{
  "best_metric": 1.3214715719223022,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.2849002849002849,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014245014245014246,
      "grad_norm": 1.0951812267303467,
      "learning_rate": 1e-05,
      "loss": 1.0342,
      "step": 1
    },
    {
      "epoch": 0.0014245014245014246,
      "eval_loss": 3.0569612979888916,
      "eval_runtime": 52.5534,
      "eval_samples_per_second": 22.491,
      "eval_steps_per_second": 5.632,
      "step": 1
    },
    {
      "epoch": 0.002849002849002849,
      "grad_norm": 1.2440255880355835,
      "learning_rate": 2e-05,
      "loss": 1.23,
      "step": 2
    },
    {
      "epoch": 0.004273504273504274,
      "grad_norm": 1.2522715330123901,
      "learning_rate": 3e-05,
      "loss": 1.1836,
      "step": 3
    },
    {
      "epoch": 0.005698005698005698,
      "grad_norm": 1.0097229480743408,
      "learning_rate": 4e-05,
      "loss": 1.1769,
      "step": 4
    },
    {
      "epoch": 0.007122507122507123,
      "grad_norm": 0.8267636895179749,
      "learning_rate": 5e-05,
      "loss": 1.1482,
      "step": 5
    },
    {
      "epoch": 0.008547008547008548,
      "grad_norm": 0.8028005957603455,
      "learning_rate": 6e-05,
      "loss": 1.1643,
      "step": 6
    },
    {
      "epoch": 0.009971509971509971,
      "grad_norm": 0.9496132135391235,
      "learning_rate": 7e-05,
      "loss": 1.1892,
      "step": 7
    },
    {
      "epoch": 0.011396011396011397,
      "grad_norm": 0.907049834728241,
      "learning_rate": 8e-05,
      "loss": 1.1762,
      "step": 8
    },
    {
      "epoch": 0.01282051282051282,
      "grad_norm": 0.8857720494270325,
      "learning_rate": 9e-05,
      "loss": 1.182,
      "step": 9
    },
    {
      "epoch": 0.014245014245014245,
      "grad_norm": 0.904456615447998,
      "learning_rate": 0.0001,
      "loss": 1.0663,
      "step": 10
    },
    {
      "epoch": 0.01566951566951567,
      "grad_norm": 0.9192906618118286,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.1089,
      "step": 11
    },
    {
      "epoch": 0.017094017094017096,
      "grad_norm": 0.8807336688041687,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.1222,
      "step": 12
    },
    {
      "epoch": 0.018518518518518517,
      "grad_norm": 0.9005780816078186,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.0595,
      "step": 13
    },
    {
      "epoch": 0.019943019943019943,
      "grad_norm": 0.8774030208587646,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.1263,
      "step": 14
    },
    {
      "epoch": 0.021367521367521368,
      "grad_norm": 0.8597561717033386,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.1282,
      "step": 15
    },
    {
      "epoch": 0.022792022792022793,
      "grad_norm": 0.8737494945526123,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.085,
      "step": 16
    },
    {
      "epoch": 0.024216524216524215,
      "grad_norm": 0.9524723887443542,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.1558,
      "step": 17
    },
    {
      "epoch": 0.02564102564102564,
      "grad_norm": 1.0169023275375366,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.2095,
      "step": 18
    },
    {
      "epoch": 0.027065527065527065,
      "grad_norm": 0.9679225087165833,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.0824,
      "step": 19
    },
    {
      "epoch": 0.02849002849002849,
      "grad_norm": 1.0456973314285278,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.1306,
      "step": 20
    },
    {
      "epoch": 0.029914529914529916,
      "grad_norm": 1.1445212364196777,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.1833,
      "step": 21
    },
    {
      "epoch": 0.03133903133903134,
      "grad_norm": 1.1573503017425537,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.1789,
      "step": 22
    },
    {
      "epoch": 0.03276353276353276,
      "grad_norm": 1.2056080102920532,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.1198,
      "step": 23
    },
    {
      "epoch": 0.03418803418803419,
      "grad_norm": 1.2797175645828247,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.1591,
      "step": 24
    },
    {
      "epoch": 0.03561253561253561,
      "grad_norm": 1.4011354446411133,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.052,
      "step": 25
    },
    {
      "epoch": 0.037037037037037035,
      "grad_norm": 1.9526931047439575,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.2072,
      "step": 26
    },
    {
      "epoch": 0.038461538461538464,
      "grad_norm": 2.649031400680542,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.2972,
      "step": 27
    },
    {
      "epoch": 0.039886039886039885,
      "grad_norm": 2.6092941761016846,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.2277,
      "step": 28
    },
    {
      "epoch": 0.04131054131054131,
      "grad_norm": 3.3454010486602783,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.4816,
      "step": 29
    },
    {
      "epoch": 0.042735042735042736,
      "grad_norm": 3.7493395805358887,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.5207,
      "step": 30
    },
    {
      "epoch": 0.04415954415954416,
      "grad_norm": 3.1265792846679688,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.2653,
      "step": 31
    },
    {
      "epoch": 0.045584045584045586,
      "grad_norm": 3.3090927600860596,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.3105,
      "step": 32
    },
    {
      "epoch": 0.04700854700854701,
      "grad_norm": 3.3450868129730225,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.4618,
      "step": 33
    },
    {
      "epoch": 0.04843304843304843,
      "grad_norm": 2.2661399841308594,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.0056,
      "step": 34
    },
    {
      "epoch": 0.04985754985754986,
      "grad_norm": 2.910419464111328,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.4462,
      "step": 35
    },
    {
      "epoch": 0.05128205128205128,
      "grad_norm": 2.9667439460754395,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.1563,
      "step": 36
    },
    {
      "epoch": 0.05270655270655271,
      "grad_norm": 3.5099833011627197,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.1158,
      "step": 37
    },
    {
      "epoch": 0.05413105413105413,
      "grad_norm": 3.4474451541900635,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.0648,
      "step": 38
    },
    {
      "epoch": 0.05555555555555555,
      "grad_norm": 4.265199184417725,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.4478,
      "step": 39
    },
    {
      "epoch": 0.05698005698005698,
      "grad_norm": 4.456178665161133,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.1621,
      "step": 40
    },
    {
      "epoch": 0.0584045584045584,
      "grad_norm": 3.706169366836548,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.0857,
      "step": 41
    },
    {
      "epoch": 0.05982905982905983,
      "grad_norm": 4.969032287597656,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.9443,
      "step": 42
    },
    {
      "epoch": 0.06125356125356125,
      "grad_norm": 5.319019794464111,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.2881,
      "step": 43
    },
    {
      "epoch": 0.06267806267806268,
      "grad_norm": 5.219686031341553,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.6457,
      "step": 44
    },
    {
      "epoch": 0.0641025641025641,
      "grad_norm": 4.730462074279785,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.6407,
      "step": 45
    },
    {
      "epoch": 0.06552706552706553,
      "grad_norm": 4.715185165405273,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.9509,
      "step": 46
    },
    {
      "epoch": 0.06695156695156695,
      "grad_norm": 5.63643217086792,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.1613,
      "step": 47
    },
    {
      "epoch": 0.06837606837606838,
      "grad_norm": 4.680100917816162,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.9855,
      "step": 48
    },
    {
      "epoch": 0.0698005698005698,
      "grad_norm": 5.361738204956055,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.4161,
      "step": 49
    },
    {
      "epoch": 0.07122507122507123,
      "grad_norm": 6.914300918579102,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.6899,
      "step": 50
    },
    {
      "epoch": 0.07122507122507123,
      "eval_loss": 1.6694526672363281,
      "eval_runtime": 52.7339,
      "eval_samples_per_second": 22.414,
      "eval_steps_per_second": 5.613,
      "step": 50
    },
    {
      "epoch": 0.07264957264957266,
      "grad_norm": 1.8794810771942139,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.2168,
      "step": 51
    },
    {
      "epoch": 0.07407407407407407,
      "grad_norm": 2.151414632797241,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4221,
      "step": 52
    },
    {
      "epoch": 0.0754985754985755,
      "grad_norm": 1.6555006504058838,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.3112,
      "step": 53
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 1.2223625183105469,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.1976,
      "step": 54
    },
    {
      "epoch": 0.07834757834757834,
      "grad_norm": 1.0195894241333008,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.2126,
      "step": 55
    },
    {
      "epoch": 0.07977207977207977,
      "grad_norm": 0.7386775612831116,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.1169,
      "step": 56
    },
    {
      "epoch": 0.0811965811965812,
      "grad_norm": 0.6235164403915405,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.0829,
      "step": 57
    },
    {
      "epoch": 0.08262108262108261,
      "grad_norm": 0.6764860153198242,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.1481,
      "step": 58
    },
    {
      "epoch": 0.08404558404558404,
      "grad_norm": 0.7909321188926697,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.1837,
      "step": 59
    },
    {
      "epoch": 0.08547008547008547,
      "grad_norm": 0.843760073184967,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.2104,
      "step": 60
    },
    {
      "epoch": 0.0868945868945869,
      "grad_norm": 0.8562942147254944,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.1524,
      "step": 61
    },
    {
      "epoch": 0.08831908831908832,
      "grad_norm": 0.843786895275116,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.1703,
      "step": 62
    },
    {
      "epoch": 0.08974358974358974,
      "grad_norm": 0.7898869514465332,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.1328,
      "step": 63
    },
    {
      "epoch": 0.09116809116809117,
      "grad_norm": 0.768024742603302,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.1505,
      "step": 64
    },
    {
      "epoch": 0.09259259259259259,
      "grad_norm": 0.7674038410186768,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.1166,
      "step": 65
    },
    {
      "epoch": 0.09401709401709402,
      "grad_norm": 0.7880211472511292,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.0265,
      "step": 66
    },
    {
      "epoch": 0.09544159544159544,
      "grad_norm": 0.8146145939826965,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.0635,
      "step": 67
    },
    {
      "epoch": 0.09686609686609686,
      "grad_norm": 0.8939031958580017,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.0924,
      "step": 68
    },
    {
      "epoch": 0.09829059829059829,
      "grad_norm": 0.9344921112060547,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.1032,
      "step": 69
    },
    {
      "epoch": 0.09971509971509972,
      "grad_norm": 1.0251350402832031,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.0943,
      "step": 70
    },
    {
      "epoch": 0.10113960113960115,
      "grad_norm": 1.0262216329574585,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.0996,
      "step": 71
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 1.050524115562439,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.03,
      "step": 72
    },
    {
      "epoch": 0.10398860398860399,
      "grad_norm": 1.1450400352478027,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.0246,
      "step": 73
    },
    {
      "epoch": 0.10541310541310542,
      "grad_norm": 1.1916637420654297,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.9999,
      "step": 74
    },
    {
      "epoch": 0.10683760683760683,
      "grad_norm": 1.4077075719833374,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1149,
      "step": 75
    },
    {
      "epoch": 0.10826210826210826,
      "grad_norm": 1.5274344682693481,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.9944,
      "step": 76
    },
    {
      "epoch": 0.10968660968660969,
      "grad_norm": 2.256171464920044,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.2873,
      "step": 77
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 2.207008123397827,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.0648,
      "step": 78
    },
    {
      "epoch": 0.11253561253561253,
      "grad_norm": 3.2979633808135986,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3021,
      "step": 79
    },
    {
      "epoch": 0.11396011396011396,
      "grad_norm": 2.376344919204712,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.0125,
      "step": 80
    },
    {
      "epoch": 0.11538461538461539,
      "grad_norm": 2.4043941497802734,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.9742,
      "step": 81
    },
    {
      "epoch": 0.1168091168091168,
      "grad_norm": 2.793799877166748,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.1254,
      "step": 82
    },
    {
      "epoch": 0.11823361823361823,
      "grad_norm": 2.7024142742156982,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2951,
      "step": 83
    },
    {
      "epoch": 0.11965811965811966,
      "grad_norm": 3.5849368572235107,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.5779,
      "step": 84
    },
    {
      "epoch": 0.12108262108262108,
      "grad_norm": 3.006340265274048,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.3645,
      "step": 85
    },
    {
      "epoch": 0.1225071225071225,
      "grad_norm": 2.277855157852173,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.0624,
      "step": 86
    },
    {
      "epoch": 0.12393162393162394,
      "grad_norm": 2.4271788597106934,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.5051,
      "step": 87
    },
    {
      "epoch": 0.12535612535612536,
      "grad_norm": 2.7788116931915283,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.7977,
      "step": 88
    },
    {
      "epoch": 0.1267806267806268,
      "grad_norm": 3.2901506423950195,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.9937,
      "step": 89
    },
    {
      "epoch": 0.1282051282051282,
      "grad_norm": 3.284877061843872,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.0343,
      "step": 90
    },
    {
      "epoch": 0.12962962962962962,
      "grad_norm": 3.519197702407837,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.2603,
      "step": 91
    },
    {
      "epoch": 0.13105413105413105,
      "grad_norm": 3.5070576667785645,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.1193,
      "step": 92
    },
    {
      "epoch": 0.13247863247863248,
      "grad_norm": 3.874929904937744,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.1551,
      "step": 93
    },
    {
      "epoch": 0.1339031339031339,
      "grad_norm": 4.789926052093506,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.4054,
      "step": 94
    },
    {
      "epoch": 0.13532763532763534,
      "grad_norm": 5.195937633514404,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.6756,
      "step": 95
    },
    {
      "epoch": 0.13675213675213677,
      "grad_norm": 4.467799663543701,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.3472,
      "step": 96
    },
    {
      "epoch": 0.13817663817663817,
      "grad_norm": 4.928045272827148,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.9974,
      "step": 97
    },
    {
      "epoch": 0.1396011396011396,
      "grad_norm": 5.623568058013916,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.399,
      "step": 98
    },
    {
      "epoch": 0.14102564102564102,
      "grad_norm": 5.925526142120361,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.9024,
      "step": 99
    },
    {
      "epoch": 0.14245014245014245,
      "grad_norm": 7.447940826416016,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.841,
      "step": 100
    },
    {
      "epoch": 0.14245014245014245,
      "eval_loss": 1.5963640213012695,
      "eval_runtime": 52.7393,
      "eval_samples_per_second": 22.412,
      "eval_steps_per_second": 5.613,
      "step": 100
    },
    {
      "epoch": 0.14387464387464388,
      "grad_norm": 1.4451898336410522,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.0918,
      "step": 101
    },
    {
      "epoch": 0.1452991452991453,
      "grad_norm": 1.5431982278823853,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.3001,
      "step": 102
    },
    {
      "epoch": 0.1467236467236467,
      "grad_norm": 1.4764037132263184,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.3387,
      "step": 103
    },
    {
      "epoch": 0.14814814814814814,
      "grad_norm": 1.3215073347091675,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.3939,
      "step": 104
    },
    {
      "epoch": 0.14957264957264957,
      "grad_norm": 0.9795156121253967,
      "learning_rate": 5e-05,
      "loss": 1.195,
      "step": 105
    },
    {
      "epoch": 0.150997150997151,
      "grad_norm": 0.8231992721557617,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2011,
      "step": 106
    },
    {
      "epoch": 0.15242165242165243,
      "grad_norm": 0.7669458985328674,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2633,
      "step": 107
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 0.7166056632995605,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.1905,
      "step": 108
    },
    {
      "epoch": 0.15527065527065528,
      "grad_norm": 0.6721870303153992,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.1175,
      "step": 109
    },
    {
      "epoch": 0.15669515669515668,
      "grad_norm": 0.6732422709465027,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1489,
      "step": 110
    },
    {
      "epoch": 0.1581196581196581,
      "grad_norm": 0.658916175365448,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.1334,
      "step": 111
    },
    {
      "epoch": 0.15954415954415954,
      "grad_norm": 0.6812751889228821,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.1186,
      "step": 112
    },
    {
      "epoch": 0.16096866096866097,
      "grad_norm": 0.7162421941757202,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.1049,
      "step": 113
    },
    {
      "epoch": 0.1623931623931624,
      "grad_norm": 0.7175534963607788,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.0275,
      "step": 114
    },
    {
      "epoch": 0.16381766381766383,
      "grad_norm": 0.8070324659347534,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1026,
      "step": 115
    },
    {
      "epoch": 0.16524216524216523,
      "grad_norm": 0.7852011322975159,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.0646,
      "step": 116
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.8045965433120728,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.9373,
      "step": 117
    },
    {
      "epoch": 0.16809116809116809,
      "grad_norm": 0.8931590914726257,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.1565,
      "step": 118
    },
    {
      "epoch": 0.16951566951566951,
      "grad_norm": 0.9178715348243713,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0861,
      "step": 119
    },
    {
      "epoch": 0.17094017094017094,
      "grad_norm": 1.0241934061050415,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.1994,
      "step": 120
    },
    {
      "epoch": 0.17236467236467237,
      "grad_norm": 1.0514984130859375,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.0728,
      "step": 121
    },
    {
      "epoch": 0.1737891737891738,
      "grad_norm": 1.114422082901001,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.12,
      "step": 122
    },
    {
      "epoch": 0.1752136752136752,
      "grad_norm": 1.0793527364730835,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.0203,
      "step": 123
    },
    {
      "epoch": 0.17663817663817663,
      "grad_norm": 1.1782469749450684,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.1031,
      "step": 124
    },
    {
      "epoch": 0.17806267806267806,
      "grad_norm": 1.4030028581619263,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.0422,
      "step": 125
    },
    {
      "epoch": 0.1794871794871795,
      "grad_norm": 1.6754884719848633,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.1101,
      "step": 126
    },
    {
      "epoch": 0.18091168091168092,
      "grad_norm": 1.6748223304748535,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.9788,
      "step": 127
    },
    {
      "epoch": 0.18233618233618235,
      "grad_norm": 2.2540535926818848,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.132,
      "step": 128
    },
    {
      "epoch": 0.18376068376068377,
      "grad_norm": 2.2037203311920166,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.0734,
      "step": 129
    },
    {
      "epoch": 0.18518518518518517,
      "grad_norm": 2.2696244716644287,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.0225,
      "step": 130
    },
    {
      "epoch": 0.1866096866096866,
      "grad_norm": 2.606853723526001,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.3488,
      "step": 131
    },
    {
      "epoch": 0.18803418803418803,
      "grad_norm": 3.2537319660186768,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.2927,
      "step": 132
    },
    {
      "epoch": 0.18945868945868946,
      "grad_norm": 2.4283738136291504,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1179,
      "step": 133
    },
    {
      "epoch": 0.1908831908831909,
      "grad_norm": 2.3190572261810303,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.2731,
      "step": 134
    },
    {
      "epoch": 0.19230769230769232,
      "grad_norm": 2.494070291519165,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2296,
      "step": 135
    },
    {
      "epoch": 0.19373219373219372,
      "grad_norm": 2.373368740081787,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.6222,
      "step": 136
    },
    {
      "epoch": 0.19515669515669515,
      "grad_norm": 2.0916032791137695,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.3034,
      "step": 137
    },
    {
      "epoch": 0.19658119658119658,
      "grad_norm": 2.3026416301727295,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.8082,
      "step": 138
    },
    {
      "epoch": 0.198005698005698,
      "grad_norm": 2.6751773357391357,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.1713,
      "step": 139
    },
    {
      "epoch": 0.19943019943019943,
      "grad_norm": 3.263648748397827,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.9226,
      "step": 140
    },
    {
      "epoch": 0.20085470085470086,
      "grad_norm": 2.890190839767456,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.9661,
      "step": 141
    },
    {
      "epoch": 0.2022792022792023,
      "grad_norm": 3.3623650074005127,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.9155,
      "step": 142
    },
    {
      "epoch": 0.2037037037037037,
      "grad_norm": 3.7464234828948975,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.2982,
      "step": 143
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 3.9489314556121826,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.0735,
      "step": 144
    },
    {
      "epoch": 0.20655270655270655,
      "grad_norm": 4.435914993286133,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.0603,
      "step": 145
    },
    {
      "epoch": 0.20797720797720798,
      "grad_norm": 5.1594743728637695,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.2006,
      "step": 146
    },
    {
      "epoch": 0.2094017094017094,
      "grad_norm": 4.007395267486572,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.056,
      "step": 147
    },
    {
      "epoch": 0.21082621082621084,
      "grad_norm": 5.316275119781494,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.989,
      "step": 148
    },
    {
      "epoch": 0.21225071225071226,
      "grad_norm": 5.518270015716553,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.973,
      "step": 149
    },
    {
      "epoch": 0.21367521367521367,
      "grad_norm": 5.537274360656738,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.084,
      "step": 150
    },
    {
      "epoch": 0.21367521367521367,
      "eval_loss": 1.34799063205719,
      "eval_runtime": 52.7878,
      "eval_samples_per_second": 22.392,
      "eval_steps_per_second": 5.607,
      "step": 150
    },
    {
      "epoch": 0.2150997150997151,
      "grad_norm": 0.4327431321144104,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.7104,
      "step": 151
    },
    {
      "epoch": 0.21652421652421652,
      "grad_norm": 0.5074938535690308,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.9456,
      "step": 152
    },
    {
      "epoch": 0.21794871794871795,
      "grad_norm": 0.510260283946991,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.0321,
      "step": 153
    },
    {
      "epoch": 0.21937321937321938,
      "grad_norm": 0.510598361492157,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.9568,
      "step": 154
    },
    {
      "epoch": 0.2207977207977208,
      "grad_norm": 0.5640057325363159,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.0487,
      "step": 155
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.5703732371330261,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.1002,
      "step": 156
    },
    {
      "epoch": 0.22364672364672364,
      "grad_norm": 0.5797698497772217,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.0821,
      "step": 157
    },
    {
      "epoch": 0.22507122507122507,
      "grad_norm": 0.5938100814819336,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.0382,
      "step": 158
    },
    {
      "epoch": 0.2264957264957265,
      "grad_norm": 0.6127101182937622,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.1586,
      "step": 159
    },
    {
      "epoch": 0.22792022792022792,
      "grad_norm": 0.6117282509803772,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.0912,
      "step": 160
    },
    {
      "epoch": 0.22934472934472935,
      "grad_norm": 0.6170830726623535,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.0151,
      "step": 161
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 0.6764513850212097,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.9979,
      "step": 162
    },
    {
      "epoch": 0.23219373219373218,
      "grad_norm": 0.7066622376441956,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.1132,
      "step": 163
    },
    {
      "epoch": 0.2336182336182336,
      "grad_norm": 0.7205550670623779,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.9935,
      "step": 164
    },
    {
      "epoch": 0.23504273504273504,
      "grad_norm": 0.7248603105545044,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.0452,
      "step": 165
    },
    {
      "epoch": 0.23646723646723647,
      "grad_norm": 0.758104145526886,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.125,
      "step": 166
    },
    {
      "epoch": 0.2378917378917379,
      "grad_norm": 0.7871737480163574,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.0546,
      "step": 167
    },
    {
      "epoch": 0.23931623931623933,
      "grad_norm": 0.8279024362564087,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.0099,
      "step": 168
    },
    {
      "epoch": 0.24074074074074073,
      "grad_norm": 0.8020943403244019,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.0516,
      "step": 169
    },
    {
      "epoch": 0.24216524216524216,
      "grad_norm": 0.9537612795829773,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.1207,
      "step": 170
    },
    {
      "epoch": 0.24358974358974358,
      "grad_norm": 0.9706317782402039,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.0968,
      "step": 171
    },
    {
      "epoch": 0.245014245014245,
      "grad_norm": 1.0593055486679077,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.1436,
      "step": 172
    },
    {
      "epoch": 0.24643874643874644,
      "grad_norm": 1.0807327032089233,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.0871,
      "step": 173
    },
    {
      "epoch": 0.24786324786324787,
      "grad_norm": 1.1584875583648682,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.1056,
      "step": 174
    },
    {
      "epoch": 0.2492877492877493,
      "grad_norm": 1.2232173681259155,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.0358,
      "step": 175
    },
    {
      "epoch": 0.25071225071225073,
      "grad_norm": 1.4096721410751343,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.0462,
      "step": 176
    },
    {
      "epoch": 0.25213675213675213,
      "grad_norm": 1.533095359802246,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.02,
      "step": 177
    },
    {
      "epoch": 0.2535612535612536,
      "grad_norm": 1.7267961502075195,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.8788,
      "step": 178
    },
    {
      "epoch": 0.254985754985755,
      "grad_norm": 2.938230037689209,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.3451,
      "step": 179
    },
    {
      "epoch": 0.2564102564102564,
      "grad_norm": 2.432194709777832,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.0302,
      "step": 180
    },
    {
      "epoch": 0.25783475783475784,
      "grad_norm": 2.264803171157837,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.1197,
      "step": 181
    },
    {
      "epoch": 0.25925925925925924,
      "grad_norm": 2.641390562057495,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.1154,
      "step": 182
    },
    {
      "epoch": 0.2606837606837607,
      "grad_norm": 2.440976858139038,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.0344,
      "step": 183
    },
    {
      "epoch": 0.2621082621082621,
      "grad_norm": 2.363471269607544,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.3344,
      "step": 184
    },
    {
      "epoch": 0.26353276353276356,
      "grad_norm": 2.391676187515259,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.0026,
      "step": 185
    },
    {
      "epoch": 0.26495726495726496,
      "grad_norm": 1.7763724327087402,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.0482,
      "step": 186
    },
    {
      "epoch": 0.26638176638176636,
      "grad_norm": 1.8221395015716553,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.5048,
      "step": 187
    },
    {
      "epoch": 0.2678062678062678,
      "grad_norm": 2.223668336868286,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.6937,
      "step": 188
    },
    {
      "epoch": 0.2692307692307692,
      "grad_norm": 2.2284491062164307,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.7243,
      "step": 189
    },
    {
      "epoch": 0.2706552706552707,
      "grad_norm": 2.6991491317749023,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.2039,
      "step": 190
    },
    {
      "epoch": 0.2720797720797721,
      "grad_norm": 2.7718915939331055,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.988,
      "step": 191
    },
    {
      "epoch": 0.27350427350427353,
      "grad_norm": 3.3055479526519775,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.8932,
      "step": 192
    },
    {
      "epoch": 0.27492877492877493,
      "grad_norm": 3.443882942199707,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.1736,
      "step": 193
    },
    {
      "epoch": 0.27635327635327633,
      "grad_norm": 4.120036602020264,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.3453,
      "step": 194
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 3.9351887702941895,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.7705,
      "step": 195
    },
    {
      "epoch": 0.2792022792022792,
      "grad_norm": 4.2510833740234375,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.8898,
      "step": 196
    },
    {
      "epoch": 0.28062678062678065,
      "grad_norm": 4.727863311767578,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.5095,
      "step": 197
    },
    {
      "epoch": 0.28205128205128205,
      "grad_norm": 4.397941589355469,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.0984,
      "step": 198
    },
    {
      "epoch": 0.28347578347578345,
      "grad_norm": 4.505115985870361,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.9002,
      "step": 199
    },
    {
      "epoch": 0.2849002849002849,
      "grad_norm": 6.337467193603516,
      "learning_rate": 0.0,
      "loss": 1.9206,
      "step": 200
    },
    {
      "epoch": 0.2849002849002849,
      "eval_loss": 1.3214715719223022,
      "eval_runtime": 52.7922,
      "eval_samples_per_second": 22.39,
      "eval_steps_per_second": 5.607,
      "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": 8.3184381232939e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}