{
  "best_metric": 0.08608997613191605,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.08994827973914998,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00044974139869574995,
      "grad_norm": 11.999661445617676,
      "learning_rate": 1e-05,
      "loss": 10.6173,
      "step": 1
    },
    {
      "epoch": 0.00044974139869574995,
      "eval_loss": 2.5602872371673584,
      "eval_runtime": 201.9893,
      "eval_samples_per_second": 18.541,
      "eval_steps_per_second": 4.639,
      "step": 1
    },
    {
      "epoch": 0.0008994827973914999,
      "grad_norm": 18.244874954223633,
      "learning_rate": 2e-05,
      "loss": 14.4428,
      "step": 2
    },
    {
      "epoch": 0.0013492241960872497,
      "grad_norm": 18.10195541381836,
      "learning_rate": 3e-05,
      "loss": 14.5523,
      "step": 3
    },
    {
      "epoch": 0.0017989655947829998,
      "grad_norm": 16.8906192779541,
      "learning_rate": 4e-05,
      "loss": 15.1014,
      "step": 4
    },
    {
      "epoch": 0.00224870699347875,
      "grad_norm": 20.2750301361084,
      "learning_rate": 5e-05,
      "loss": 15.0885,
      "step": 5
    },
    {
      "epoch": 0.0026984483921744995,
      "grad_norm": 19.730945587158203,
      "learning_rate": 6e-05,
      "loss": 15.1841,
      "step": 6
    },
    {
      "epoch": 0.0031481897908702495,
      "grad_norm": 26.96023941040039,
      "learning_rate": 7e-05,
      "loss": 15.6539,
      "step": 7
    },
    {
      "epoch": 0.0035979311895659996,
      "grad_norm": 14.041892051696777,
      "learning_rate": 8e-05,
      "loss": 9.3399,
      "step": 8
    },
    {
      "epoch": 0.00404767258826175,
      "grad_norm": 12.261756896972656,
      "learning_rate": 9e-05,
      "loss": 3.4857,
      "step": 9
    },
    {
      "epoch": 0.0044974139869575,
      "grad_norm": 11.446184158325195,
      "learning_rate": 0.0001,
      "loss": 2.7381,
      "step": 10
    },
    {
      "epoch": 0.00494715538565325,
      "grad_norm": 6.41536283493042,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.8225,
      "step": 11
    },
    {
      "epoch": 0.005396896784348999,
      "grad_norm": 5.386502742767334,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.8364,
      "step": 12
    },
    {
      "epoch": 0.005846638183044749,
      "grad_norm": 4.892361640930176,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.2371,
      "step": 13
    },
    {
      "epoch": 0.006296379581740499,
      "grad_norm": 3.4953689575195312,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.9307,
      "step": 14
    },
    {
      "epoch": 0.006746120980436249,
      "grad_norm": 2.8037469387054443,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.8449,
      "step": 15
    },
    {
      "epoch": 0.007195862379131999,
      "grad_norm": 3.504249095916748,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.5873,
      "step": 16
    },
    {
      "epoch": 0.007645603777827749,
      "grad_norm": 2.9529964923858643,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.6727,
      "step": 17
    },
    {
      "epoch": 0.0080953451765235,
      "grad_norm": 2.2594285011291504,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.5205,
      "step": 18
    },
    {
      "epoch": 0.00854508657521925,
      "grad_norm": 2.9907758235931396,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.7388,
      "step": 19
    },
    {
      "epoch": 0.008994827973915,
      "grad_norm": 2.6715469360351562,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.522,
      "step": 20
    },
    {
      "epoch": 0.00944456937261075,
      "grad_norm": 4.26343297958374,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.6605,
      "step": 21
    },
    {
      "epoch": 0.0098943107713065,
      "grad_norm": 3.155764102935791,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.732,
      "step": 22
    },
    {
      "epoch": 0.01034405217000225,
      "grad_norm": 2.41500186920166,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.6712,
      "step": 23
    },
    {
      "epoch": 0.010793793568697998,
      "grad_norm": 1.7916475534439087,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.5122,
      "step": 24
    },
    {
      "epoch": 0.011243534967393748,
      "grad_norm": 1.9910305738449097,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.6616,
      "step": 25
    },
    {
      "epoch": 0.011693276366089498,
      "grad_norm": 2.8561861515045166,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.6131,
      "step": 26
    },
    {
      "epoch": 0.012143017764785248,
      "grad_norm": 6.196378707885742,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.7687,
      "step": 27
    },
    {
      "epoch": 0.012592759163480998,
      "grad_norm": 9.783409118652344,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.5217,
      "step": 28
    },
    {
      "epoch": 0.013042500562176748,
      "grad_norm": 1.8568822145462036,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.3759,
      "step": 29
    },
    {
      "epoch": 0.013492241960872498,
      "grad_norm": 2.173732280731201,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.9058,
      "step": 30
    },
    {
      "epoch": 0.013941983359568248,
      "grad_norm": 2.3693814277648926,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.8297,
      "step": 31
    },
    {
      "epoch": 0.014391724758263998,
      "grad_norm": 4.888921737670898,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.0706,
      "step": 32
    },
    {
      "epoch": 0.014841466156959748,
      "grad_norm": 5.716813564300537,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.0319,
      "step": 33
    },
    {
      "epoch": 0.015291207555655498,
      "grad_norm": 2.7340168952941895,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.4382,
      "step": 34
    },
    {
      "epoch": 0.01574094895435125,
      "grad_norm": 2.036818027496338,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.36,
      "step": 35
    },
    {
      "epoch": 0.016190690353047,
      "grad_norm": 1.1884934902191162,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.2437,
      "step": 36
    },
    {
      "epoch": 0.01664043175174275,
      "grad_norm": 1.4083915948867798,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.2313,
      "step": 37
    },
    {
      "epoch": 0.0170901731504385,
      "grad_norm": 0.8769119381904602,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.077,
      "step": 38
    },
    {
      "epoch": 0.01753991454913425,
      "grad_norm": 3.0471885204315186,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.3445,
      "step": 39
    },
    {
      "epoch": 0.01798965594783,
      "grad_norm": 1.4649059772491455,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.2184,
      "step": 40
    },
    {
      "epoch": 0.01843939734652575,
      "grad_norm": 3.399675130844116,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.1999,
      "step": 41
    },
    {
      "epoch": 0.0188891387452215,
      "grad_norm": 1.1934828758239746,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.1314,
      "step": 42
    },
    {
      "epoch": 0.01933888014391725,
      "grad_norm": 0.9715436697006226,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.1467,
      "step": 43
    },
    {
      "epoch": 0.019788621542613,
      "grad_norm": 1.0965895652770996,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.1272,
      "step": 44
    },
    {
      "epoch": 0.02023836294130875,
      "grad_norm": 1.6038917303085327,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.2163,
      "step": 45
    },
    {
      "epoch": 0.0206881043400045,
      "grad_norm": 1.843549370765686,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.2697,
      "step": 46
    },
    {
      "epoch": 0.021137845738700246,
      "grad_norm": 1.1421908140182495,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.0914,
      "step": 47
    },
    {
      "epoch": 0.021587587137395996,
      "grad_norm": 1.1461985111236572,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.1423,
      "step": 48
    },
    {
      "epoch": 0.022037328536091746,
      "grad_norm": 1.065673828125,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.1579,
      "step": 49
    },
    {
      "epoch": 0.022487069934787496,
      "grad_norm": 2.8399734497070312,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4404,
      "step": 50
    },
    {
      "epoch": 0.022487069934787496,
      "eval_loss": 0.4560151696205139,
      "eval_runtime": 203.2767,
      "eval_samples_per_second": 18.423,
      "eval_steps_per_second": 4.609,
      "step": 50
    },
    {
      "epoch": 0.022936811333483246,
      "grad_norm": 12.12181282043457,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.5931,
      "step": 51
    },
    {
      "epoch": 0.023386552732178996,
      "grad_norm": 10.235139846801758,
      "learning_rate": 8.842005554284296e-05,
      "loss": 5.0561,
      "step": 52
    },
    {
      "epoch": 0.023836294130874746,
      "grad_norm": 14.308311462402344,
      "learning_rate": 8.788574348801675e-05,
      "loss": 5.1274,
      "step": 53
    },
    {
      "epoch": 0.024286035529570496,
      "grad_norm": 10.70067310333252,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.6855,
      "step": 54
    },
    {
      "epoch": 0.024735776928266246,
      "grad_norm": 37.494049072265625,
      "learning_rate": 8.678619553365659e-05,
      "loss": 5.1325,
      "step": 55
    },
    {
      "epoch": 0.025185518326961996,
      "grad_norm": 11.039992332458496,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.4534,
      "step": 56
    },
    {
      "epoch": 0.025635259725657746,
      "grad_norm": 11.280861854553223,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.2581,
      "step": 57
    },
    {
      "epoch": 0.026085001124353496,
      "grad_norm": 12.357115745544434,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.9033,
      "step": 58
    },
    {
      "epoch": 0.026534742523049246,
      "grad_norm": 5.195148944854736,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.7927,
      "step": 59
    },
    {
      "epoch": 0.026984483921744996,
      "grad_norm": 4.997382640838623,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.2314,
      "step": 60
    },
    {
      "epoch": 0.027434225320440746,
      "grad_norm": 6.4407806396484375,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.5699,
      "step": 61
    },
    {
      "epoch": 0.027883966719136496,
      "grad_norm": 3.009465456008911,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.6786,
      "step": 62
    },
    {
      "epoch": 0.028333708117832247,
      "grad_norm": 1.9331899881362915,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.5576,
      "step": 63
    },
    {
      "epoch": 0.028783449516527997,
      "grad_norm": 1.6227545738220215,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.5796,
      "step": 64
    },
    {
      "epoch": 0.029233190915223747,
      "grad_norm": 1.3804194927215576,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.3511,
      "step": 65
    },
    {
      "epoch": 0.029682932313919497,
      "grad_norm": 1.6237164735794067,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.4173,
      "step": 66
    },
    {
      "epoch": 0.030132673712615247,
      "grad_norm": 1.5237326622009277,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.4576,
      "step": 67
    },
    {
      "epoch": 0.030582415111310997,
      "grad_norm": 1.6522092819213867,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3648,
      "step": 68
    },
    {
      "epoch": 0.031032156510006747,
      "grad_norm": 1.2098571062088013,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.2397,
      "step": 69
    },
    {
      "epoch": 0.0314818979087025,
      "grad_norm": 1.4940675497055054,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.3326,
      "step": 70
    },
    {
      "epoch": 0.031931639307398244,
      "grad_norm": 1.5978797674179077,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.4967,
      "step": 71
    },
    {
      "epoch": 0.032381380706094,
      "grad_norm": 0.9962940812110901,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.237,
      "step": 72
    },
    {
      "epoch": 0.032831122104789744,
      "grad_norm": 1.1361626386642456,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.3214,
      "step": 73
    },
    {
      "epoch": 0.0332808635034855,
      "grad_norm": 1.6479883193969727,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.3604,
      "step": 74
    },
    {
      "epoch": 0.033730604902181244,
      "grad_norm": 1.6413830518722534,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.537,
      "step": 75
    },
    {
      "epoch": 0.034180346300877,
      "grad_norm": 1.4625543355941772,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.3052,
      "step": 76
    },
    {
      "epoch": 0.034630087699572744,
      "grad_norm": 1.5516142845153809,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.2548,
      "step": 77
    },
    {
      "epoch": 0.0350798290982685,
      "grad_norm": 1.256709337234497,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.3805,
      "step": 78
    },
    {
      "epoch": 0.035529570496964244,
      "grad_norm": 1.1262892484664917,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.2461,
      "step": 79
    },
    {
      "epoch": 0.03597931189566,
      "grad_norm": 1.6752142906188965,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.438,
      "step": 80
    },
    {
      "epoch": 0.036429053294355744,
      "grad_norm": 3.0445713996887207,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.9077,
      "step": 81
    },
    {
      "epoch": 0.0368787946930515,
      "grad_norm": 2.7846665382385254,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.6396,
      "step": 82
    },
    {
      "epoch": 0.037328536091747244,
      "grad_norm": 2.3965539932250977,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.6681,
      "step": 83
    },
    {
      "epoch": 0.037778277490443,
      "grad_norm": 1.8316521644592285,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.421,
      "step": 84
    },
    {
      "epoch": 0.038228018889138744,
      "grad_norm": 1.8929729461669922,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.2464,
      "step": 85
    },
    {
      "epoch": 0.0386777602878345,
      "grad_norm": 0.9921917915344238,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.199,
      "step": 86
    },
    {
      "epoch": 0.039127501686530244,
      "grad_norm": 0.5218179225921631,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.1281,
      "step": 87
    },
    {
      "epoch": 0.039577243085226,
      "grad_norm": 0.7603991627693176,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3143,
      "step": 88
    },
    {
      "epoch": 0.040026984483921745,
      "grad_norm": 0.9146785736083984,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.1797,
      "step": 89
    },
    {
      "epoch": 0.0404767258826175,
      "grad_norm": 3.629509449005127,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.4972,
      "step": 90
    },
    {
      "epoch": 0.040926467281313245,
      "grad_norm": 1.9315358400344849,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.3064,
      "step": 91
    },
    {
      "epoch": 0.041376208680009,
      "grad_norm": 0.860022783279419,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.171,
      "step": 92
    },
    {
      "epoch": 0.041825950078704745,
      "grad_norm": 0.549447774887085,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.1401,
      "step": 93
    },
    {
      "epoch": 0.04227569147740049,
      "grad_norm": 0.5060736536979675,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.1453,
      "step": 94
    },
    {
      "epoch": 0.042725432876096245,
      "grad_norm": 0.45919734239578247,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.0682,
      "step": 95
    },
    {
      "epoch": 0.04317517427479199,
      "grad_norm": 0.3154193162918091,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.1073,
      "step": 96
    },
    {
      "epoch": 0.043624915673487745,
      "grad_norm": 0.38796675205230713,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.121,
      "step": 97
    },
    {
      "epoch": 0.04407465707218349,
      "grad_norm": 1.1888880729675293,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3309,
      "step": 98
    },
    {
      "epoch": 0.044524398470879245,
      "grad_norm": 0.901296079158783,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.296,
      "step": 99
    },
    {
      "epoch": 0.04497413986957499,
      "grad_norm": 0.6436352729797363,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3143,
      "step": 100
    },
    {
      "epoch": 0.04497413986957499,
      "eval_loss": 0.15746021270751953,
      "eval_runtime": 203.3918,
      "eval_samples_per_second": 18.413,
      "eval_steps_per_second": 4.607,
      "step": 100
    },
    {
      "epoch": 0.045423881268270745,
      "grad_norm": 4.299271583557129,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.7905,
      "step": 101
    },
    {
      "epoch": 0.04587362266696649,
      "grad_norm": 6.018694877624512,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.389,
      "step": 102
    },
    {
      "epoch": 0.046323364065662245,
      "grad_norm": 4.68121862411499,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.5465,
      "step": 103
    },
    {
      "epoch": 0.04677310546435799,
      "grad_norm": 5.616169452667236,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.1954,
      "step": 104
    },
    {
      "epoch": 0.047222846863053745,
      "grad_norm": 4.181501388549805,
      "learning_rate": 5e-05,
      "loss": 1.5948,
      "step": 105
    },
    {
      "epoch": 0.04767258826174949,
      "grad_norm": 4.1742329597473145,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2398,
      "step": 106
    },
    {
      "epoch": 0.048122329660445246,
      "grad_norm": 5.769238471984863,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.8842,
      "step": 107
    },
    {
      "epoch": 0.04857207105914099,
      "grad_norm": 1.6297969818115234,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.4904,
      "step": 108
    },
    {
      "epoch": 0.049021812457836746,
      "grad_norm": 2.725541353225708,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4696,
      "step": 109
    },
    {
      "epoch": 0.04947155385653249,
      "grad_norm": 3.012354850769043,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.6448,
      "step": 110
    },
    {
      "epoch": 0.049921295255228246,
      "grad_norm": 1.756006121635437,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.733,
      "step": 111
    },
    {
      "epoch": 0.05037103665392399,
      "grad_norm": 1.4145863056182861,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.5818,
      "step": 112
    },
    {
      "epoch": 0.050820778052619746,
      "grad_norm": 1.603208065032959,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.6479,
      "step": 113
    },
    {
      "epoch": 0.05127051945131549,
      "grad_norm": 1.4745620489120483,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3272,
      "step": 114
    },
    {
      "epoch": 0.051720260850011246,
      "grad_norm": 0.8649131655693054,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2336,
      "step": 115
    },
    {
      "epoch": 0.05217000224870699,
      "grad_norm": 1.5313912630081177,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3927,
      "step": 116
    },
    {
      "epoch": 0.052619743647402746,
      "grad_norm": 2.051571846008301,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3947,
      "step": 117
    },
    {
      "epoch": 0.05306948504609849,
      "grad_norm": 1.0870023965835571,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.3093,
      "step": 118
    },
    {
      "epoch": 0.053519226444794246,
      "grad_norm": 0.9088997840881348,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.1957,
      "step": 119
    },
    {
      "epoch": 0.05396896784348999,
      "grad_norm": 1.3456634283065796,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.1694,
      "step": 120
    },
    {
      "epoch": 0.054418709242185746,
      "grad_norm": 1.5115083456039429,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.2486,
      "step": 121
    },
    {
      "epoch": 0.05486845064088149,
      "grad_norm": 0.9957089424133301,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3279,
      "step": 122
    },
    {
      "epoch": 0.055318192039577246,
      "grad_norm": 0.871972382068634,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.2601,
      "step": 123
    },
    {
      "epoch": 0.05576793343827299,
      "grad_norm": 1.3917059898376465,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.3278,
      "step": 124
    },
    {
      "epoch": 0.05621767483696874,
      "grad_norm": 1.323182463645935,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.2366,
      "step": 125
    },
    {
      "epoch": 0.05666741623566449,
      "grad_norm": 1.936692476272583,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.2676,
      "step": 126
    },
    {
      "epoch": 0.05711715763436024,
      "grad_norm": 2.8535847663879395,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.2812,
      "step": 127
    },
    {
      "epoch": 0.05756689903305599,
      "grad_norm": 1.051253318786621,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.1826,
      "step": 128
    },
    {
      "epoch": 0.05801664043175174,
      "grad_norm": 1.3316032886505127,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.2051,
      "step": 129
    },
    {
      "epoch": 0.05846638183044749,
      "grad_norm": 1.0143500566482544,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3097,
      "step": 130
    },
    {
      "epoch": 0.05891612322914324,
      "grad_norm": 1.079813003540039,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.2279,
      "step": 131
    },
    {
      "epoch": 0.05936586462783899,
      "grad_norm": 1.3066439628601074,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.3842,
      "step": 132
    },
    {
      "epoch": 0.05981560602653474,
      "grad_norm": 1.9721635580062866,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.4572,
      "step": 133
    },
    {
      "epoch": 0.060265347425230494,
      "grad_norm": 2.701054811477661,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.4551,
      "step": 134
    },
    {
      "epoch": 0.06071508882392624,
      "grad_norm": 1.252236247062683,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.2149,
      "step": 135
    },
    {
      "epoch": 0.061164830222621994,
      "grad_norm": 0.8284734487533569,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.2108,
      "step": 136
    },
    {
      "epoch": 0.06161457162131774,
      "grad_norm": 0.5045490264892578,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.0899,
      "step": 137
    },
    {
      "epoch": 0.062064313020013494,
      "grad_norm": 0.4537599980831146,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.1139,
      "step": 138
    },
    {
      "epoch": 0.06251405441870925,
      "grad_norm": 0.8299667835235596,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0637,
      "step": 139
    },
    {
      "epoch": 0.062963795817405,
      "grad_norm": 0.8221279382705688,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.284,
      "step": 140
    },
    {
      "epoch": 0.06341353721610074,
      "grad_norm": 1.9402421712875366,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.1774,
      "step": 141
    },
    {
      "epoch": 0.06386327861479649,
      "grad_norm": 0.7924624681472778,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.1697,
      "step": 142
    },
    {
      "epoch": 0.06431302001349225,
      "grad_norm": 0.6073143482208252,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.2193,
      "step": 143
    },
    {
      "epoch": 0.064762761412188,
      "grad_norm": 0.7139266729354858,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.1481,
      "step": 144
    },
    {
      "epoch": 0.06521250281088374,
      "grad_norm": 1.0343050956726074,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.1871,
      "step": 145
    },
    {
      "epoch": 0.06566224420957949,
      "grad_norm": 0.37282031774520874,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.0981,
      "step": 146
    },
    {
      "epoch": 0.06611198560827525,
      "grad_norm": 0.6088612079620361,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.1449,
      "step": 147
    },
    {
      "epoch": 0.066561727006971,
      "grad_norm": 1.0094534158706665,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2721,
      "step": 148
    },
    {
      "epoch": 0.06701146840566674,
      "grad_norm": 0.9021718502044678,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.2798,
      "step": 149
    },
    {
      "epoch": 0.06746120980436249,
      "grad_norm": 0.5983198285102844,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.3237,
      "step": 150
    },
    {
      "epoch": 0.06746120980436249,
      "eval_loss": 0.09369870275259018,
      "eval_runtime": 203.3624,
      "eval_samples_per_second": 18.415,
      "eval_steps_per_second": 4.608,
      "step": 150
    },
    {
      "epoch": 0.06791095120305825,
      "grad_norm": 4.786923885345459,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.4455,
      "step": 151
    },
    {
      "epoch": 0.068360692601754,
      "grad_norm": 3.380749464035034,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.3601,
      "step": 152
    },
    {
      "epoch": 0.06881043400044974,
      "grad_norm": 2.7484781742095947,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.7531,
      "step": 153
    },
    {
      "epoch": 0.06926017539914549,
      "grad_norm": 4.33957576751709,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.5817,
      "step": 154
    },
    {
      "epoch": 0.06970991679784125,
      "grad_norm": 2.7756197452545166,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.6104,
      "step": 155
    },
    {
      "epoch": 0.070159658196537,
      "grad_norm": 2.2321743965148926,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.5223,
      "step": 156
    },
    {
      "epoch": 0.07060939959523274,
      "grad_norm": 1.4635720252990723,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.4029,
      "step": 157
    },
    {
      "epoch": 0.07105914099392849,
      "grad_norm": 1.9950910806655884,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.2904,
      "step": 158
    },
    {
      "epoch": 0.07150888239262423,
      "grad_norm": 1.1130157709121704,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.229,
      "step": 159
    },
    {
      "epoch": 0.07195862379132,
      "grad_norm": 1.522284746170044,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.2675,
      "step": 160
    },
    {
      "epoch": 0.07240836519001574,
      "grad_norm": 1.232682228088379,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.3086,
      "step": 161
    },
    {
      "epoch": 0.07285810658871149,
      "grad_norm": 1.0816649198532104,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2263,
      "step": 162
    },
    {
      "epoch": 0.07330784798740723,
      "grad_norm": 2.5467801094055176,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0293,
      "step": 163
    },
    {
      "epoch": 0.073757589386103,
      "grad_norm": 0.9722093343734741,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2555,
      "step": 164
    },
    {
      "epoch": 0.07420733078479874,
      "grad_norm": 0.8932598233222961,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.27,
      "step": 165
    },
    {
      "epoch": 0.07465707218349449,
      "grad_norm": 0.6046376824378967,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.1398,
      "step": 166
    },
    {
      "epoch": 0.07510681358219023,
      "grad_norm": 1.0055515766143799,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3703,
      "step": 167
    },
    {
      "epoch": 0.075556554980886,
      "grad_norm": 1.8218284845352173,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2656,
      "step": 168
    },
    {
      "epoch": 0.07600629637958174,
      "grad_norm": 1.3842567205429077,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.3727,
      "step": 169
    },
    {
      "epoch": 0.07645603777827749,
      "grad_norm": 0.8669049143791199,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.1526,
      "step": 170
    },
    {
      "epoch": 0.07690577917697324,
      "grad_norm": 1.5502482652664185,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.2949,
      "step": 171
    },
    {
      "epoch": 0.077355520575669,
      "grad_norm": 2.590786933898926,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.3063,
      "step": 172
    },
    {
      "epoch": 0.07780526197436474,
      "grad_norm": 1.4980239868164062,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.371,
      "step": 173
    },
    {
      "epoch": 0.07825500337306049,
      "grad_norm": 0.7792003154754639,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.2598,
      "step": 174
    },
    {
      "epoch": 0.07870474477175624,
      "grad_norm": 1.3562554121017456,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.2571,
      "step": 175
    },
    {
      "epoch": 0.079154486170452,
      "grad_norm": 0.8238479495048523,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.1862,
      "step": 176
    },
    {
      "epoch": 0.07960422756914774,
      "grad_norm": 1.8872851133346558,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.4148,
      "step": 177
    },
    {
      "epoch": 0.08005396896784349,
      "grad_norm": 1.4717414379119873,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.2145,
      "step": 178
    },
    {
      "epoch": 0.08050371036653924,
      "grad_norm": 0.545353353023529,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0709,
      "step": 179
    },
    {
      "epoch": 0.080953451765235,
      "grad_norm": 0.45792168378829956,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.1301,
      "step": 180
    },
    {
      "epoch": 0.08140319316393074,
      "grad_norm": 0.7124459743499756,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.1829,
      "step": 181
    },
    {
      "epoch": 0.08185293456262649,
      "grad_norm": 0.6707390546798706,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.1799,
      "step": 182
    },
    {
      "epoch": 0.08230267596132224,
      "grad_norm": 1.2587655782699585,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.1871,
      "step": 183
    },
    {
      "epoch": 0.082752417360018,
      "grad_norm": 1.6115909814834595,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.2172,
      "step": 184
    },
    {
      "epoch": 0.08320215875871374,
      "grad_norm": 1.0401527881622314,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.1235,
      "step": 185
    },
    {
      "epoch": 0.08365190015740949,
      "grad_norm": 0.7353770136833191,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0885,
      "step": 186
    },
    {
      "epoch": 0.08410164155610524,
      "grad_norm": 1.084867000579834,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.2207,
      "step": 187
    },
    {
      "epoch": 0.08455138295480098,
      "grad_norm": 1.1870498657226562,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2729,
      "step": 188
    },
    {
      "epoch": 0.08500112435349674,
      "grad_norm": 0.44437530636787415,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.1118,
      "step": 189
    },
    {
      "epoch": 0.08545086575219249,
      "grad_norm": 0.4486014246940613,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0976,
      "step": 190
    },
    {
      "epoch": 0.08590060715088824,
      "grad_norm": 0.715753436088562,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.1443,
      "step": 191
    },
    {
      "epoch": 0.08635034854958398,
      "grad_norm": 0.6170939207077026,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.1325,
      "step": 192
    },
    {
      "epoch": 0.08680008994827974,
      "grad_norm": 0.5585030913352966,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.1417,
      "step": 193
    },
    {
      "epoch": 0.08724983134697549,
      "grad_norm": 0.3143051266670227,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.088,
      "step": 194
    },
    {
      "epoch": 0.08769957274567124,
      "grad_norm": 0.7443788647651672,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.1938,
      "step": 195
    },
    {
      "epoch": 0.08814931414436698,
      "grad_norm": 0.5507656931877136,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.1522,
      "step": 196
    },
    {
      "epoch": 0.08859905554306274,
      "grad_norm": 0.6213487982749939,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.2454,
      "step": 197
    },
    {
      "epoch": 0.08904879694175849,
      "grad_norm": 0.6985729336738586,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.1538,
      "step": 198
    },
    {
      "epoch": 0.08949853834045424,
      "grad_norm": 1.1249070167541504,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2568,
      "step": 199
    },
    {
      "epoch": 0.08994827973914998,
      "grad_norm": 1.0557364225387573,
      "learning_rate": 0.0,
      "loss": 0.2892,
      "step": 200
    },
    {
      "epoch": 0.08994827973914998,
      "eval_loss": 0.08608997613191605,
      "eval_runtime": 203.166,
      "eval_samples_per_second": 18.433,
      "eval_steps_per_second": 4.612,
      "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": 1.7420184239210496e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}