{
  "best_metric": 10.211272239685059,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 0.15597582374731916,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0007798791187365958,
      "grad_norm": 2.526876926422119,
      "learning_rate": 1e-05,
      "loss": 41.5167,
      "step": 1
    },
    {
      "epoch": 0.0007798791187365958,
      "eval_loss": 10.37878131866455,
      "eval_runtime": 5.7405,
      "eval_samples_per_second": 376.276,
      "eval_steps_per_second": 94.069,
      "step": 1
    },
    {
      "epoch": 0.0015597582374731916,
      "grad_norm": 2.687814950942993,
      "learning_rate": 2e-05,
      "loss": 41.525,
      "step": 2
    },
    {
      "epoch": 0.0023396373562097877,
      "grad_norm": 2.2887372970581055,
      "learning_rate": 3e-05,
      "loss": 41.5225,
      "step": 3
    },
    {
      "epoch": 0.003119516474946383,
      "grad_norm": 3.1145741939544678,
      "learning_rate": 4e-05,
      "loss": 41.4807,
      "step": 4
    },
    {
      "epoch": 0.003899395593682979,
      "grad_norm": 2.3700356483459473,
      "learning_rate": 5e-05,
      "loss": 41.5166,
      "step": 5
    },
    {
      "epoch": 0.004679274712419575,
      "grad_norm": 2.0899479389190674,
      "learning_rate": 6e-05,
      "loss": 41.4965,
      "step": 6
    },
    {
      "epoch": 0.00545915383115617,
      "grad_norm": 1.755570650100708,
      "learning_rate": 7e-05,
      "loss": 41.5462,
      "step": 7
    },
    {
      "epoch": 0.006239032949892766,
      "grad_norm": 1.9807515144348145,
      "learning_rate": 8e-05,
      "loss": 41.4568,
      "step": 8
    },
    {
      "epoch": 0.007018912068629362,
      "grad_norm": 2.4661405086517334,
      "learning_rate": 9e-05,
      "loss": 41.4844,
      "step": 9
    },
    {
      "epoch": 0.007798791187365958,
      "grad_norm": 2.6777143478393555,
      "learning_rate": 0.0001,
      "loss": 41.4631,
      "step": 10
    },
    {
      "epoch": 0.008578670306102555,
      "grad_norm": 3.1971161365509033,
      "learning_rate": 9.999316524962345e-05,
      "loss": 41.4276,
      "step": 11
    },
    {
      "epoch": 0.00935854942483915,
      "grad_norm": 2.541069269180298,
      "learning_rate": 9.997266286704631e-05,
      "loss": 41.4671,
      "step": 12
    },
    {
      "epoch": 0.010138428543575747,
      "grad_norm": 2.8264119625091553,
      "learning_rate": 9.993849845741524e-05,
      "loss": 41.4152,
      "step": 13
    },
    {
      "epoch": 0.01091830766231234,
      "grad_norm": 2.8780510425567627,
      "learning_rate": 9.989068136093873e-05,
      "loss": 41.4178,
      "step": 14
    },
    {
      "epoch": 0.011698186781048937,
      "grad_norm": 3.1355903148651123,
      "learning_rate": 9.98292246503335e-05,
      "loss": 41.3888,
      "step": 15
    },
    {
      "epoch": 0.012478065899785533,
      "grad_norm": 3.3326852321624756,
      "learning_rate": 9.975414512725057e-05,
      "loss": 41.4038,
      "step": 16
    },
    {
      "epoch": 0.013257945018522128,
      "grad_norm": 3.73624587059021,
      "learning_rate": 9.966546331768191e-05,
      "loss": 41.3445,
      "step": 17
    },
    {
      "epoch": 0.014037824137258724,
      "grad_norm": 3.433025598526001,
      "learning_rate": 9.956320346634876e-05,
      "loss": 41.3715,
      "step": 18
    },
    {
      "epoch": 0.01481770325599532,
      "grad_norm": 3.941873788833618,
      "learning_rate": 9.944739353007344e-05,
      "loss": 41.3349,
      "step": 19
    },
    {
      "epoch": 0.015597582374731916,
      "grad_norm": 4.041134357452393,
      "learning_rate": 9.931806517013612e-05,
      "loss": 41.3081,
      "step": 20
    },
    {
      "epoch": 0.016377461493468512,
      "grad_norm": 4.013212203979492,
      "learning_rate": 9.917525374361912e-05,
      "loss": 41.3256,
      "step": 21
    },
    {
      "epoch": 0.01715734061220511,
      "grad_norm": 4.252299785614014,
      "learning_rate": 9.901899829374047e-05,
      "loss": 41.2868,
      "step": 22
    },
    {
      "epoch": 0.017937219730941704,
      "grad_norm": 4.342537879943848,
      "learning_rate": 9.884934153917997e-05,
      "loss": 41.2649,
      "step": 23
    },
    {
      "epoch": 0.0187170988496783,
      "grad_norm": 3.9591593742370605,
      "learning_rate": 9.86663298624003e-05,
      "loss": 41.2473,
      "step": 24
    },
    {
      "epoch": 0.019496977968414896,
      "grad_norm": 3.9896693229675293,
      "learning_rate": 9.847001329696653e-05,
      "loss": 41.2313,
      "step": 25
    },
    {
      "epoch": 0.020276857087151493,
      "grad_norm": 4.3405442237854,
      "learning_rate": 9.826044551386744e-05,
      "loss": 41.1687,
      "step": 26
    },
    {
      "epoch": 0.021056736205888087,
      "grad_norm": 4.12937068939209,
      "learning_rate": 9.803768380684242e-05,
      "loss": 41.1849,
      "step": 27
    },
    {
      "epoch": 0.02183661532462468,
      "grad_norm": 3.9627113342285156,
      "learning_rate": 9.780178907671789e-05,
      "loss": 41.184,
      "step": 28
    },
    {
      "epoch": 0.02261649444336128,
      "grad_norm": 3.8954954147338867,
      "learning_rate": 9.755282581475769e-05,
      "loss": 41.1243,
      "step": 29
    },
    {
      "epoch": 0.023396373562097873,
      "grad_norm": 3.5618081092834473,
      "learning_rate": 9.729086208503174e-05,
      "loss": 41.1225,
      "step": 30
    },
    {
      "epoch": 0.02417625268083447,
      "grad_norm": 3.580491542816162,
      "learning_rate": 9.701596950580806e-05,
      "loss": 41.0724,
      "step": 31
    },
    {
      "epoch": 0.024956131799571065,
      "grad_norm": 3.376166343688965,
      "learning_rate": 9.672822322997305e-05,
      "loss": 41.0798,
      "step": 32
    },
    {
      "epoch": 0.025736010918307663,
      "grad_norm": 3.463730573654175,
      "learning_rate": 9.642770192448536e-05,
      "loss": 41.0663,
      "step": 33
    },
    {
      "epoch": 0.026515890037044257,
      "grad_norm": 3.1344120502471924,
      "learning_rate": 9.611448774886924e-05,
      "loss": 41.029,
      "step": 34
    },
    {
      "epoch": 0.027295769155780854,
      "grad_norm": 2.918123483657837,
      "learning_rate": 9.578866633275288e-05,
      "loss": 41.015,
      "step": 35
    },
    {
      "epoch": 0.02807564827451745,
      "grad_norm": 2.6530044078826904,
      "learning_rate": 9.545032675245813e-05,
      "loss": 41.0162,
      "step": 36
    },
    {
      "epoch": 0.028855527393254046,
      "grad_norm": 2.7092931270599365,
      "learning_rate": 9.509956150664796e-05,
      "loss": 40.9695,
      "step": 37
    },
    {
      "epoch": 0.02963540651199064,
      "grad_norm": 2.078944683074951,
      "learning_rate": 9.473646649103818e-05,
      "loss": 41.017,
      "step": 38
    },
    {
      "epoch": 0.030415285630727238,
      "grad_norm": 2.156764030456543,
      "learning_rate": 9.43611409721806e-05,
      "loss": 41.0209,
      "step": 39
    },
    {
      "epoch": 0.031195164749463832,
      "grad_norm": 2.1703855991363525,
      "learning_rate": 9.397368756032445e-05,
      "loss": 40.955,
      "step": 40
    },
    {
      "epoch": 0.03197504386820043,
      "grad_norm": 2.1769373416900635,
      "learning_rate": 9.357421218136386e-05,
      "loss": 40.9234,
      "step": 41
    },
    {
      "epoch": 0.032754922986937024,
      "grad_norm": 1.9560120105743408,
      "learning_rate": 9.316282404787871e-05,
      "loss": 40.9359,
      "step": 42
    },
    {
      "epoch": 0.03353480210567362,
      "grad_norm": 1.6645371913909912,
      "learning_rate": 9.273963562927695e-05,
      "loss": 40.9089,
      "step": 43
    },
    {
      "epoch": 0.03431468122441022,
      "grad_norm": 1.7558664083480835,
      "learning_rate": 9.230476262104677e-05,
      "loss": 40.9149,
      "step": 44
    },
    {
      "epoch": 0.03509456034314681,
      "grad_norm": 1.5846197605133057,
      "learning_rate": 9.185832391312644e-05,
      "loss": 40.9409,
      "step": 45
    },
    {
      "epoch": 0.03587443946188341,
      "grad_norm": 1.5088289976119995,
      "learning_rate": 9.140044155740101e-05,
      "loss": 40.8849,
      "step": 46
    },
    {
      "epoch": 0.03665431858062,
      "grad_norm": 1.580735683441162,
      "learning_rate": 9.093124073433463e-05,
      "loss": 40.9917,
      "step": 47
    },
    {
      "epoch": 0.0374341976993566,
      "grad_norm": 1.1486490964889526,
      "learning_rate": 9.045084971874738e-05,
      "loss": 40.9312,
      "step": 48
    },
    {
      "epoch": 0.0382140768180932,
      "grad_norm": 1.9551911354064941,
      "learning_rate": 8.995939984474624e-05,
      "loss": 40.8176,
      "step": 49
    },
    {
      "epoch": 0.03899395593682979,
      "grad_norm": 1.2632198333740234,
      "learning_rate": 8.945702546981969e-05,
      "loss": 40.8667,
      "step": 50
    },
    {
      "epoch": 0.03899395593682979,
      "eval_loss": 10.228333473205566,
      "eval_runtime": 5.6935,
      "eval_samples_per_second": 379.381,
      "eval_steps_per_second": 94.845,
      "step": 50
    },
    {
      "epoch": 0.039773835055566385,
      "grad_norm": 1.1985833644866943,
      "learning_rate": 8.894386393810563e-05,
      "loss": 40.9282,
      "step": 51
    },
    {
      "epoch": 0.040553714174302986,
      "grad_norm": 1.084412693977356,
      "learning_rate": 8.842005554284296e-05,
      "loss": 40.9223,
      "step": 52
    },
    {
      "epoch": 0.04133359329303958,
      "grad_norm": 1.1719279289245605,
      "learning_rate": 8.788574348801675e-05,
      "loss": 40.8965,
      "step": 53
    },
    {
      "epoch": 0.042113472411776175,
      "grad_norm": 1.0154069662094116,
      "learning_rate": 8.73410738492077e-05,
      "loss": 40.9217,
      "step": 54
    },
    {
      "epoch": 0.04289335153051277,
      "grad_norm": 1.8919508457183838,
      "learning_rate": 8.678619553365659e-05,
      "loss": 40.7622,
      "step": 55
    },
    {
      "epoch": 0.04367323064924936,
      "grad_norm": 1.046948790550232,
      "learning_rate": 8.622126023955446e-05,
      "loss": 40.9274,
      "step": 56
    },
    {
      "epoch": 0.044453109767985964,
      "grad_norm": 1.038284420967102,
      "learning_rate": 8.564642241456986e-05,
      "loss": 40.9788,
      "step": 57
    },
    {
      "epoch": 0.04523298888672256,
      "grad_norm": 1.0178009271621704,
      "learning_rate": 8.506183921362443e-05,
      "loss": 41.0024,
      "step": 58
    },
    {
      "epoch": 0.04601286800545915,
      "grad_norm": 1.6775829792022705,
      "learning_rate": 8.44676704559283e-05,
      "loss": 41.0032,
      "step": 59
    },
    {
      "epoch": 0.04679274712419575,
      "grad_norm": 1.646486759185791,
      "learning_rate": 8.386407858128706e-05,
      "loss": 40.9614,
      "step": 60
    },
    {
      "epoch": 0.04757262624293235,
      "grad_norm": 1.539516806602478,
      "learning_rate": 8.32512286056924e-05,
      "loss": 40.9156,
      "step": 61
    },
    {
      "epoch": 0.04835250536166894,
      "grad_norm": 1.772049069404602,
      "learning_rate": 8.262928807620843e-05,
      "loss": 40.8579,
      "step": 62
    },
    {
      "epoch": 0.049132384480405536,
      "grad_norm": 1.654160499572754,
      "learning_rate": 8.199842702516583e-05,
      "loss": 40.8495,
      "step": 63
    },
    {
      "epoch": 0.04991226359914213,
      "grad_norm": 0.799794614315033,
      "learning_rate": 8.135881792367686e-05,
      "loss": 40.8855,
      "step": 64
    },
    {
      "epoch": 0.05069214271787873,
      "grad_norm": 0.8519944548606873,
      "learning_rate": 8.07106356344834e-05,
      "loss": 40.8998,
      "step": 65
    },
    {
      "epoch": 0.051472021836615325,
      "grad_norm": 0.7602493762969971,
      "learning_rate": 8.005405736415126e-05,
      "loss": 40.9352,
      "step": 66
    },
    {
      "epoch": 0.05225190095535192,
      "grad_norm": 0.6944881677627563,
      "learning_rate": 7.938926261462366e-05,
      "loss": 40.9072,
      "step": 67
    },
    {
      "epoch": 0.053031780074088514,
      "grad_norm": 0.7723879218101501,
      "learning_rate": 7.871643313414718e-05,
      "loss": 40.8872,
      "step": 68
    },
    {
      "epoch": 0.053811659192825115,
      "grad_norm": 0.6100687384605408,
      "learning_rate": 7.803575286758364e-05,
      "loss": 40.9039,
      "step": 69
    },
    {
      "epoch": 0.05459153831156171,
      "grad_norm": 0.7049018740653992,
      "learning_rate": 7.734740790612136e-05,
      "loss": 40.876,
      "step": 70
    },
    {
      "epoch": 0.0553714174302983,
      "grad_norm": 0.735948383808136,
      "learning_rate": 7.66515864363997e-05,
      "loss": 40.861,
      "step": 71
    },
    {
      "epoch": 0.0561512965490349,
      "grad_norm": 0.6083983778953552,
      "learning_rate": 7.594847868906076e-05,
      "loss": 40.9168,
      "step": 72
    },
    {
      "epoch": 0.0569311756677715,
      "grad_norm": 0.6275166273117065,
      "learning_rate": 7.52382768867422e-05,
      "loss": 40.8447,
      "step": 73
    },
    {
      "epoch": 0.05771105478650809,
      "grad_norm": 0.6555567383766174,
      "learning_rate": 7.452117519152542e-05,
      "loss": 40.8841,
      "step": 74
    },
    {
      "epoch": 0.05849093390524469,
      "grad_norm": 0.6510383486747742,
      "learning_rate": 7.379736965185368e-05,
      "loss": 40.8733,
      "step": 75
    },
    {
      "epoch": 0.05927081302398128,
      "grad_norm": 0.6892179846763611,
      "learning_rate": 7.30670581489344e-05,
      "loss": 40.9606,
      "step": 76
    },
    {
      "epoch": 0.06005069214271788,
      "grad_norm": 0.6081066727638245,
      "learning_rate": 7.233044034264034e-05,
      "loss": 40.8553,
      "step": 77
    },
    {
      "epoch": 0.060830571261454476,
      "grad_norm": 0.5948057174682617,
      "learning_rate": 7.158771761692464e-05,
      "loss": 40.9052,
      "step": 78
    },
    {
      "epoch": 0.06161045038019107,
      "grad_norm": 0.61478191614151,
      "learning_rate": 7.083909302476453e-05,
      "loss": 40.9019,
      "step": 79
    },
    {
      "epoch": 0.062390329498927664,
      "grad_norm": 0.6337127089500427,
      "learning_rate": 7.008477123264848e-05,
      "loss": 40.9099,
      "step": 80
    },
    {
      "epoch": 0.06317020861766426,
      "grad_norm": 0.5935375690460205,
      "learning_rate": 6.932495846462261e-05,
      "loss": 40.9038,
      "step": 81
    },
    {
      "epoch": 0.06395008773640086,
      "grad_norm": 0.5433638691902161,
      "learning_rate": 6.855986244591104e-05,
      "loss": 40.8429,
      "step": 82
    },
    {
      "epoch": 0.06472996685513745,
      "grad_norm": 0.62641441822052,
      "learning_rate": 6.778969234612584e-05,
      "loss": 40.8965,
      "step": 83
    },
    {
      "epoch": 0.06550984597387405,
      "grad_norm": 0.47481104731559753,
      "learning_rate": 6.701465872208216e-05,
      "loss": 40.8687,
      "step": 84
    },
    {
      "epoch": 0.06628972509261065,
      "grad_norm": 0.6178433895111084,
      "learning_rate": 6.623497346023418e-05,
      "loss": 40.8422,
      "step": 85
    },
    {
      "epoch": 0.06706960421134724,
      "grad_norm": 0.700564980506897,
      "learning_rate": 6.545084971874738e-05,
      "loss": 40.8797,
      "step": 86
    },
    {
      "epoch": 0.06784948333008384,
      "grad_norm": 0.5639402866363525,
      "learning_rate": 6.466250186922325e-05,
      "loss": 40.8988,
      "step": 87
    },
    {
      "epoch": 0.06862936244882044,
      "grad_norm": 0.6164994835853577,
      "learning_rate": 6.387014543809223e-05,
      "loss": 40.8912,
      "step": 88
    },
    {
      "epoch": 0.06940924156755703,
      "grad_norm": 0.6429092288017273,
      "learning_rate": 6.307399704769099e-05,
      "loss": 40.8972,
      "step": 89
    },
    {
      "epoch": 0.07018912068629363,
      "grad_norm": 0.7178722620010376,
      "learning_rate": 6.227427435703997e-05,
      "loss": 40.8878,
      "step": 90
    },
    {
      "epoch": 0.07096899980503021,
      "grad_norm": 0.6236583590507507,
      "learning_rate": 6.147119600233758e-05,
      "loss": 40.8803,
      "step": 91
    },
    {
      "epoch": 0.07174887892376682,
      "grad_norm": 0.7891796231269836,
      "learning_rate": 6.066498153718735e-05,
      "loss": 40.9036,
      "step": 92
    },
    {
      "epoch": 0.07252875804250342,
      "grad_norm": 0.6549966931343079,
      "learning_rate": 5.985585137257401e-05,
      "loss": 40.8578,
      "step": 93
    },
    {
      "epoch": 0.07330863716124,
      "grad_norm": 0.8011507391929626,
      "learning_rate": 5.90440267166055e-05,
      "loss": 40.9054,
      "step": 94
    },
    {
      "epoch": 0.0740885162799766,
      "grad_norm": 0.6258664131164551,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 40.9008,
      "step": 95
    },
    {
      "epoch": 0.0748683953987132,
      "grad_norm": 0.7289450168609619,
      "learning_rate": 5.74131823855921e-05,
      "loss": 40.8771,
      "step": 96
    },
    {
      "epoch": 0.07564827451744979,
      "grad_norm": 0.9092827439308167,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 40.9348,
      "step": 97
    },
    {
      "epoch": 0.0764281536361864,
      "grad_norm": 0.7976567149162292,
      "learning_rate": 5.577423184847932e-05,
      "loss": 40.8296,
      "step": 98
    },
    {
      "epoch": 0.07720803275492298,
      "grad_norm": 0.8471447229385376,
      "learning_rate": 5.495227651252315e-05,
      "loss": 40.8128,
      "step": 99
    },
    {
      "epoch": 0.07798791187365958,
      "grad_norm": 1.034164547920227,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 40.9444,
      "step": 100
    },
    {
      "epoch": 0.07798791187365958,
      "eval_loss": 10.217576026916504,
      "eval_runtime": 5.7151,
      "eval_samples_per_second": 377.943,
      "eval_steps_per_second": 94.486,
      "step": 100
    },
    {
      "epoch": 0.07876779099239618,
      "grad_norm": 0.7746984958648682,
      "learning_rate": 5.330452921628497e-05,
      "loss": 40.8819,
      "step": 101
    },
    {
      "epoch": 0.07954767011113277,
      "grad_norm": 0.7593967318534851,
      "learning_rate": 5.247918773366112e-05,
      "loss": 40.9227,
      "step": 102
    },
    {
      "epoch": 0.08032754922986937,
      "grad_norm": 0.7995108366012573,
      "learning_rate": 5.165316846586541e-05,
      "loss": 40.8819,
      "step": 103
    },
    {
      "epoch": 0.08110742834860597,
      "grad_norm": 0.7618635296821594,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 40.8453,
      "step": 104
    },
    {
      "epoch": 0.08188730746734256,
      "grad_norm": 2.1329658031463623,
      "learning_rate": 5e-05,
      "loss": 40.6406,
      "step": 105
    },
    {
      "epoch": 0.08266718658607916,
      "grad_norm": 0.7373039722442627,
      "learning_rate": 4.917330276168208e-05,
      "loss": 40.8358,
      "step": 106
    },
    {
      "epoch": 0.08344706570481575,
      "grad_norm": 0.8535414338111877,
      "learning_rate": 4.834683153413459e-05,
      "loss": 40.8877,
      "step": 107
    },
    {
      "epoch": 0.08422694482355235,
      "grad_norm": 0.8930816650390625,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 40.9748,
      "step": 108
    },
    {
      "epoch": 0.08500682394228895,
      "grad_norm": 1.1487185955047607,
      "learning_rate": 4.669547078371504e-05,
      "loss": 40.9777,
      "step": 109
    },
    {
      "epoch": 0.08578670306102554,
      "grad_norm": 1.3498154878616333,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 40.9811,
      "step": 110
    },
    {
      "epoch": 0.08656658217976214,
      "grad_norm": 1.991074800491333,
      "learning_rate": 4.504772348747687e-05,
      "loss": 40.8493,
      "step": 111
    },
    {
      "epoch": 0.08734646129849873,
      "grad_norm": 2.3195078372955322,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 40.759,
      "step": 112
    },
    {
      "epoch": 0.08812634041723533,
      "grad_norm": 0.5596413016319275,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 40.8378,
      "step": 113
    },
    {
      "epoch": 0.08890621953597193,
      "grad_norm": 0.6436564922332764,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 40.8706,
      "step": 114
    },
    {
      "epoch": 0.08968609865470852,
      "grad_norm": 0.6691244840621948,
      "learning_rate": 4.17702704859633e-05,
      "loss": 40.8766,
      "step": 115
    },
    {
      "epoch": 0.09046597777344512,
      "grad_norm": 0.6754659414291382,
      "learning_rate": 4.095597328339452e-05,
      "loss": 40.8155,
      "step": 116
    },
    {
      "epoch": 0.09124585689218172,
      "grad_norm": 0.7361174821853638,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 40.7997,
      "step": 117
    },
    {
      "epoch": 0.0920257360109183,
      "grad_norm": 0.790183424949646,
      "learning_rate": 3.933501846281267e-05,
      "loss": 40.8418,
      "step": 118
    },
    {
      "epoch": 0.0928056151296549,
      "grad_norm": 0.5840756893157959,
      "learning_rate": 3.852880399766243e-05,
      "loss": 40.8756,
      "step": 119
    },
    {
      "epoch": 0.0935854942483915,
      "grad_norm": 0.5648621320724487,
      "learning_rate": 3.772572564296005e-05,
      "loss": 40.8742,
      "step": 120
    },
    {
      "epoch": 0.0943653733671281,
      "grad_norm": 0.6849337816238403,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 40.8714,
      "step": 121
    },
    {
      "epoch": 0.0951452524858647,
      "grad_norm": 0.8350796103477478,
      "learning_rate": 3.612985456190778e-05,
      "loss": 40.7928,
      "step": 122
    },
    {
      "epoch": 0.09592513160460128,
      "grad_norm": 0.6076328754425049,
      "learning_rate": 3.533749813077677e-05,
      "loss": 40.8374,
      "step": 123
    },
    {
      "epoch": 0.09670501072333788,
      "grad_norm": 0.6455417275428772,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 40.8406,
      "step": 124
    },
    {
      "epoch": 0.09748488984207448,
      "grad_norm": 0.6414567828178406,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 40.7887,
      "step": 125
    },
    {
      "epoch": 0.09826476896081107,
      "grad_norm": 0.6727384328842163,
      "learning_rate": 3.298534127791785e-05,
      "loss": 40.8339,
      "step": 126
    },
    {
      "epoch": 0.09904464807954767,
      "grad_norm": 0.7361812591552734,
      "learning_rate": 3.221030765387417e-05,
      "loss": 40.8182,
      "step": 127
    },
    {
      "epoch": 0.09982452719828426,
      "grad_norm": 0.564975380897522,
      "learning_rate": 3.144013755408895e-05,
      "loss": 40.8264,
      "step": 128
    },
    {
      "epoch": 0.10060440631702086,
      "grad_norm": 0.5750811696052551,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 40.834,
      "step": 129
    },
    {
      "epoch": 0.10138428543575746,
      "grad_norm": 0.5753947496414185,
      "learning_rate": 2.991522876735154e-05,
      "loss": 40.8516,
      "step": 130
    },
    {
      "epoch": 0.10216416455449405,
      "grad_norm": 0.5774562954902649,
      "learning_rate": 2.916090697523549e-05,
      "loss": 40.8232,
      "step": 131
    },
    {
      "epoch": 0.10294404367323065,
      "grad_norm": 0.6086456775665283,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 40.7957,
      "step": 132
    },
    {
      "epoch": 0.10372392279196725,
      "grad_norm": 0.8998701572418213,
      "learning_rate": 2.766955965735968e-05,
      "loss": 40.7871,
      "step": 133
    },
    {
      "epoch": 0.10450380191070384,
      "grad_norm": 0.6536602973937988,
      "learning_rate": 2.693294185106562e-05,
      "loss": 40.8849,
      "step": 134
    },
    {
      "epoch": 0.10528368102944044,
      "grad_norm": 0.5051807165145874,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 40.82,
      "step": 135
    },
    {
      "epoch": 0.10606356014817703,
      "grad_norm": 0.6145546436309814,
      "learning_rate": 2.547882480847461e-05,
      "loss": 40.8889,
      "step": 136
    },
    {
      "epoch": 0.10684343926691363,
      "grad_norm": 0.5775119662284851,
      "learning_rate": 2.476172311325783e-05,
      "loss": 40.8549,
      "step": 137
    },
    {
      "epoch": 0.10762331838565023,
      "grad_norm": 0.7312015891075134,
      "learning_rate": 2.405152131093926e-05,
      "loss": 40.7823,
      "step": 138
    },
    {
      "epoch": 0.10840319750438682,
      "grad_norm": 0.5825233459472656,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 40.819,
      "step": 139
    },
    {
      "epoch": 0.10918307662312342,
      "grad_norm": 0.7027179598808289,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 40.8551,
      "step": 140
    },
    {
      "epoch": 0.10996295574186,
      "grad_norm": 0.6119231581687927,
      "learning_rate": 2.196424713241637e-05,
      "loss": 40.8047,
      "step": 141
    },
    {
      "epoch": 0.1107428348605966,
      "grad_norm": 0.9758428335189819,
      "learning_rate": 2.128356686585282e-05,
      "loss": 40.849,
      "step": 142
    },
    {
      "epoch": 0.11152271397933321,
      "grad_norm": 0.5225445032119751,
      "learning_rate": 2.061073738537635e-05,
      "loss": 40.8348,
      "step": 143
    },
    {
      "epoch": 0.1123025930980698,
      "grad_norm": 1.0314196348190308,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 40.9268,
      "step": 144
    },
    {
      "epoch": 0.1130824722168064,
      "grad_norm": 0.7246149778366089,
      "learning_rate": 1.928936436551661e-05,
      "loss": 40.8103,
      "step": 145
    },
    {
      "epoch": 0.113862351335543,
      "grad_norm": 0.5631200671195984,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 40.8089,
      "step": 146
    },
    {
      "epoch": 0.11464223045427958,
      "grad_norm": 0.6935296654701233,
      "learning_rate": 1.800157297483417e-05,
      "loss": 40.8477,
      "step": 147
    },
    {
      "epoch": 0.11542210957301619,
      "grad_norm": 0.8210736513137817,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 40.8292,
      "step": 148
    },
    {
      "epoch": 0.11620198869175277,
      "grad_norm": 0.8503446578979492,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 40.8332,
      "step": 149
    },
    {
      "epoch": 0.11698186781048937,
      "grad_norm": 0.8683088421821594,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 40.8357,
      "step": 150
    },
    {
      "epoch": 0.11698186781048937,
      "eval_loss": 10.211272239685059,
      "eval_runtime": 5.6676,
      "eval_samples_per_second": 381.112,
      "eval_steps_per_second": 95.278,
      "step": 150
    },
    {
      "epoch": 0.11776174692922597,
      "grad_norm": 0.5858197808265686,
      "learning_rate": 1.553232954407171e-05,
      "loss": 40.8766,
      "step": 151
    },
    {
      "epoch": 0.11854162604796256,
      "grad_norm": 0.5694209933280945,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 40.8895,
      "step": 152
    },
    {
      "epoch": 0.11932150516669916,
      "grad_norm": 0.7384527921676636,
      "learning_rate": 1.435357758543015e-05,
      "loss": 40.8667,
      "step": 153
    },
    {
      "epoch": 0.12010138428543576,
      "grad_norm": 0.6088923215866089,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 40.7573,
      "step": 154
    },
    {
      "epoch": 0.12088126340417235,
      "grad_norm": 2.4358131885528564,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 40.5275,
      "step": 155
    },
    {
      "epoch": 0.12166114252290895,
      "grad_norm": 0.5830925107002258,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 40.8207,
      "step": 156
    },
    {
      "epoch": 0.12244102164164554,
      "grad_norm": 0.7969471216201782,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 40.9062,
      "step": 157
    },
    {
      "epoch": 0.12322090076038214,
      "grad_norm": 0.7817261815071106,
      "learning_rate": 1.157994445715706e-05,
      "loss": 40.9253,
      "step": 158
    },
    {
      "epoch": 0.12400077987911874,
      "grad_norm": 1.426339030265808,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 41.0472,
      "step": 159
    },
    {
      "epoch": 0.12478065899785533,
      "grad_norm": 1.3720293045043945,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 40.9191,
      "step": 160
    },
    {
      "epoch": 0.12556053811659193,
      "grad_norm": 2.7072181701660156,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 40.9597,
      "step": 161
    },
    {
      "epoch": 0.12634041723532852,
      "grad_norm": 1.7928940057754517,
      "learning_rate": 9.549150281252633e-06,
      "loss": 40.829,
      "step": 162
    },
    {
      "epoch": 0.12712029635406513,
      "grad_norm": 0.5371671915054321,
      "learning_rate": 9.068759265665384e-06,
      "loss": 40.8395,
      "step": 163
    },
    {
      "epoch": 0.12790017547280172,
      "grad_norm": 0.5449931621551514,
      "learning_rate": 8.599558442598998e-06,
      "loss": 40.8481,
      "step": 164
    },
    {
      "epoch": 0.1286800545915383,
      "grad_norm": 0.5795981884002686,
      "learning_rate": 8.141676086873572e-06,
      "loss": 40.7797,
      "step": 165
    },
    {
      "epoch": 0.1294599337102749,
      "grad_norm": 0.4264478385448456,
      "learning_rate": 7.695237378953223e-06,
      "loss": 40.8576,
      "step": 166
    },
    {
      "epoch": 0.1302398128290115,
      "grad_norm": 0.6426555514335632,
      "learning_rate": 7.260364370723044e-06,
      "loss": 40.8285,
      "step": 167
    },
    {
      "epoch": 0.1310196919477481,
      "grad_norm": 0.6536228656768799,
      "learning_rate": 6.837175952121306e-06,
      "loss": 40.7384,
      "step": 168
    },
    {
      "epoch": 0.13179957106648468,
      "grad_norm": 0.6299805045127869,
      "learning_rate": 6.425787818636131e-06,
      "loss": 40.8384,
      "step": 169
    },
    {
      "epoch": 0.1325794501852213,
      "grad_norm": 0.7330281734466553,
      "learning_rate": 6.026312439675552e-06,
      "loss": 40.8606,
      "step": 170
    },
    {
      "epoch": 0.13335932930395789,
      "grad_norm": 0.6534645557403564,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 40.8382,
      "step": 171
    },
    {
      "epoch": 0.13413920842269447,
      "grad_norm": 0.7620009183883667,
      "learning_rate": 5.263533508961827e-06,
      "loss": 40.8482,
      "step": 172
    },
    {
      "epoch": 0.1349190875414311,
      "grad_norm": 0.47706276178359985,
      "learning_rate": 4.900438493352055e-06,
      "loss": 40.7956,
      "step": 173
    },
    {
      "epoch": 0.13569896666016767,
      "grad_norm": 0.5959166884422302,
      "learning_rate": 4.549673247541875e-06,
      "loss": 40.8008,
      "step": 174
    },
    {
      "epoch": 0.13647884577890426,
      "grad_norm": 0.509042501449585,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 40.7646,
      "step": 175
    },
    {
      "epoch": 0.13725872489764088,
      "grad_norm": 0.4765230715274811,
      "learning_rate": 3.885512251130763e-06,
      "loss": 40.8103,
      "step": 176
    },
    {
      "epoch": 0.13803860401637746,
      "grad_norm": 0.5187248587608337,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 40.7728,
      "step": 177
    },
    {
      "epoch": 0.13881848313511405,
      "grad_norm": 0.4397425949573517,
      "learning_rate": 3.271776770026963e-06,
      "loss": 40.7758,
      "step": 178
    },
    {
      "epoch": 0.13959836225385067,
      "grad_norm": 0.4881274104118347,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 40.7738,
      "step": 179
    },
    {
      "epoch": 0.14037824137258725,
      "grad_norm": 0.65187668800354,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 40.8658,
      "step": 180
    },
    {
      "epoch": 0.14115812049132384,
      "grad_norm": 0.5115351676940918,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 40.7763,
      "step": 181
    },
    {
      "epoch": 0.14193799961006043,
      "grad_norm": 0.5020357370376587,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 40.7786,
      "step": 182
    },
    {
      "epoch": 0.14271787872879704,
      "grad_norm": 0.7339043617248535,
      "learning_rate": 1.962316193157593e-06,
      "loss": 40.7989,
      "step": 183
    },
    {
      "epoch": 0.14349775784753363,
      "grad_norm": 0.6433815956115723,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 40.8709,
      "step": 184
    },
    {
      "epoch": 0.14427763696627022,
      "grad_norm": 0.48516517877578735,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 40.8185,
      "step": 185
    },
    {
      "epoch": 0.14505751608500683,
      "grad_norm": 0.6620453000068665,
      "learning_rate": 1.333670137599713e-06,
      "loss": 40.8463,
      "step": 186
    },
    {
      "epoch": 0.14583739520374342,
      "grad_norm": 0.6768563985824585,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 40.8282,
      "step": 187
    },
    {
      "epoch": 0.14661727432248,
      "grad_norm": 0.5989657640457153,
      "learning_rate": 9.810017062595322e-07,
      "loss": 40.8064,
      "step": 188
    },
    {
      "epoch": 0.14739715344121662,
      "grad_norm": 0.7759832739830017,
      "learning_rate": 8.247462563808817e-07,
      "loss": 40.8545,
      "step": 189
    },
    {
      "epoch": 0.1481770325599532,
      "grad_norm": 0.5701521039009094,
      "learning_rate": 6.819348298638839e-07,
      "loss": 40.8302,
      "step": 190
    },
    {
      "epoch": 0.1489569116786898,
      "grad_norm": 0.5439790487289429,
      "learning_rate": 5.526064699265753e-07,
      "loss": 40.8794,
      "step": 191
    },
    {
      "epoch": 0.1497367907974264,
      "grad_norm": 0.6554091572761536,
      "learning_rate": 4.367965336512403e-07,
      "loss": 40.8716,
      "step": 192
    },
    {
      "epoch": 0.150516669916163,
      "grad_norm": 0.9526164531707764,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 40.7741,
      "step": 193
    },
    {
      "epoch": 0.15129654903489959,
      "grad_norm": 0.6600074172019958,
      "learning_rate": 2.458548727494292e-07,
      "loss": 40.7696,
      "step": 194
    },
    {
      "epoch": 0.15207642815363617,
      "grad_norm": 0.6150196194648743,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 40.8571,
      "step": 195
    },
    {
      "epoch": 0.1528563072723728,
      "grad_norm": 0.903807520866394,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 40.7745,
      "step": 196
    },
    {
      "epoch": 0.15363618639110938,
      "grad_norm": 0.6509922742843628,
      "learning_rate": 6.150154258476315e-08,
      "loss": 40.8401,
      "step": 197
    },
    {
      "epoch": 0.15441606550984596,
      "grad_norm": 0.6398036479949951,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 40.8783,
      "step": 198
    },
    {
      "epoch": 0.15519594462858258,
      "grad_norm": 0.7212160229682922,
      "learning_rate": 6.834750376549792e-09,
      "loss": 40.7388,
      "step": 199
    },
    {
      "epoch": 0.15597582374731916,
      "grad_norm": 1.0435220003128052,
      "learning_rate": 0.0,
      "loss": 40.7138,
      "step": 200
    },
    {
      "epoch": 0.15597582374731916,
      "eval_loss": 10.211485862731934,
      "eval_runtime": 5.6548,
      "eval_samples_per_second": 381.974,
      "eval_steps_per_second": 95.494,
      "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": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 53326379483136.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}