_norm": 0.4791109561920166,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.2054,
      "step": 37
    },
    {
      "epoch": 0.042925727195707426,
      "grad_norm": 0.46742773056030273,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.2421,
      "step": 38
    },
    {
      "epoch": 0.04405535159559446,
      "grad_norm": 0.47654256224632263,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.2031,
      "step": 39
    },
    {
      "epoch": 0.045184975995481505,
      "grad_norm": 0.5024082660675049,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.2832,
      "step": 40
    },
    {
      "epoch": 0.04631460039536854,
      "grad_norm": 0.5106438994407654,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.3643,
      "step": 41
    },
    {
      "epoch": 0.04744422479525558,
      "grad_norm": 0.5419034361839294,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.356,
      "step": 42
    },
    {
      "epoch": 0.04857384919514261,
      "grad_norm": 0.5574733018875122,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.4562,
      "step": 43
    },
    {
      "epoch": 0.049703473595029656,
      "grad_norm": 0.5316269397735596,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.3021,
      "step": 44
    },
    {
      "epoch": 0.05083309799491669,
      "grad_norm": 0.5879673361778259,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.3575,
      "step": 45
    },
    {
      "epoch": 0.05196272239480373,
      "grad_norm": 0.6572689414024353,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.4258,
      "step": 46
    },
    {
      "epoch": 0.05309234679469076,
      "grad_norm": 0.7166992425918579,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4093,
      "step": 47
    },
    {
      "epoch": 0.054221971194577806,
      "grad_norm": 0.738736629486084,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.4699,
      "step": 48
    },
    {
      "epoch": 0.05535159559446484,
      "grad_norm": 0.8827663660049438,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.4088,
      "step": 49
    },
    {
      "epoch": 0.05648121999435188,
      "grad_norm": 1.1569432020187378,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.4324,
      "step": 50
    },
    {
      "epoch": 0.05648121999435188,
      "eval_loss": 1.2543538808822632,
      "eval_runtime": 129.1395,
      "eval_samples_per_second": 11.546,
      "eval_steps_per_second": 2.888,
      "step": 50
    },
    {
      "epoch": 0.057610844394238914,
      "grad_norm": 1.414048194885254,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.1817,
      "step": 51
    },
    {
      "epoch": 0.058740468794125957,
      "grad_norm": 0.8697742223739624,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.2029,
      "step": 52
    },
    {
      "epoch": 0.05987009319401299,
      "grad_norm": 0.43536263704299927,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.2405,
      "step": 53
    },
    {
      "epoch": 0.06099971759390003,
      "grad_norm": 0.37985551357269287,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.3161,
      "step": 54
    },
    {
      "epoch": 0.062129341993787064,
      "grad_norm": 0.41267821192741394,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.3803,
      "step": 55
    },
    {
      "epoch": 0.06325896639367411,
      "grad_norm": 0.3816688656806946,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.2733,
      "step": 56
    },
    {
      "epoch": 0.06438859079356114,
      "grad_norm": 0.348086953163147,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.0677,
      "step": 57
    },
    {
      "epoch": 0.06551821519344818,
      "grad_norm": 0.36134061217308044,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.0529,
      "step": 58
    },
    {
      "epoch": 0.06664783959333521,
      "grad_norm": 0.38435545563697815,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.1269,
      "step": 59
    },
    {
      "epoch": 0.06777746399322225,
      "grad_norm": 0.3482604920864105,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.9619,
      "step": 60
    },
    {
      "epoch": 0.06890708839310929,
      "grad_norm": 0.34352102875709534,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.9393,
      "step": 61
    },
    {
      "epoch": 0.07003671279299632,
      "grad_norm": 0.3560313582420349,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.9764,
      "step": 62
    },
    {
      "epoch": 0.07116633719288337,
      "grad_norm": 0.3476743698120117,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.027,
      "step": 63
    },
    {
      "epoch": 0.07229596159277041,
      "grad_norm": 0.35056135058403015,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.9587,
      "step": 64
    },
    {
      "epoch": 0.07342558599265744,
      "grad_norm": 0.3702610433101654,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.0169,
      "step": 65
    },
    {
      "epoch": 0.07455521039254448,
      "grad_norm": 0.3690338730812073,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.0473,
      "step": 66
    },
    {
      "epoch": 0.07568483479243152,
      "grad_norm": 0.3667239248752594,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.0059,
      "step": 67
    },
    {
      "epoch": 0.07681445919231855,
      "grad_norm": 0.36757180094718933,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.045,
      "step": 68
    },
    {
      "epoch": 0.07794408359220559,
      "grad_norm": 0.386473149061203,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.0232,
      "step": 69
    },
    {
      "epoch": 0.07907370799209262,
      "grad_norm": 0.3790735900402069,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.1096,
      "step": 70
    },
    {
      "epoch": 0.08020333239197967,
      "grad_norm": 0.3658673167228699,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.133,
      "step": 71
    },
    {
      "epoch": 0.08133295679186671,
      "grad_norm": 0.38201308250427246,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.0425,
      "step": 72
    },
    {
      "epoch": 0.08246258119175374,
      "grad_norm": 0.3764412999153137,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.1804,
      "step": 73
    },
    {
      "epoch": 0.08359220559164078,
      "grad_norm": 0.40265172719955444,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.0292,
      "step": 74
    },
    {
      "epoch": 0.08472182999152782,
      "grad_norm": 0.38062238693237305,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.9383,
      "step": 75
    },
    {
      "epoch": 0.08585145439141485,
      "grad_norm": 0.3853003680706024,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.0792,
      "step": 76
    },
    {
      "epoch": 0.08698107879130189,
      "grad_norm": 0.40098947286605835,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.0833,
      "step": 77
    },
    {
      "epoch": 0.08811070319118892,
      "grad_norm": 0.39075568318367004,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.9935,
      "step": 78
    },
    {
      "epoch": 0.08924032759107597,
      "grad_norm": 0.4095045030117035,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1492,
      "step": 79
    },
    {
      "epoch": 0.09036995199096301,
      "grad_norm": 0.41808947920799255,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.0681,
      "step": 80
    },
    {
      "epoch": 0.09149957639085005,
      "grad_norm": 0.4202672839164734,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.0516,
      "step": 81
    },
    {
      "epoch": 0.09262920079073708,
      "grad_norm": 0.4508487582206726,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.1386,
      "step": 82
    },
    {
      "epoch": 0.09375882519062412,
      "grad_norm": 0.42843881249427795,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.0954,
      "step": 83
    },
    {
      "epoch": 0.09488844959051115,
      "grad_norm": 0.46038341522216797,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.2282,
      "step": 84
    },
    {
      "epoch": 0.09601807399039819,
      "grad_norm": 0.4499610960483551,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.2062,
      "step": 85
    },
    {
      "epoch": 0.09714769839028523,
      "grad_norm": 0.4425991177558899,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.2628,
      "step": 86
    },
    {
      "epoch": 0.09827732279017226,
      "grad_norm": 0.4847813546657562,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.2947,
      "step": 87
    },
    {
      "epoch": 0.09940694719005931,
      "grad_norm": 0.46998468041419983,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.2693,
      "step": 88
    },
    {
      "epoch": 0.10053657158994635,
      "grad_norm": 0.5071948766708374,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.2611,
      "step": 89
    },
    {
      "epoch": 0.10166619598983338,
      "grad_norm": 0.5407974123954773,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.2743,
      "step": 90
    },
    {
      "epoch": 0.10279582038972042,
      "grad_norm": 0.5305125713348389,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.3381,
      "step": 91
    },
    {
      "epoch": 0.10392544478960745,
      "grad_norm": 0.541960597038269,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2561,
      "step": 92
    },
    {
      "epoch": 0.10505506918949449,
      "grad_norm": 0.5386022329330444,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.3456,
      "step": 93
    },
    {
      "epoch": 0.10618469358938153,
      "grad_norm": 0.5853845477104187,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.4307,
      "step": 94
    },
    {
      "epoch": 0.10731431798926856,
      "grad_norm": 0.5873990654945374,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.4466,
      "step": 95
    },
    {
      "epoch": 0.10844394238915561,
      "grad_norm": 0.6204474568367004,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.463,
      "step": 96
    },
    {
      "epoch": 0.10957356678904265,
      "grad_norm": 0.6788293123245239,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.384,
      "step": 97
    },
    {
      "epoch": 0.11070319118892968,
      "grad_norm": 0.7152941226959229,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.3518,
      "step": 98
    },
    {
      "epoch": 0.11183281558881672,
      "grad_norm": 0.8938578367233276,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.4256,
      "step": 99
    },
    {
      "epoch": 0.11296243998870376,
      "grad_norm": 0.9971409440040588,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.2386,
      "step": 100
    },
    {
      "epoch": 0.11296243998870376,
      "eval_loss": 1.180202841758728,
      "eval_runtime": 129.0712,
      "eval_samples_per_second": 11.552,
      "eval_steps_per_second": 2.89,
      "step": 100
    },
    {
      "epoch": 0.11409206438859079,
      "grad_norm": 0.685209333896637,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.2389,
      "step": 101
    },
    {
      "epoch": 0.11522168878847783,
      "grad_norm": 0.5941435098648071,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2478,
      "step": 102
    },
    {
      "epoch": 0.11635131318836486,
      "grad_norm": 0.42345911264419556,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.3687,
      "step": 103
    },
    {
      "epoch": 0.11748093758825191,
      "grad_norm": 0.38297444581985474,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.3719,
      "step": 104
    },
    {
      "epoch": 0.11861056198813895,
      "grad_norm": 0.3841160833835602,
      "learning_rate": 5e-05,
      "loss": 1.4525,
      "step": 105
    },
    {
      "epoch": 0.11974018638802598,
      "grad_norm": 0.3808828890323639,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.1691,
      "step": 106
    },
    {
      "epoch": 0.12086981078791302,
      "grad_norm": 0.36592814326286316,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.0353,
      "step": 107
    },
    {
      "epoch": 0.12199943518780006,
      "grad_norm": 0.3603374660015106,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.0364,
      "step": 108
    },
    {
      "epoch": 0.12312905958768709,
      "grad_norm": 0.34239304065704346,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.0466,
      "step": 109
    },
    {
      "epoch": 0.12425868398757413,
      "grad_norm": 0.35895609855651855,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1497,
      "step": 110
    },
    {
      "epoch": 0.12538830838746118,
      "grad_norm": 0.4060536324977875,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.0201,
      "step": 111
    },
    {
      "epoch": 0.12651793278734821,
      "grad_norm": 0.4092485010623932,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.9991,
      "step": 112
    },
    {
      "epoch": 0.12764755718723525,
      "grad_norm": 0.38795047998428345,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.8854,
      "step": 113
    },
    {
      "epoch": 0.12877718158712229,
      "grad_norm": 0.35938042402267456,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.9463,
      "step": 114
    },
    {
      "epoch": 0.12990680598700932,
      "grad_norm": 0.3654451370239258,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.9106,
      "step": 115
    },
    {
      "epoch": 0.13103643038689636,
      "grad_norm": 0.36022642254829407,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.0342,
      "step": 116
    },
    {
      "epoch": 0.1321660547867834,
      "grad_norm": 0.3541191518306732,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.8922,
      "step": 117
    },
    {
      "epoch": 0.13329567918667043,
      "grad_norm": 0.3726769983768463,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.0815,
      "step": 118
    },
    {
      "epoch": 0.13442530358655747,
      "grad_norm": 0.377860963344574,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.8906,
      "step": 119
    },
    {
      "epoch": 0.1355549279864445,
      "grad_norm": 0.37433651089668274,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.007,
      "step": 120
    },
    {
      "epoch": 0.13668455238633154,
      "grad_norm": 0.38958725333213806,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.0222,
      "step": 121
    },
    {
      "epoch": 0.13781417678621857,
      "grad_norm": 0.3903820514678955,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.1088,
      "step": 122
    },
    {
      "epoch": 0.1389438011861056,
      "grad_norm": 0.4269711673259735,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.1574,
      "step": 123
    },
    {
      "epoch": 0.14007342558599264,
      "grad_norm": 0.37161242961883545,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.9996,
      "step": 124
    },
    {
      "epoch": 0.1412030499858797,
      "grad_norm": 0.3638422191143036,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.0562,
      "step": 125
    },
    {
      "epoch": 0.14233267438576674,
      "grad_norm": 0.3776945173740387,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.0835,
      "step": 126
    },
    {
      "epoch": 0.14346229878565378,
      "grad_norm": 0.40536466240882874,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.2252,
      "step": 127
    },
    {
      "epoch": 0.14459192318554082,
      "grad_norm": 0.38575565814971924,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.0098,
      "step": 128
    },
    {
      "epoch": 0.14572154758542785,
      "grad_norm": 0.4363740086555481,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.0363,
      "step": 129
    },
    {
      "epoch": 0.1468511719853149,
      "grad_norm": 0.4551375210285187,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.0869,
      "step": 130
    },
    {
      "epoch": 0.14798079638520192,
      "grad_norm": 0.41762956976890564,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1185,
      "step": 131
    },
    {
      "epoch": 0.14911042078508896,
      "grad_norm": 0.40056923031806946,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.0876,
      "step": 132
    },
    {
      "epoch": 0.150240045184976,
      "grad_norm": 0.41666334867477417,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1312,
      "step": 133
    },
    {
      "epoch": 0.15136966958486303,
      "grad_norm": 0.3918934166431427,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.0351,
      "step": 134
    },
    {
      "epoch": 0.15249929398475007,
      "grad_norm": 0.40713757276535034,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.0452,
      "step": 135
    },
    {
      "epoch": 0.1536289183846371,
      "grad_norm": 0.5111376643180847,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.1983,
      "step": 136
    },
    {
      "epoch": 0.15475854278452414,
      "grad_norm": 0.46056586503982544,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.2397,
      "step": 137
    },
    {
      "epoch": 0.15588816718441117,
      "grad_norm": 0.4692816734313965,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.2454,
      "step": 138
    },
    {
      "epoch": 0.1570177915842982,
      "grad_norm": 0.46887555718421936,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.2822,
      "step": 139
    },
    {
      "epoch": 0.15814741598418525,
      "grad_norm": 0.46481066942214966,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.301,
      "step": 140
    },
    {
      "epoch": 0.15927704038407228,
      "grad_norm": 0.4652707874774933,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.3534,
      "step": 141
    },
    {
      "epoch": 0.16040666478395935,
      "grad_norm": 0.4918733537197113,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.3281,
      "step": 142
    },
    {
      "epoch": 0.16153628918384638,
      "grad_norm": 0.5426083207130432,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.3306,
      "step": 143
    },
    {
      "epoch": 0.16266591358373342,
      "grad_norm": 0.575157880783081,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.3213,
      "step": 144
    },
    {
      "epoch": 0.16379553798362045,
      "grad_norm": 0.5867725014686584,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.4017,
      "step": 145
    },
    {
      "epoch": 0.1649251623835075,
      "grad_norm": 0.6145371198654175,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.4364,
      "step": 146
    },
    {
      "epoch": 0.16605478678339453,
      "grad_norm": 0.6413678526878357,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.4528,
      "step": 147
    },
    {
      "epoch": 0.16718441118328156,
      "grad_norm": 0.6936803460121155,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.3236,
      "step": 148
    },
    {
      "epoch": 0.1683140355831686,
      "grad_norm": 0.820589005947113,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.4062,
      "step": 149
    },
    {
      "epoch": 0.16944365998305563,
      "grad_norm": 1.1195847988128662,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.4636,
      "step": 150
    },
    {
      "epoch": 0.16944365998305563,
      "eval_loss": 1.1399078369140625,
      "eval_runtime": 129.0521,
      "eval_samples_per_second": 11.553,
      "eval_steps_per_second": 2.89,
      "step": 150
    },
    {
      "epoch": 0.17057328438294267,
      "grad_norm": 0.32628974318504333,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.0091,
      "step": 151
    },
    {
      "epoch": 0.1717029087828297,
      "grad_norm": 0.36658379435539246,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.0987,
      "step": 152
    },
    {
      "epoch": 0.17283253318271674,
      "grad_norm": 0.3987427055835724,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.1903,
      "step": 153
    },
    {
      "epoch": 0.17396215758260378,
      "grad_norm": 0.3822471797466278,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.2651,
      "step": 154
    },
    {
      "epoch": 0.1750917819824908,
      "grad_norm": 0.42838650941848755,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4846,
      "step": 155
    },
    {
      "epoch": 0.17622140638237785,
      "grad_norm": 0.3765850067138672,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.1628,
      "step": 156
    },
    {
      "epoch": 0.17735103078226488,
      "grad_norm": 0.4056573808193207,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.9762,
      "step": 157
    },
    {
      "epoch": 0.17848065518215195,
      "grad_norm": 0.3515107035636902,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.0156,
      "step": 158
    },
    {
      "epoch": 0.17961027958203898,
      "grad_norm": 0.3641347289085388,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.9826,
      "step": 159
    },
    {
      "epoch": 0.18073990398192602,
      "grad_norm": 0.3337154686450958,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.8433,
      "step": 160
    },
    {
      "epoch": 0.18186952838181306,
      "grad_norm": 0.3699985444545746,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.9765,
      "step": 161
    },
    {
      "epoch": 0.1829991527817001,
      "grad_norm": 0.3419792950153351,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0475,
      "step": 162
    },
    {
      "epoch": 0.18412877718158713,
      "grad_norm": 0.37137681245803833,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.8738,
      "step": 163
    },
    {
      "epoch": 0.18525840158147416,
      "grad_norm": 0.3740991950035095,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.0206,
      "step": 164
    },
    {
      "epoch": 0.1863880259813612,
      "grad_norm": 0.34303179383277893,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.0116,
      "step": 165
    },
    {
      "epoch": 0.18751765038124824,
      "grad_norm": 0.3419070243835449,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.0343,
      "step": 166
    },
    {
      "epoch": 0.18864727478113527,
      "grad_norm": 0.36161795258522034,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.0867,
      "step": 167
    },
    {
      "epoch": 0.1897768991810223,
      "grad_norm": 0.3431941270828247,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.9489,
      "step": 168
    },
    {
      "epoch": 0.19090652358090934,
      "grad_norm": 0.348385214805603,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.0282,
      "step": 169
    },
    {
      "epoch": 0.19203614798079638,
      "grad_norm": 0.36738184094429016,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.9859,
      "step": 170
    },
    {
      "epoch": 0.19316577238068341,
      "grad_norm": 0.3770311176776886,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.9725,
      "step": 171
    },
    {
      "epoch": 0.19429539678057045,
      "grad_norm": 0.3571227192878723,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.0002,
      "step": 172
    },
    {
      "epoch": 0.1954250211804575,
      "grad_norm": 0.36427995562553406,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.992,
      "step": 173
    },
    {
      "epoch": 0.19655464558034452,
      "grad_norm": 0.351335346698761,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.9732,
      "step": 174
    },
    {
      "epoch": 0.1976842699802316,
      "grad_norm": 0.37934452295303345,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.0971,
      "step": 175
    },
    {
      "epoch": 0.19881389438011862,
      "grad_norm": 0.3769500255584717,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9996,
      "step": 176
    },
    {
      "epoch": 0.19994351878000566,
      "grad_norm": 0.37651577591896057,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.0713,
      "step": 177
    },
    {
      "epoch": 0.2010731431798927,
      "grad_norm": 0.41816914081573486,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.1059,
      "step": 178
    },
    {
      "epoch": 0.20220276757977973,
      "grad_norm": 0.3885565996170044,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.014,
      "step": 179
    },
    {
      "epoch": 0.20333239197966677,
      "grad_norm": 0.3977658748626709,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.1802,
      "step": 180
    },
    {
      "epoch": 0.2044620163795538,
      "grad_norm": 0.38080862164497375,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.0424,
      "step": 181
    },
    {
      "epoch": 0.20559164077944084,
      "grad_norm": 0.3959409296512604,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.0945,
      "step": 182
    },
    {
      "epoch": 0.20672126517932787,
      "grad_norm": 0.4136783480644226,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.0244,
      "step": 183
    },
    {
      "epoch": 0.2078508895792149,
      "grad_norm": 0.42429646849632263,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.1523,
      "step": 184
    },
    {
      "epoch": 0.20898051397910195,
      "grad_norm": 0.42179545760154724,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.143,
      "step": 185
    },
    {
      "epoch": 0.21011013837898898,
      "grad_norm": 0.41918957233428955,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1137,
      "step": 186
    },
    {
      "epoch": 0.21123976277887602,
      "grad_norm": 0.4748425781726837,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.2565,
      "step": 187
    },
    {
      "epoch": 0.21236938717876305,
      "grad_norm": 0.44485995173454285,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.1829,
      "step": 188
    },
    {
      "epoch": 0.2134990115786501,
      "grad_norm": 0.43751591444015503,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.1641,
      "step": 189
    },
    {
      "epoch": 0.21462863597853712,
      "grad_norm": 0.4534774720668793,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.2634,
      "step": 190
    },
    {
      "epoch": 0.21575826037842416,
      "grad_norm": 0.4973703622817993,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.3311,
      "step": 191
    },
    {
      "epoch": 0.21688788477831122,
      "grad_norm": 0.5174738168716431,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.2405,
      "step": 192
    },
    {
      "epoch": 0.21801750917819826,
      "grad_norm": 0.594620943069458,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.3448,
      "step": 193
    },
    {
      "epoch": 0.2191471335780853,
      "grad_norm": 0.5844096541404724,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.3335,
      "step": 194
    },
    {
      "epoch": 0.22027675797797233,
      "grad_norm": 0.6043779850006104,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.3511,
      "step": 195
    },
    {
      "epoch": 0.22140638237785937,
      "grad_norm": 0.5952646136283875,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.3622,
      "step": 196
    },
    {
      "epoch": 0.2225360067777464,
      "grad_norm": 0.735372006893158,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.3572,
      "step": 197
    },
    {
      "epoch": 0.22366563117763344,
      "grad_norm": 0.717737078666687,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.3361,
      "step": 198
    },
    {
      "epoch": 0.22479525557752048,
      "grad_norm": 0.9252230525016785,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.4052,
      "step": 199
    },
    {
      "epoch": 0.2259248799774075,
      "grad_norm": 1.1963423490524292,
      "learning_rate": 0.0,
      "loss": 1.3363,
      "step": 200
    },
    {
      "epoch": 0.2259248799774075,
      "eval_loss": 1.1319868564605713,
      "eval_runtime": 128.4758,
      "eval_samples_per_second": 11.605,
      "eval_steps_per_second": 2.903,
      "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": 3.360213996656394e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}