{
  "best_metric": 0.2749819755554199,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.6739679865206403,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003369839932603201,
      "grad_norm": 7.523158073425293,
      "learning_rate": 1e-05,
      "loss": 2.1494,
      "step": 1
    },
    {
      "epoch": 0.003369839932603201,
      "eval_loss": 0.8092716336250305,
      "eval_runtime": 119.8404,
      "eval_samples_per_second": 4.172,
      "eval_steps_per_second": 1.043,
      "step": 1
    },
    {
      "epoch": 0.006739679865206402,
      "grad_norm": 9.174283027648926,
      "learning_rate": 2e-05,
      "loss": 2.7371,
      "step": 2
    },
    {
      "epoch": 0.010109519797809604,
      "grad_norm": 7.828287124633789,
      "learning_rate": 3e-05,
      "loss": 2.5091,
      "step": 3
    },
    {
      "epoch": 0.013479359730412805,
      "grad_norm": 5.485031604766846,
      "learning_rate": 4e-05,
      "loss": 2.3461,
      "step": 4
    },
    {
      "epoch": 0.016849199663016005,
      "grad_norm": 5.057663440704346,
      "learning_rate": 5e-05,
      "loss": 2.1335,
      "step": 5
    },
    {
      "epoch": 0.020219039595619208,
      "grad_norm": 5.425457954406738,
      "learning_rate": 6e-05,
      "loss": 2.2379,
      "step": 6
    },
    {
      "epoch": 0.02358887952822241,
      "grad_norm": 4.443009376525879,
      "learning_rate": 7e-05,
      "loss": 1.9256,
      "step": 7
    },
    {
      "epoch": 0.02695871946082561,
      "grad_norm": 4.284821033477783,
      "learning_rate": 8e-05,
      "loss": 2.0171,
      "step": 8
    },
    {
      "epoch": 0.030328559393428812,
      "grad_norm": 3.9193997383117676,
      "learning_rate": 9e-05,
      "loss": 1.7516,
      "step": 9
    },
    {
      "epoch": 0.03369839932603201,
      "grad_norm": 4.5055975914001465,
      "learning_rate": 0.0001,
      "loss": 1.8936,
      "step": 10
    },
    {
      "epoch": 0.03706823925863521,
      "grad_norm": 3.605220317840576,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.7015,
      "step": 11
    },
    {
      "epoch": 0.040438079191238416,
      "grad_norm": 3.5049567222595215,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.8522,
      "step": 12
    },
    {
      "epoch": 0.04380791912384162,
      "grad_norm": 4.1059160232543945,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.8485,
      "step": 13
    },
    {
      "epoch": 0.04717775905644482,
      "grad_norm": 3.140507698059082,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.6875,
      "step": 14
    },
    {
      "epoch": 0.05054759898904802,
      "grad_norm": 3.499692440032959,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.8699,
      "step": 15
    },
    {
      "epoch": 0.05391743892165122,
      "grad_norm": 2.9643075466156006,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.7781,
      "step": 16
    },
    {
      "epoch": 0.05728727885425442,
      "grad_norm": 3.485825300216675,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.6629,
      "step": 17
    },
    {
      "epoch": 0.060657118786857624,
      "grad_norm": 3.2578351497650146,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.7217,
      "step": 18
    },
    {
      "epoch": 0.06402695871946082,
      "grad_norm": 3.0005578994750977,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.693,
      "step": 19
    },
    {
      "epoch": 0.06739679865206402,
      "grad_norm": 2.777435779571533,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.6064,
      "step": 20
    },
    {
      "epoch": 0.07076663858466722,
      "grad_norm": 3.0843496322631836,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6554,
      "step": 21
    },
    {
      "epoch": 0.07413647851727043,
      "grad_norm": 3.3692944049835205,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.5929,
      "step": 22
    },
    {
      "epoch": 0.07750631844987363,
      "grad_norm": 3.13559889793396,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.6566,
      "step": 23
    },
    {
      "epoch": 0.08087615838247683,
      "grad_norm": 3.035297155380249,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.7637,
      "step": 24
    },
    {
      "epoch": 0.08424599831508003,
      "grad_norm": 3.0559241771698,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6777,
      "step": 25
    },
    {
      "epoch": 0.08761583824768324,
      "grad_norm": 3.1593425273895264,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5599,
      "step": 26
    },
    {
      "epoch": 0.09098567818028644,
      "grad_norm": 3.1773195266723633,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.5623,
      "step": 27
    },
    {
      "epoch": 0.09435551811288964,
      "grad_norm": 3.58475399017334,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.5323,
      "step": 28
    },
    {
      "epoch": 0.09772535804549284,
      "grad_norm": 3.061366558074951,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.482,
      "step": 29
    },
    {
      "epoch": 0.10109519797809605,
      "grad_norm": 2.8615095615386963,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.2556,
      "step": 30
    },
    {
      "epoch": 0.10446503791069923,
      "grad_norm": 2.979785680770874,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.3782,
      "step": 31
    },
    {
      "epoch": 0.10783487784330244,
      "grad_norm": 3.0386674404144287,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.2312,
      "step": 32
    },
    {
      "epoch": 0.11120471777590564,
      "grad_norm": 2.6157619953155518,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2005,
      "step": 33
    },
    {
      "epoch": 0.11457455770850884,
      "grad_norm": 2.7591357231140137,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.4317,
      "step": 34
    },
    {
      "epoch": 0.11794439764111204,
      "grad_norm": 2.922539710998535,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.5454,
      "step": 35
    },
    {
      "epoch": 0.12131423757371525,
      "grad_norm": 2.922325611114502,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5959,
      "step": 36
    },
    {
      "epoch": 0.12468407750631845,
      "grad_norm": 2.637524366378784,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.5938,
      "step": 37
    },
    {
      "epoch": 0.12805391743892164,
      "grad_norm": 2.856714963912964,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5612,
      "step": 38
    },
    {
      "epoch": 0.13142375737152484,
      "grad_norm": 2.888352870941162,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.683,
      "step": 39
    },
    {
      "epoch": 0.13479359730412804,
      "grad_norm": 2.799309492111206,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.6642,
      "step": 40
    },
    {
      "epoch": 0.13816343723673125,
      "grad_norm": 2.946946859359741,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.7212,
      "step": 41
    },
    {
      "epoch": 0.14153327716933445,
      "grad_norm": 2.733062982559204,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.5566,
      "step": 42
    },
    {
      "epoch": 0.14490311710193765,
      "grad_norm": 2.725644588470459,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.5424,
      "step": 43
    },
    {
      "epoch": 0.14827295703454085,
      "grad_norm": 2.6782684326171875,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.4512,
      "step": 44
    },
    {
      "epoch": 0.15164279696714406,
      "grad_norm": 3.1654129028320312,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.4352,
      "step": 45
    },
    {
      "epoch": 0.15501263689974726,
      "grad_norm": 4.080636501312256,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5494,
      "step": 46
    },
    {
      "epoch": 0.15838247683235046,
      "grad_norm": 2.9639883041381836,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4076,
      "step": 47
    },
    {
      "epoch": 0.16175231676495366,
      "grad_norm": 2.5637450218200684,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3265,
      "step": 48
    },
    {
      "epoch": 0.16512215669755687,
      "grad_norm": 2.4928646087646484,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3455,
      "step": 49
    },
    {
      "epoch": 0.16849199663016007,
      "grad_norm": 3.090738296508789,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.5222,
      "step": 50
    },
    {
      "epoch": 0.16849199663016007,
      "eval_loss": 0.36440208554267883,
      "eval_runtime": 121.1871,
      "eval_samples_per_second": 4.126,
      "eval_steps_per_second": 1.031,
      "step": 50
    },
    {
      "epoch": 0.17186183656276327,
      "grad_norm": 2.4865734577178955,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.5456,
      "step": 51
    },
    {
      "epoch": 0.17523167649536647,
      "grad_norm": 1.9161837100982666,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.5705,
      "step": 52
    },
    {
      "epoch": 0.17860151642796968,
      "grad_norm": 3.304349660873413,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5712,
      "step": 53
    },
    {
      "epoch": 0.18197135636057288,
      "grad_norm": 2.0122663974761963,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.5889,
      "step": 54
    },
    {
      "epoch": 0.18534119629317608,
      "grad_norm": 2.1787772178649902,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.4923,
      "step": 55
    },
    {
      "epoch": 0.18871103622577928,
      "grad_norm": 2.114611864089966,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.5003,
      "step": 56
    },
    {
      "epoch": 0.19208087615838249,
      "grad_norm": 2.0512936115264893,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.485,
      "step": 57
    },
    {
      "epoch": 0.1954507160909857,
      "grad_norm": 2.0864319801330566,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4577,
      "step": 58
    },
    {
      "epoch": 0.1988205560235889,
      "grad_norm": 1.9597128629684448,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.4234,
      "step": 59
    },
    {
      "epoch": 0.2021903959561921,
      "grad_norm": 1.8872239589691162,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3397,
      "step": 60
    },
    {
      "epoch": 0.2055602358887953,
      "grad_norm": 1.9661518335342407,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.2693,
      "step": 61
    },
    {
      "epoch": 0.20893007582139847,
      "grad_norm": 1.8932143449783325,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.1117,
      "step": 62
    },
    {
      "epoch": 0.21229991575400167,
      "grad_norm": 2.0526702404022217,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.3426,
      "step": 63
    },
    {
      "epoch": 0.21566975568660487,
      "grad_norm": 2.1341538429260254,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.3113,
      "step": 64
    },
    {
      "epoch": 0.21903959561920808,
      "grad_norm": 2.251939535140991,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.6047,
      "step": 65
    },
    {
      "epoch": 0.22240943555181128,
      "grad_norm": 2.1989572048187256,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.5021,
      "step": 66
    },
    {
      "epoch": 0.22577927548441448,
      "grad_norm": 2.166799306869507,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5386,
      "step": 67
    },
    {
      "epoch": 0.22914911541701768,
      "grad_norm": 2.1859495639801025,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.4406,
      "step": 68
    },
    {
      "epoch": 0.2325189553496209,
      "grad_norm": 2.120924234390259,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.3459,
      "step": 69
    },
    {
      "epoch": 0.2358887952822241,
      "grad_norm": 2.228654384613037,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.6667,
      "step": 70
    },
    {
      "epoch": 0.2392586352148273,
      "grad_norm": 2.1573808193206787,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.2445,
      "step": 71
    },
    {
      "epoch": 0.2426284751474305,
      "grad_norm": 2.063796281814575,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.2835,
      "step": 72
    },
    {
      "epoch": 0.2459983150800337,
      "grad_norm": 2.065737247467041,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.3699,
      "step": 73
    },
    {
      "epoch": 0.2493681550126369,
      "grad_norm": 2.0461854934692383,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.1077,
      "step": 74
    },
    {
      "epoch": 0.2527379949452401,
      "grad_norm": 1.9452458620071411,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.0426,
      "step": 75
    },
    {
      "epoch": 0.2561078348778433,
      "grad_norm": 2.243985414505005,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.268,
      "step": 76
    },
    {
      "epoch": 0.2594776748104465,
      "grad_norm": 2.390695810317993,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.2456,
      "step": 77
    },
    {
      "epoch": 0.2628475147430497,
      "grad_norm": 2.1825804710388184,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.0941,
      "step": 78
    },
    {
      "epoch": 0.2662173546756529,
      "grad_norm": 2.441941738128662,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1738,
      "step": 79
    },
    {
      "epoch": 0.2695871946082561,
      "grad_norm": 2.0857393741607666,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.0275,
      "step": 80
    },
    {
      "epoch": 0.2729570345408593,
      "grad_norm": 2.1092944145202637,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.9757,
      "step": 81
    },
    {
      "epoch": 0.2763268744734625,
      "grad_norm": 2.0380167961120605,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.8574,
      "step": 82
    },
    {
      "epoch": 0.2796967144060657,
      "grad_norm": 1.917100191116333,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.9452,
      "step": 83
    },
    {
      "epoch": 0.2830665543386689,
      "grad_norm": 2.122340202331543,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.1275,
      "step": 84
    },
    {
      "epoch": 0.2864363942712721,
      "grad_norm": 2.1233088970184326,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.1109,
      "step": 85
    },
    {
      "epoch": 0.2898062342038753,
      "grad_norm": 2.2637298107147217,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.3176,
      "step": 86
    },
    {
      "epoch": 0.2931760741364785,
      "grad_norm": 2.145240545272827,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.3032,
      "step": 87
    },
    {
      "epoch": 0.2965459140690817,
      "grad_norm": 2.2836382389068604,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.2905,
      "step": 88
    },
    {
      "epoch": 0.2999157540016849,
      "grad_norm": 2.2280659675598145,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.2851,
      "step": 89
    },
    {
      "epoch": 0.3032855939342881,
      "grad_norm": 2.3336851596832275,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.3529,
      "step": 90
    },
    {
      "epoch": 0.3066554338668913,
      "grad_norm": 2.3799586296081543,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.4304,
      "step": 91
    },
    {
      "epoch": 0.3100252737994945,
      "grad_norm": 2.417768716812134,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.3969,
      "step": 92
    },
    {
      "epoch": 0.3133951137320977,
      "grad_norm": 2.566636085510254,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.35,
      "step": 93
    },
    {
      "epoch": 0.3167649536647009,
      "grad_norm": 2.5842761993408203,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.3297,
      "step": 94
    },
    {
      "epoch": 0.3201347935973041,
      "grad_norm": 2.478417158126831,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.9393,
      "step": 95
    },
    {
      "epoch": 0.3235046335299073,
      "grad_norm": 2.5178377628326416,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.3365,
      "step": 96
    },
    {
      "epoch": 0.32687447346251053,
      "grad_norm": 2.2588064670562744,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.2522,
      "step": 97
    },
    {
      "epoch": 0.33024431339511373,
      "grad_norm": 2.3114778995513916,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.2872,
      "step": 98
    },
    {
      "epoch": 0.33361415332771693,
      "grad_norm": 2.2465057373046875,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.0875,
      "step": 99
    },
    {
      "epoch": 0.33698399326032014,
      "grad_norm": 2.4964780807495117,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.2759,
      "step": 100
    },
    {
      "epoch": 0.33698399326032014,
      "eval_loss": 0.31484583020210266,
      "eval_runtime": 121.1416,
      "eval_samples_per_second": 4.127,
      "eval_steps_per_second": 1.032,
      "step": 100
    },
    {
      "epoch": 0.34035383319292334,
      "grad_norm": 2.046468496322632,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4105,
      "step": 101
    },
    {
      "epoch": 0.34372367312552654,
      "grad_norm": 1.7251356840133667,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2823,
      "step": 102
    },
    {
      "epoch": 0.34709351305812974,
      "grad_norm": 1.703196406364441,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.4311,
      "step": 103
    },
    {
      "epoch": 0.35046335299073295,
      "grad_norm": 1.6881848573684692,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.4711,
      "step": 104
    },
    {
      "epoch": 0.35383319292333615,
      "grad_norm": 1.69926917552948,
      "learning_rate": 5e-05,
      "loss": 1.4165,
      "step": 105
    },
    {
      "epoch": 0.35720303285593935,
      "grad_norm": 1.7950522899627686,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.3449,
      "step": 106
    },
    {
      "epoch": 0.36057287278854255,
      "grad_norm": 1.7758194208145142,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2668,
      "step": 107
    },
    {
      "epoch": 0.36394271272114576,
      "grad_norm": 1.7041988372802734,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.4054,
      "step": 108
    },
    {
      "epoch": 0.36731255265374896,
      "grad_norm": 1.9329830408096313,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4071,
      "step": 109
    },
    {
      "epoch": 0.37068239258635216,
      "grad_norm": 1.873896598815918,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5791,
      "step": 110
    },
    {
      "epoch": 0.37405223251895536,
      "grad_norm": 1.7990448474884033,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.2499,
      "step": 111
    },
    {
      "epoch": 0.37742207245155857,
      "grad_norm": 1.9483150243759155,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.2249,
      "step": 112
    },
    {
      "epoch": 0.38079191238416177,
      "grad_norm": 1.7141283750534058,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.17,
      "step": 113
    },
    {
      "epoch": 0.38416175231676497,
      "grad_norm": 2.013573408126831,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.3728,
      "step": 114
    },
    {
      "epoch": 0.3875315922493682,
      "grad_norm": 1.7621855735778809,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1382,
      "step": 115
    },
    {
      "epoch": 0.3909014321819714,
      "grad_norm": 1.9862143993377686,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.3786,
      "step": 116
    },
    {
      "epoch": 0.3942712721145746,
      "grad_norm": 1.7510665655136108,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.2855,
      "step": 117
    },
    {
      "epoch": 0.3976411120471778,
      "grad_norm": 1.9236868619918823,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.311,
      "step": 118
    },
    {
      "epoch": 0.401010951979781,
      "grad_norm": 2.0435805320739746,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2986,
      "step": 119
    },
    {
      "epoch": 0.4043807919123842,
      "grad_norm": 1.9279718399047852,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3819,
      "step": 120
    },
    {
      "epoch": 0.4077506318449874,
      "grad_norm": 1.8936488628387451,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.3616,
      "step": 121
    },
    {
      "epoch": 0.4111204717775906,
      "grad_norm": 1.8582466840744019,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2168,
      "step": 122
    },
    {
      "epoch": 0.4144903117101938,
      "grad_norm": 1.9604997634887695,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2091,
      "step": 123
    },
    {
      "epoch": 0.41786015164279694,
      "grad_norm": 2.0173380374908447,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.1826,
      "step": 124
    },
    {
      "epoch": 0.42122999157540014,
      "grad_norm": 2.1094472408294678,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.3863,
      "step": 125
    },
    {
      "epoch": 0.42459983150800334,
      "grad_norm": 1.9218369722366333,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.821,
      "step": 126
    },
    {
      "epoch": 0.42796967144060655,
      "grad_norm": 1.9228280782699585,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.1225,
      "step": 127
    },
    {
      "epoch": 0.43133951137320975,
      "grad_norm": 1.993282675743103,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.0312,
      "step": 128
    },
    {
      "epoch": 0.43470935130581295,
      "grad_norm": 1.912155032157898,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.8526,
      "step": 129
    },
    {
      "epoch": 0.43807919123841615,
      "grad_norm": 1.8283792734146118,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.8728,
      "step": 130
    },
    {
      "epoch": 0.44144903117101936,
      "grad_norm": 2.0509607791900635,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1437,
      "step": 131
    },
    {
      "epoch": 0.44481887110362256,
      "grad_norm": 2.023038387298584,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.1974,
      "step": 132
    },
    {
      "epoch": 0.44818871103622576,
      "grad_norm": 1.787955403327942,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.9063,
      "step": 133
    },
    {
      "epoch": 0.45155855096882896,
      "grad_norm": 2.0747554302215576,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.0845,
      "step": 134
    },
    {
      "epoch": 0.45492839090143217,
      "grad_norm": 1.923215389251709,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.8595,
      "step": 135
    },
    {
      "epoch": 0.45829823083403537,
      "grad_norm": 2.1409788131713867,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.4269,
      "step": 136
    },
    {
      "epoch": 0.46166807076663857,
      "grad_norm": 2.012211561203003,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.1905,
      "step": 137
    },
    {
      "epoch": 0.4650379106992418,
      "grad_norm": 1.985556721687317,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.1023,
      "step": 138
    },
    {
      "epoch": 0.468407750631845,
      "grad_norm": 2.3808369636535645,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.3271,
      "step": 139
    },
    {
      "epoch": 0.4717775905644482,
      "grad_norm": 2.1127185821533203,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.1657,
      "step": 140
    },
    {
      "epoch": 0.4751474304970514,
      "grad_norm": 2.1625142097473145,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2499,
      "step": 141
    },
    {
      "epoch": 0.4785172704296546,
      "grad_norm": 2.438977003097534,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.2194,
      "step": 142
    },
    {
      "epoch": 0.4818871103622578,
      "grad_norm": 2.300933361053467,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1324,
      "step": 143
    },
    {
      "epoch": 0.485256950294861,
      "grad_norm": 2.5897982120513916,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.1773,
      "step": 144
    },
    {
      "epoch": 0.4886267902274642,
      "grad_norm": 2.3003430366516113,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.0685,
      "step": 145
    },
    {
      "epoch": 0.4919966301600674,
      "grad_norm": 2.559908151626587,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.0089,
      "step": 146
    },
    {
      "epoch": 0.4953664700926706,
      "grad_norm": 2.485055446624756,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.1171,
      "step": 147
    },
    {
      "epoch": 0.4987363100252738,
      "grad_norm": 2.325895309448242,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1202,
      "step": 148
    },
    {
      "epoch": 0.502106149957877,
      "grad_norm": 2.435990333557129,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.1628,
      "step": 149
    },
    {
      "epoch": 0.5054759898904801,
      "grad_norm": 2.55873441696167,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.0563,
      "step": 150
    },
    {
      "epoch": 0.5054759898904801,
      "eval_loss": 0.2869250476360321,
      "eval_runtime": 121.1344,
      "eval_samples_per_second": 4.128,
      "eval_steps_per_second": 1.032,
      "step": 150
    },
    {
      "epoch": 0.5088458298230834,
      "grad_norm": 2.1456289291381836,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.4068,
      "step": 151
    },
    {
      "epoch": 0.5122156697556866,
      "grad_norm": 1.8963745832443237,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.2149,
      "step": 152
    },
    {
      "epoch": 0.5155855096882898,
      "grad_norm": 1.690665602684021,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.2368,
      "step": 153
    },
    {
      "epoch": 0.518955349620893,
      "grad_norm": 1.5876457691192627,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.0824,
      "step": 154
    },
    {
      "epoch": 0.5223251895534962,
      "grad_norm": 1.5829848051071167,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.1664,
      "step": 155
    },
    {
      "epoch": 0.5256950294860994,
      "grad_norm": 1.556316614151001,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.1729,
      "step": 156
    },
    {
      "epoch": 0.5290648694187026,
      "grad_norm": 1.6850814819335938,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.2733,
      "step": 157
    },
    {
      "epoch": 0.5324347093513058,
      "grad_norm": 1.6765884160995483,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.1687,
      "step": 158
    },
    {
      "epoch": 0.535804549283909,
      "grad_norm": 1.716776967048645,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.327,
      "step": 159
    },
    {
      "epoch": 0.5391743892165122,
      "grad_norm": 1.8959401845932007,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.4084,
      "step": 160
    },
    {
      "epoch": 0.5425442291491154,
      "grad_norm": 1.7681440114974976,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.1947,
      "step": 161
    },
    {
      "epoch": 0.5459140690817186,
      "grad_norm": 1.64078688621521,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0822,
      "step": 162
    },
    {
      "epoch": 0.5492839090143218,
      "grad_norm": 1.666814923286438,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.1674,
      "step": 163
    },
    {
      "epoch": 0.552653748946925,
      "grad_norm": 1.51601243019104,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.074,
      "step": 164
    },
    {
      "epoch": 0.5560235888795282,
      "grad_norm": 1.8284496068954468,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.1988,
      "step": 165
    },
    {
      "epoch": 0.5593934288121314,
      "grad_norm": 1.6202670335769653,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.895,
      "step": 166
    },
    {
      "epoch": 0.5627632687447346,
      "grad_norm": 1.6051350831985474,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.8968,
      "step": 167
    },
    {
      "epoch": 0.5661331086773378,
      "grad_norm": 1.9224071502685547,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.222,
      "step": 168
    },
    {
      "epoch": 0.569502948609941,
      "grad_norm": 1.893166422843933,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.329,
      "step": 169
    },
    {
      "epoch": 0.5728727885425442,
      "grad_norm": 2.0426459312438965,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.3968,
      "step": 170
    },
    {
      "epoch": 0.5762426284751474,
      "grad_norm": 1.878231167793274,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.1258,
      "step": 171
    },
    {
      "epoch": 0.5796124684077506,
      "grad_norm": 1.9137085676193237,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.246,
      "step": 172
    },
    {
      "epoch": 0.5829823083403538,
      "grad_norm": 1.8763216733932495,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.2401,
      "step": 173
    },
    {
      "epoch": 0.586352148272957,
      "grad_norm": 1.8498156070709229,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0621,
      "step": 174
    },
    {
      "epoch": 0.5897219882055602,
      "grad_norm": 1.985643982887268,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.0639,
      "step": 175
    },
    {
      "epoch": 0.5930918281381634,
      "grad_norm": 1.6695940494537354,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.8519,
      "step": 176
    },
    {
      "epoch": 0.5964616680707666,
      "grad_norm": 1.8775286674499512,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.0259,
      "step": 177
    },
    {
      "epoch": 0.5998315080033698,
      "grad_norm": 2.0667293071746826,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.3024,
      "step": 178
    },
    {
      "epoch": 0.603201347935973,
      "grad_norm": 1.8653072118759155,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.8309,
      "step": 179
    },
    {
      "epoch": 0.6065711878685762,
      "grad_norm": 1.793724775314331,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.8761,
      "step": 180
    },
    {
      "epoch": 0.6099410278011794,
      "grad_norm": 1.8889003992080688,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.0853,
      "step": 181
    },
    {
      "epoch": 0.6133108677337826,
      "grad_norm": 1.8247442245483398,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.7295,
      "step": 182
    },
    {
      "epoch": 0.6166807076663858,
      "grad_norm": 2.1758697032928467,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.186,
      "step": 183
    },
    {
      "epoch": 0.620050547598989,
      "grad_norm": 1.765748143196106,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.8961,
      "step": 184
    },
    {
      "epoch": 0.6234203875315922,
      "grad_norm": 2.0728254318237305,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.2267,
      "step": 185
    },
    {
      "epoch": 0.6267902274641954,
      "grad_norm": 2.1173336505889893,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1803,
      "step": 186
    },
    {
      "epoch": 0.6301600673967986,
      "grad_norm": 1.9707828760147095,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.1133,
      "step": 187
    },
    {
      "epoch": 0.6335299073294018,
      "grad_norm": 2.062408924102783,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.9996,
      "step": 188
    },
    {
      "epoch": 0.636899747262005,
      "grad_norm": 2.142023801803589,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.2721,
      "step": 189
    },
    {
      "epoch": 0.6402695871946082,
      "grad_norm": 2.1368825435638428,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.2297,
      "step": 190
    },
    {
      "epoch": 0.6436394271272114,
      "grad_norm": 2.1877987384796143,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.2347,
      "step": 191
    },
    {
      "epoch": 0.6470092670598147,
      "grad_norm": 2.381326913833618,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3015,
      "step": 192
    },
    {
      "epoch": 0.6503791069924179,
      "grad_norm": 2.439781427383423,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1942,
      "step": 193
    },
    {
      "epoch": 0.6537489469250211,
      "grad_norm": 2.1794843673706055,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.9994,
      "step": 194
    },
    {
      "epoch": 0.6571187868576243,
      "grad_norm": 2.6946897506713867,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.8862,
      "step": 195
    },
    {
      "epoch": 0.6604886267902275,
      "grad_norm": 2.566250801086426,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2465,
      "step": 196
    },
    {
      "epoch": 0.6638584667228307,
      "grad_norm": 2.3748836517333984,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.0982,
      "step": 197
    },
    {
      "epoch": 0.6672283066554339,
      "grad_norm": 2.516383171081543,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.1929,
      "step": 198
    },
    {
      "epoch": 0.6705981465880371,
      "grad_norm": 2.4011261463165283,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9815,
      "step": 199
    },
    {
      "epoch": 0.6739679865206403,
      "grad_norm": 2.6091206073760986,
      "learning_rate": 0.0,
      "loss": 1.0339,
      "step": 200
    },
    {
      "epoch": 0.6739679865206403,
      "eval_loss": 0.2749819755554199,
      "eval_runtime": 120.9532,
      "eval_samples_per_second": 4.134,
      "eval_steps_per_second": 1.033,
      "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": 5.673907105636024e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}