{
  "best_metric": 1.0226194858551025,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 1.593625498007968,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00796812749003984,
      "grad_norm": 0.3242502510547638,
      "learning_rate": 1e-05,
      "loss": 1.1109,
      "step": 1
    },
    {
      "epoch": 0.00796812749003984,
      "eval_loss": 1.365441918373108,
      "eval_runtime": 16.3569,
      "eval_samples_per_second": 12.961,
      "eval_steps_per_second": 3.24,
      "step": 1
    },
    {
      "epoch": 0.01593625498007968,
      "grad_norm": 0.3418625593185425,
      "learning_rate": 2e-05,
      "loss": 1.0309,
      "step": 2
    },
    {
      "epoch": 0.02390438247011952,
      "grad_norm": 0.40970972180366516,
      "learning_rate": 3e-05,
      "loss": 1.3099,
      "step": 3
    },
    {
      "epoch": 0.03187250996015936,
      "grad_norm": 0.3997035026550293,
      "learning_rate": 4e-05,
      "loss": 1.1565,
      "step": 4
    },
    {
      "epoch": 0.0398406374501992,
      "grad_norm": 0.4518078565597534,
      "learning_rate": 5e-05,
      "loss": 1.1639,
      "step": 5
    },
    {
      "epoch": 0.04780876494023904,
      "grad_norm": 0.4510937035083771,
      "learning_rate": 6e-05,
      "loss": 1.2969,
      "step": 6
    },
    {
      "epoch": 0.055776892430278883,
      "grad_norm": 0.3678322732448578,
      "learning_rate": 7e-05,
      "loss": 1.1943,
      "step": 7
    },
    {
      "epoch": 0.06374501992031872,
      "grad_norm": 0.39165693521499634,
      "learning_rate": 8e-05,
      "loss": 1.0959,
      "step": 8
    },
    {
      "epoch": 0.07171314741035857,
      "grad_norm": 0.47110119462013245,
      "learning_rate": 9e-05,
      "loss": 1.2726,
      "step": 9
    },
    {
      "epoch": 0.0796812749003984,
      "grad_norm": 0.47733739018440247,
      "learning_rate": 0.0001,
      "loss": 1.1794,
      "step": 10
    },
    {
      "epoch": 0.08764940239043825,
      "grad_norm": 0.3454241454601288,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.2443,
      "step": 11
    },
    {
      "epoch": 0.09561752988047809,
      "grad_norm": 0.3162645697593689,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.2034,
      "step": 12
    },
    {
      "epoch": 0.10358565737051793,
      "grad_norm": 0.26847437024116516,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.2383,
      "step": 13
    },
    {
      "epoch": 0.11155378486055777,
      "grad_norm": 0.27468669414520264,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.2556,
      "step": 14
    },
    {
      "epoch": 0.11952191235059761,
      "grad_norm": 0.27332568168640137,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.161,
      "step": 15
    },
    {
      "epoch": 0.12749003984063745,
      "grad_norm": 0.28017857670783997,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.1358,
      "step": 16
    },
    {
      "epoch": 0.13545816733067728,
      "grad_norm": 0.26711124181747437,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.2288,
      "step": 17
    },
    {
      "epoch": 0.14342629482071714,
      "grad_norm": 0.27929574251174927,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.2273,
      "step": 18
    },
    {
      "epoch": 0.15139442231075698,
      "grad_norm": 0.2879405915737152,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.1586,
      "step": 19
    },
    {
      "epoch": 0.1593625498007968,
      "grad_norm": 0.3169874846935272,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.125,
      "step": 20
    },
    {
      "epoch": 0.16733067729083664,
      "grad_norm": 0.35508668422698975,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1428,
      "step": 21
    },
    {
      "epoch": 0.1752988047808765,
      "grad_norm": 0.32874295115470886,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.1935,
      "step": 22
    },
    {
      "epoch": 0.18326693227091634,
      "grad_norm": 0.4287513494491577,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.1858,
      "step": 23
    },
    {
      "epoch": 0.19123505976095617,
      "grad_norm": 0.36414965987205505,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.2555,
      "step": 24
    },
    {
      "epoch": 0.199203187250996,
      "grad_norm": 0.3727114796638489,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.0102,
      "step": 25
    },
    {
      "epoch": 0.20717131474103587,
      "grad_norm": 0.367774099111557,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.1076,
      "step": 26
    },
    {
      "epoch": 0.2151394422310757,
      "grad_norm": 0.4915139675140381,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.18,
      "step": 27
    },
    {
      "epoch": 0.22310756972111553,
      "grad_norm": 0.4582114815711975,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.0745,
      "step": 28
    },
    {
      "epoch": 0.23107569721115537,
      "grad_norm": 0.5187856554985046,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0381,
      "step": 29
    },
    {
      "epoch": 0.23904382470119523,
      "grad_norm": 0.5487841963768005,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.9235,
      "step": 30
    },
    {
      "epoch": 0.24701195219123506,
      "grad_norm": 0.8531146049499512,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.9476,
      "step": 31
    },
    {
      "epoch": 0.2549800796812749,
      "grad_norm": 0.35870563983917236,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.0256,
      "step": 32
    },
    {
      "epoch": 0.26294820717131473,
      "grad_norm": 0.3366445302963257,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.9943,
      "step": 33
    },
    {
      "epoch": 0.27091633466135456,
      "grad_norm": 0.28196606040000916,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.9394,
      "step": 34
    },
    {
      "epoch": 0.2788844621513944,
      "grad_norm": 0.2857818305492401,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.9038,
      "step": 35
    },
    {
      "epoch": 0.2868525896414343,
      "grad_norm": 0.2223840206861496,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.0171,
      "step": 36
    },
    {
      "epoch": 0.2948207171314741,
      "grad_norm": 0.21041598916053772,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.1172,
      "step": 37
    },
    {
      "epoch": 0.30278884462151395,
      "grad_norm": 0.20920740067958832,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.1528,
      "step": 38
    },
    {
      "epoch": 0.3107569721115538,
      "grad_norm": 0.24661560356616974,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.1908,
      "step": 39
    },
    {
      "epoch": 0.3187250996015936,
      "grad_norm": 0.21268917620182037,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.1528,
      "step": 40
    },
    {
      "epoch": 0.32669322709163345,
      "grad_norm": 0.19971424341201782,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.1878,
      "step": 41
    },
    {
      "epoch": 0.3346613545816733,
      "grad_norm": 0.21613259613513947,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.1505,
      "step": 42
    },
    {
      "epoch": 0.3426294820717131,
      "grad_norm": 0.22882933914661407,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.1222,
      "step": 43
    },
    {
      "epoch": 0.350597609561753,
      "grad_norm": 0.23151880502700806,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.0374,
      "step": 44
    },
    {
      "epoch": 0.35856573705179284,
      "grad_norm": 0.2249683141708374,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.0508,
      "step": 45
    },
    {
      "epoch": 0.3665338645418327,
      "grad_norm": 0.2502020299434662,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.1087,
      "step": 46
    },
    {
      "epoch": 0.3745019920318725,
      "grad_norm": 0.25505223870277405,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.2002,
      "step": 47
    },
    {
      "epoch": 0.38247011952191234,
      "grad_norm": 0.2566724717617035,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.1278,
      "step": 48
    },
    {
      "epoch": 0.3904382470119522,
      "grad_norm": 0.24747489392757416,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.1013,
      "step": 49
    },
    {
      "epoch": 0.398406374501992,
      "grad_norm": 0.24226625263690948,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.9954,
      "step": 50
    },
    {
      "epoch": 0.398406374501992,
      "eval_loss": 1.069889783859253,
      "eval_runtime": 16.9228,
      "eval_samples_per_second": 12.528,
      "eval_steps_per_second": 3.132,
      "step": 50
    },
    {
      "epoch": 0.4063745019920319,
      "grad_norm": 0.24536524713039398,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.0053,
      "step": 51
    },
    {
      "epoch": 0.41434262948207173,
      "grad_norm": 0.29171469807624817,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.1316,
      "step": 52
    },
    {
      "epoch": 0.42231075697211157,
      "grad_norm": 0.3208681344985962,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.1426,
      "step": 53
    },
    {
      "epoch": 0.4302788844621514,
      "grad_norm": 0.314033567905426,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.1474,
      "step": 54
    },
    {
      "epoch": 0.43824701195219123,
      "grad_norm": 0.3222683370113373,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.0466,
      "step": 55
    },
    {
      "epoch": 0.44621513944223107,
      "grad_norm": 0.34940779209136963,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.0925,
      "step": 56
    },
    {
      "epoch": 0.4541832669322709,
      "grad_norm": 0.43452611565589905,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.1631,
      "step": 57
    },
    {
      "epoch": 0.46215139442231074,
      "grad_norm": 0.396918922662735,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.9979,
      "step": 58
    },
    {
      "epoch": 0.4701195219123506,
      "grad_norm": 0.4055061340332031,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.0608,
      "step": 59
    },
    {
      "epoch": 0.47808764940239046,
      "grad_norm": 0.4347855746746063,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.9215,
      "step": 60
    },
    {
      "epoch": 0.4860557768924303,
      "grad_norm": 0.4913939833641052,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.8338,
      "step": 61
    },
    {
      "epoch": 0.4940239043824701,
      "grad_norm": 0.8004218339920044,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.8727,
      "step": 62
    },
    {
      "epoch": 0.50199203187251,
      "grad_norm": 0.25151973962783813,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.9377,
      "step": 63
    },
    {
      "epoch": 0.5099601593625498,
      "grad_norm": 0.24308063089847565,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.9317,
      "step": 64
    },
    {
      "epoch": 0.5179282868525896,
      "grad_norm": 0.24601133167743683,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.9164,
      "step": 65
    },
    {
      "epoch": 0.5258964143426295,
      "grad_norm": 0.23069192469120026,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.9585,
      "step": 66
    },
    {
      "epoch": 0.5338645418326693,
      "grad_norm": 0.21055878698825836,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.9402,
      "step": 67
    },
    {
      "epoch": 0.5418326693227091,
      "grad_norm": 0.20026247203350067,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.0347,
      "step": 68
    },
    {
      "epoch": 0.549800796812749,
      "grad_norm": 0.17947176098823547,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.0756,
      "step": 69
    },
    {
      "epoch": 0.5577689243027888,
      "grad_norm": 0.18427321314811707,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.0247,
      "step": 70
    },
    {
      "epoch": 0.5657370517928287,
      "grad_norm": 0.18460749089717865,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.9558,
      "step": 71
    },
    {
      "epoch": 0.5737051792828686,
      "grad_norm": 0.18814508616924286,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1269,
      "step": 72
    },
    {
      "epoch": 0.5816733067729084,
      "grad_norm": 0.2025289535522461,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.0483,
      "step": 73
    },
    {
      "epoch": 0.5896414342629482,
      "grad_norm": 0.1902155578136444,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.1514,
      "step": 74
    },
    {
      "epoch": 0.5976095617529881,
      "grad_norm": 0.2015172243118286,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.0052,
      "step": 75
    },
    {
      "epoch": 0.6055776892430279,
      "grad_norm": 0.20598344504833221,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.1138,
      "step": 76
    },
    {
      "epoch": 0.6135458167330677,
      "grad_norm": 0.2212064266204834,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.1703,
      "step": 77
    },
    {
      "epoch": 0.6215139442231076,
      "grad_norm": 0.23657497763633728,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.067,
      "step": 78
    },
    {
      "epoch": 0.6294820717131474,
      "grad_norm": 0.23551279306411743,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.0648,
      "step": 79
    },
    {
      "epoch": 0.6374501992031872,
      "grad_norm": 0.25298237800598145,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.1297,
      "step": 80
    },
    {
      "epoch": 0.6454183266932271,
      "grad_norm": 0.26222771406173706,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.0816,
      "step": 81
    },
    {
      "epoch": 0.6533864541832669,
      "grad_norm": 0.2847362756729126,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.2205,
      "step": 82
    },
    {
      "epoch": 0.6613545816733067,
      "grad_norm": 0.26608216762542725,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.1272,
      "step": 83
    },
    {
      "epoch": 0.6693227091633466,
      "grad_norm": 0.2973073422908783,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.0214,
      "step": 84
    },
    {
      "epoch": 0.6772908366533864,
      "grad_norm": 0.2927739918231964,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.0131,
      "step": 85
    },
    {
      "epoch": 0.6852589641434262,
      "grad_norm": 0.3021952211856842,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.97,
      "step": 86
    },
    {
      "epoch": 0.6932270916334662,
      "grad_norm": 0.33981063961982727,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.0521,
      "step": 87
    },
    {
      "epoch": 0.701195219123506,
      "grad_norm": 0.3503754734992981,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.1179,
      "step": 88
    },
    {
      "epoch": 0.7091633466135459,
      "grad_norm": 0.3996816575527191,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.068,
      "step": 89
    },
    {
      "epoch": 0.7171314741035857,
      "grad_norm": 0.3864859342575073,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.0865,
      "step": 90
    },
    {
      "epoch": 0.7250996015936255,
      "grad_norm": 0.478983998298645,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.0067,
      "step": 91
    },
    {
      "epoch": 0.7330677290836654,
      "grad_norm": 0.46570664644241333,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.7802,
      "step": 92
    },
    {
      "epoch": 0.7410358565737052,
      "grad_norm": 0.8785074949264526,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.9159,
      "step": 93
    },
    {
      "epoch": 0.749003984063745,
      "grad_norm": 0.1802893877029419,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.9978,
      "step": 94
    },
    {
      "epoch": 0.7569721115537849,
      "grad_norm": 0.17755185067653656,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.8913,
      "step": 95
    },
    {
      "epoch": 0.7649402390438247,
      "grad_norm": 0.18117305636405945,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.8738,
      "step": 96
    },
    {
      "epoch": 0.7729083665338645,
      "grad_norm": 0.17376860976219177,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.867,
      "step": 97
    },
    {
      "epoch": 0.7808764940239044,
      "grad_norm": 0.19229505956172943,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.0108,
      "step": 98
    },
    {
      "epoch": 0.7888446215139442,
      "grad_norm": 0.17664861679077148,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.0283,
      "step": 99
    },
    {
      "epoch": 0.796812749003984,
      "grad_norm": 0.15817706286907196,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.9772,
      "step": 100
    },
    {
      "epoch": 0.796812749003984,
      "eval_loss": 1.0484580993652344,
      "eval_runtime": 16.6647,
      "eval_samples_per_second": 12.721,
      "eval_steps_per_second": 3.18,
      "step": 100
    },
    {
      "epoch": 0.8047808764940239,
      "grad_norm": 0.3082982003688812,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.9515,
      "step": 101
    },
    {
      "epoch": 0.8127490039840638,
      "grad_norm": 0.1807137280702591,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.1024,
      "step": 102
    },
    {
      "epoch": 0.8207171314741036,
      "grad_norm": 0.17332789301872253,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.1216,
      "step": 103
    },
    {
      "epoch": 0.8286852589641435,
      "grad_norm": 0.17725008726119995,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.0249,
      "step": 104
    },
    {
      "epoch": 0.8366533864541833,
      "grad_norm": 0.1845608800649643,
      "learning_rate": 5e-05,
      "loss": 0.9936,
      "step": 105
    },
    {
      "epoch": 0.8446215139442231,
      "grad_norm": 0.24144724011421204,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.0673,
      "step": 106
    },
    {
      "epoch": 0.852589641434263,
      "grad_norm": 0.23575012385845184,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.214,
      "step": 107
    },
    {
      "epoch": 0.8605577689243028,
      "grad_norm": 0.2337813675403595,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.0538,
      "step": 108
    },
    {
      "epoch": 0.8685258964143426,
      "grad_norm": 0.22413738071918488,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.065,
      "step": 109
    },
    {
      "epoch": 0.8764940239043825,
      "grad_norm": 0.22103345394134521,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.0272,
      "step": 110
    },
    {
      "epoch": 0.8844621513944223,
      "grad_norm": 0.25055572390556335,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.1709,
      "step": 111
    },
    {
      "epoch": 0.8924302788844621,
      "grad_norm": 0.24400641024112701,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.9701,
      "step": 112
    },
    {
      "epoch": 0.900398406374502,
      "grad_norm": 0.26395559310913086,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.0888,
      "step": 113
    },
    {
      "epoch": 0.9083665338645418,
      "grad_norm": 0.3101651072502136,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.1487,
      "step": 114
    },
    {
      "epoch": 0.9163346613545816,
      "grad_norm": 0.280747652053833,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1171,
      "step": 115
    },
    {
      "epoch": 0.9243027888446215,
      "grad_norm": 0.40413448214530945,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.1585,
      "step": 116
    },
    {
      "epoch": 0.9322709163346613,
      "grad_norm": 0.3763766288757324,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.9307,
      "step": 117
    },
    {
      "epoch": 0.9402390438247012,
      "grad_norm": 0.36883345246315,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.0472,
      "step": 118
    },
    {
      "epoch": 0.9482071713147411,
      "grad_norm": 0.3642585873603821,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.9853,
      "step": 119
    },
    {
      "epoch": 0.9561752988047809,
      "grad_norm": 0.43915310502052307,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.1615,
      "step": 120
    },
    {
      "epoch": 0.9641434262948207,
      "grad_norm": 0.4310603737831116,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.0356,
      "step": 121
    },
    {
      "epoch": 0.9721115537848606,
      "grad_norm": 0.47499728202819824,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.0468,
      "step": 122
    },
    {
      "epoch": 0.9800796812749004,
      "grad_norm": 0.46576792001724243,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.8957,
      "step": 123
    },
    {
      "epoch": 0.9880478087649402,
      "grad_norm": 0.7518230676651001,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.8025,
      "step": 124
    },
    {
      "epoch": 0.9960159362549801,
      "grad_norm": 0.21898995339870453,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.897,
      "step": 125
    },
    {
      "epoch": 1.00398406374502,
      "grad_norm": 0.49494481086730957,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.4167,
      "step": 126
    },
    {
      "epoch": 1.0119521912350598,
      "grad_norm": 0.14274008572101593,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.924,
      "step": 127
    },
    {
      "epoch": 1.0199203187250996,
      "grad_norm": 0.16929715871810913,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.8777,
      "step": 128
    },
    {
      "epoch": 1.0278884462151394,
      "grad_norm": 0.14793212711811066,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.7854,
      "step": 129
    },
    {
      "epoch": 1.0358565737051793,
      "grad_norm": 0.18344572186470032,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.9512,
      "step": 130
    },
    {
      "epoch": 1.043824701195219,
      "grad_norm": 0.15657907724380493,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.9446,
      "step": 131
    },
    {
      "epoch": 1.051792828685259,
      "grad_norm": 0.15426430106163025,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.9213,
      "step": 132
    },
    {
      "epoch": 1.0597609561752988,
      "grad_norm": 0.20302581787109375,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.0127,
      "step": 133
    },
    {
      "epoch": 1.0677290836653386,
      "grad_norm": 0.164063960313797,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.0012,
      "step": 134
    },
    {
      "epoch": 1.0756972111553784,
      "grad_norm": 0.1688258945941925,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.0375,
      "step": 135
    },
    {
      "epoch": 1.0836653386454183,
      "grad_norm": 0.18732470273971558,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.0868,
      "step": 136
    },
    {
      "epoch": 1.091633466135458,
      "grad_norm": 0.1671607792377472,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.0697,
      "step": 137
    },
    {
      "epoch": 1.099601593625498,
      "grad_norm": 0.17614442110061646,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.1221,
      "step": 138
    },
    {
      "epoch": 1.1075697211155378,
      "grad_norm": 0.18757355213165283,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.9856,
      "step": 139
    },
    {
      "epoch": 1.1155378486055776,
      "grad_norm": 0.20143137872219086,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.0339,
      "step": 140
    },
    {
      "epoch": 1.1235059760956174,
      "grad_norm": 0.19974087178707123,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.0533,
      "step": 141
    },
    {
      "epoch": 1.1314741035856573,
      "grad_norm": 0.2140830159187317,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.0704,
      "step": 142
    },
    {
      "epoch": 1.139442231075697,
      "grad_norm": 0.2263874113559723,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.0187,
      "step": 143
    },
    {
      "epoch": 1.1474103585657371,
      "grad_norm": 0.2369435876607895,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.0188,
      "step": 144
    },
    {
      "epoch": 1.155378486055777,
      "grad_norm": 0.2371990829706192,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.9108,
      "step": 145
    },
    {
      "epoch": 1.1633466135458168,
      "grad_norm": 0.25344139337539673,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.9491,
      "step": 146
    },
    {
      "epoch": 1.1713147410358566,
      "grad_norm": 0.26361140608787537,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.028,
      "step": 147
    },
    {
      "epoch": 1.1792828685258965,
      "grad_norm": 0.2927485704421997,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.9666,
      "step": 148
    },
    {
      "epoch": 1.1872509960159363,
      "grad_norm": 0.2782008945941925,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.9223,
      "step": 149
    },
    {
      "epoch": 1.1952191235059761,
      "grad_norm": 0.30430302023887634,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.9633,
      "step": 150
    },
    {
      "epoch": 1.1952191235059761,
      "eval_loss": 1.0226194858551025,
      "eval_runtime": 16.9268,
      "eval_samples_per_second": 12.525,
      "eval_steps_per_second": 3.131,
      "step": 150
    },
    {
      "epoch": 1.203187250996016,
      "grad_norm": 0.33463671803474426,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.0122,
      "step": 151
    },
    {
      "epoch": 1.2111553784860558,
      "grad_norm": 0.33508387207984924,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.9326,
      "step": 152
    },
    {
      "epoch": 1.2191235059760956,
      "grad_norm": 0.3682953119277954,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.8744,
      "step": 153
    },
    {
      "epoch": 1.2270916334661355,
      "grad_norm": 0.3922181725502014,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.8008,
      "step": 154
    },
    {
      "epoch": 1.2350597609561753,
      "grad_norm": 0.45484086871147156,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.7886,
      "step": 155
    },
    {
      "epoch": 1.2430278884462151,
      "grad_norm": 0.4931739270687103,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.6079,
      "step": 156
    },
    {
      "epoch": 1.250996015936255,
      "grad_norm": 0.30890247225761414,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.6766,
      "step": 157
    },
    {
      "epoch": 1.2589641434262948,
      "grad_norm": 0.1521446853876114,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.8739,
      "step": 158
    },
    {
      "epoch": 1.2669322709163346,
      "grad_norm": 0.16086290776729584,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.8773,
      "step": 159
    },
    {
      "epoch": 1.2749003984063745,
      "grad_norm": 0.16591285169124603,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.771,
      "step": 160
    },
    {
      "epoch": 1.2828685258964143,
      "grad_norm": 0.167489156126976,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.8209,
      "step": 161
    },
    {
      "epoch": 1.2908366533864541,
      "grad_norm": 0.1751011461019516,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.8373,
      "step": 162
    },
    {
      "epoch": 1.298804780876494,
      "grad_norm": 0.18078137934207916,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0102,
      "step": 163
    },
    {
      "epoch": 1.3067729083665338,
      "grad_norm": 0.17670126259326935,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.9915,
      "step": 164
    },
    {
      "epoch": 1.3147410358565736,
      "grad_norm": 0.18374687433242798,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.9959,
      "step": 165
    },
    {
      "epoch": 1.3227091633466135,
      "grad_norm": 0.18274135887622833,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.1196,
      "step": 166
    },
    {
      "epoch": 1.3306772908366533,
      "grad_norm": 0.20548294484615326,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.0851,
      "step": 167
    },
    {
      "epoch": 1.3386454183266931,
      "grad_norm": 0.18688830733299255,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.9967,
      "step": 168
    },
    {
      "epoch": 1.3466135458167332,
      "grad_norm": 0.20327039062976837,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.0931,
      "step": 169
    },
    {
      "epoch": 1.354581673306773,
      "grad_norm": 0.22560401260852814,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.0555,
      "step": 170
    },
    {
      "epoch": 1.3625498007968129,
      "grad_norm": 0.2253095805644989,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.047,
      "step": 171
    },
    {
      "epoch": 1.3705179282868527,
      "grad_norm": 0.21239764988422394,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.9306,
      "step": 172
    },
    {
      "epoch": 1.3784860557768925,
      "grad_norm": 0.25015777349472046,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.1112,
      "step": 173
    },
    {
      "epoch": 1.3864541832669324,
      "grad_norm": 0.24629725515842438,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0676,
      "step": 174
    },
    {
      "epoch": 1.3944223107569722,
      "grad_norm": 0.25692859292030334,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.9786,
      "step": 175
    },
    {
      "epoch": 1.402390438247012,
      "grad_norm": 0.2834259271621704,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9875,
      "step": 176
    },
    {
      "epoch": 1.4103585657370519,
      "grad_norm": 0.2717365622520447,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.9367,
      "step": 177
    },
    {
      "epoch": 1.4183266932270917,
      "grad_norm": 0.30141326785087585,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.987,
      "step": 178
    },
    {
      "epoch": 1.4262948207171315,
      "grad_norm": 0.3208613693714142,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.992,
      "step": 179
    },
    {
      "epoch": 1.4342629482071714,
      "grad_norm": 0.3182251453399658,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.8824,
      "step": 180
    },
    {
      "epoch": 1.4422310756972112,
      "grad_norm": 0.32122623920440674,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.7913,
      "step": 181
    },
    {
      "epoch": 1.450199203187251,
      "grad_norm": 0.3295499086380005,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.831,
      "step": 182
    },
    {
      "epoch": 1.4581673306772909,
      "grad_norm": 0.37478864192962646,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.0263,
      "step": 183
    },
    {
      "epoch": 1.4661354581673307,
      "grad_norm": 0.3989088237285614,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.9787,
      "step": 184
    },
    {
      "epoch": 1.4741035856573705,
      "grad_norm": 0.40363550186157227,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.8225,
      "step": 185
    },
    {
      "epoch": 1.4820717131474104,
      "grad_norm": 0.42043498158454895,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.6117,
      "step": 186
    },
    {
      "epoch": 1.4900398406374502,
      "grad_norm": 0.5256233811378479,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.5101,
      "step": 187
    },
    {
      "epoch": 1.49800796812749,
      "grad_norm": 0.3452662229537964,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.6279,
      "step": 188
    },
    {
      "epoch": 1.5059760956175299,
      "grad_norm": 0.1466367244720459,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.875,
      "step": 189
    },
    {
      "epoch": 1.5139442231075697,
      "grad_norm": 0.1269664466381073,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.7988,
      "step": 190
    },
    {
      "epoch": 1.5219123505976095,
      "grad_norm": 0.13701358437538147,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.8339,
      "step": 191
    },
    {
      "epoch": 1.5298804780876494,
      "grad_norm": 0.14291813969612122,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.9205,
      "step": 192
    },
    {
      "epoch": 1.5378486055776892,
      "grad_norm": 0.16319553554058075,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.9625,
      "step": 193
    },
    {
      "epoch": 1.545816733067729,
      "grad_norm": 0.14466412365436554,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.8159,
      "step": 194
    },
    {
      "epoch": 1.5537848605577689,
      "grad_norm": 0.1631213277578354,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.0096,
      "step": 195
    },
    {
      "epoch": 1.5617529880478087,
      "grad_norm": 0.16483645141124725,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.9732,
      "step": 196
    },
    {
      "epoch": 1.5697211155378485,
      "grad_norm": 0.16057440638542175,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.8549,
      "step": 197
    },
    {
      "epoch": 1.5776892430278884,
      "grad_norm": 0.19115550816059113,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.011,
      "step": 198
    },
    {
      "epoch": 1.5856573705179282,
      "grad_norm": 0.17043732106685638,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9675,
      "step": 199
    },
    {
      "epoch": 1.593625498007968,
      "grad_norm": 0.18523328006267548,
      "learning_rate": 0.0,
      "loss": 0.9881,
      "step": 200
    },
    {
      "epoch": 1.593625498007968,
      "eval_loss": 1.0271413326263428,
      "eval_runtime": 16.9408,
      "eval_samples_per_second": 12.514,
      "eval_steps_per_second": 3.129,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 2.95059547226112e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}