{
  "best_metric": 0.08815223723649979,
  "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": 11.735479354858398,
      "learning_rate": 1e-05,
      "loss": 10.6173,
      "step": 1
    },
    {
      "epoch": 0.00044974139869574995,
      "eval_loss": 2.5602872371673584,
      "eval_runtime": 201.6375,
      "eval_samples_per_second": 18.573,
      "eval_steps_per_second": 4.647,
      "step": 1
    },
    {
      "epoch": 0.0008994827973914999,
      "grad_norm": 17.807926177978516,
      "learning_rate": 2e-05,
      "loss": 14.4428,
      "step": 2
    },
    {
      "epoch": 0.0013492241960872497,
      "grad_norm": 17.837289810180664,
      "learning_rate": 3e-05,
      "loss": 14.5793,
      "step": 3
    },
    {
      "epoch": 0.0017989655947829998,
      "grad_norm": 16.72642707824707,
      "learning_rate": 4e-05,
      "loss": 15.1181,
      "step": 4
    },
    {
      "epoch": 0.00224870699347875,
      "grad_norm": 19.632795333862305,
      "learning_rate": 5e-05,
      "loss": 15.1309,
      "step": 5
    },
    {
      "epoch": 0.0026984483921744995,
      "grad_norm": 19.37847137451172,
      "learning_rate": 6e-05,
      "loss": 15.2668,
      "step": 6
    },
    {
      "epoch": 0.0031481897908702495,
      "grad_norm": 26.49122428894043,
      "learning_rate": 7e-05,
      "loss": 15.7215,
      "step": 7
    },
    {
      "epoch": 0.0035979311895659996,
      "grad_norm": 14.679391860961914,
      "learning_rate": 8e-05,
      "loss": 9.3807,
      "step": 8
    },
    {
      "epoch": 0.00404767258826175,
      "grad_norm": 12.275378227233887,
      "learning_rate": 9e-05,
      "loss": 3.5329,
      "step": 9
    },
    {
      "epoch": 0.0044974139869575,
      "grad_norm": 11.591130256652832,
      "learning_rate": 0.0001,
      "loss": 2.7893,
      "step": 10
    },
    {
      "epoch": 0.00494715538565325,
      "grad_norm": 6.519595623016357,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.8468,
      "step": 11
    },
    {
      "epoch": 0.005396896784348999,
      "grad_norm": 5.2675700187683105,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.8381,
      "step": 12
    },
    {
      "epoch": 0.005846638183044749,
      "grad_norm": 4.8740949630737305,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.2442,
      "step": 13
    },
    {
      "epoch": 0.006296379581740499,
      "grad_norm": 3.3301310539245605,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.9507,
      "step": 14
    },
    {
      "epoch": 0.006746120980436249,
      "grad_norm": 2.8382670879364014,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.847,
      "step": 15
    },
    {
      "epoch": 0.007195862379131999,
      "grad_norm": 3.4895591735839844,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.6031,
      "step": 16
    },
    {
      "epoch": 0.007645603777827749,
      "grad_norm": 2.971111536026001,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.6742,
      "step": 17
    },
    {
      "epoch": 0.0080953451765235,
      "grad_norm": 2.177934408187866,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.5172,
      "step": 18
    },
    {
      "epoch": 0.00854508657521925,
      "grad_norm": 2.864173412322998,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.7161,
      "step": 19
    },
    {
      "epoch": 0.008994827973915,
      "grad_norm": 2.579519748687744,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.5086,
      "step": 20
    },
    {
      "epoch": 0.00944456937261075,
      "grad_norm": 4.2306132316589355,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.6602,
      "step": 21
    },
    {
      "epoch": 0.0098943107713065,
      "grad_norm": 2.6955723762512207,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.7386,
      "step": 22
    },
    {
      "epoch": 0.01034405217000225,
      "grad_norm": 2.6514077186584473,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.6599,
      "step": 23
    },
    {
      "epoch": 0.010793793568697998,
      "grad_norm": 1.8549970388412476,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.5226,
      "step": 24
    },
    {
      "epoch": 0.011243534967393748,
      "grad_norm": 1.9528262615203857,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.6709,
      "step": 25
    },
    {
      "epoch": 0.011693276366089498,
      "grad_norm": 2.2940781116485596,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.616,
      "step": 26
    },
    {
      "epoch": 0.012143017764785248,
      "grad_norm": 6.190703392028809,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.739,
      "step": 27
    },
    {
      "epoch": 0.012592759163480998,
      "grad_norm": 3.067837715148926,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.499,
      "step": 28
    },
    {
      "epoch": 0.013042500562176748,
      "grad_norm": 1.2731962203979492,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.3533,
      "step": 29
    },
    {
      "epoch": 0.013492241960872498,
      "grad_norm": 2.0637011528015137,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.8369,
      "step": 30
    },
    {
      "epoch": 0.013941983359568248,
      "grad_norm": 2.2823944091796875,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.8267,
      "step": 31
    },
    {
      "epoch": 0.014391724758263998,
      "grad_norm": 5.36124324798584,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.1588,
      "step": 32
    },
    {
      "epoch": 0.014841466156959748,
      "grad_norm": 4.739502429962158,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.049,
      "step": 33
    },
    {
      "epoch": 0.015291207555655498,
      "grad_norm": 2.8808159828186035,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.4704,
      "step": 34
    },
    {
      "epoch": 0.01574094895435125,
      "grad_norm": 2.1416738033294678,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.4021,
      "step": 35
    },
    {
      "epoch": 0.016190690353047,
      "grad_norm": 1.1888728141784668,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.2505,
      "step": 36
    },
    {
      "epoch": 0.01664043175174275,
      "grad_norm": 1.31410551071167,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.2359,
      "step": 37
    },
    {
      "epoch": 0.0170901731504385,
      "grad_norm": 0.9498269557952881,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.0917,
      "step": 38
    },
    {
      "epoch": 0.01753991454913425,
      "grad_norm": 2.7599101066589355,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.3757,
      "step": 39
    },
    {
      "epoch": 0.01798965594783,
      "grad_norm": 1.1273548603057861,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.2274,
      "step": 40
    },
    {
      "epoch": 0.01843939734652575,
      "grad_norm": 3.2617902755737305,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.219,
      "step": 41
    },
    {
      "epoch": 0.0188891387452215,
      "grad_norm": 0.9467918276786804,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.1196,
      "step": 42
    },
    {
      "epoch": 0.01933888014391725,
      "grad_norm": 0.9428938627243042,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.1577,
      "step": 43
    },
    {
      "epoch": 0.019788621542613,
      "grad_norm": 0.9939823150634766,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.1289,
      "step": 44
    },
    {
      "epoch": 0.02023836294130875,
      "grad_norm": 1.3999722003936768,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.2146,
      "step": 45
    },
    {
      "epoch": 0.0206881043400045,
      "grad_norm": 1.936253309249878,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.2816,
      "step": 46
    },
    {
      "epoch": 0.021137845738700246,
      "grad_norm": 1.0651367902755737,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.0799,
      "step": 47
    },
    {
      "epoch": 0.021587587137395996,
      "grad_norm": 1.3475674390792847,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.1354,
      "step": 48
    },
    {
      "epoch": 0.022037328536091746,
      "grad_norm": 1.221743106842041,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.13,
      "step": 49
    },
    {
      "epoch": 0.022487069934787496,
      "grad_norm": 3.161536931991577,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4896,
      "step": 50
    },
    {
      "epoch": 0.022487069934787496,
      "eval_loss": 0.38648658990859985,
      "eval_runtime": 203.0827,
      "eval_samples_per_second": 18.441,
      "eval_steps_per_second": 4.614,
      "step": 50
    },
    {
      "epoch": 0.022936811333483246,
      "grad_norm": 8.87120246887207,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.2464,
      "step": 51
    },
    {
      "epoch": 0.023386552732178996,
      "grad_norm": 10.309056282043457,
      "learning_rate": 8.842005554284296e-05,
      "loss": 4.8396,
      "step": 52
    },
    {
      "epoch": 0.023836294130874746,
      "grad_norm": 11.742121696472168,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.8737,
      "step": 53
    },
    {
      "epoch": 0.024286035529570496,
      "grad_norm": 11.009238243103027,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.4629,
      "step": 54
    },
    {
      "epoch": 0.024735776928266246,
      "grad_norm": 12.108832359313965,
      "learning_rate": 8.678619553365659e-05,
      "loss": 4.9016,
      "step": 55
    },
    {
      "epoch": 0.025185518326961996,
      "grad_norm": 10.640168190002441,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.1305,
      "step": 56
    },
    {
      "epoch": 0.025635259725657746,
      "grad_norm": 10.313965797424316,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.9402,
      "step": 57
    },
    {
      "epoch": 0.026085001124353496,
      "grad_norm": 10.491543769836426,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.4536,
      "step": 58
    },
    {
      "epoch": 0.026534742523049246,
      "grad_norm": 5.127934455871582,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.592,
      "step": 59
    },
    {
      "epoch": 0.026984483921744996,
      "grad_norm": 7.034383296966553,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3663,
      "step": 60
    },
    {
      "epoch": 0.027434225320440746,
      "grad_norm": 8.371705055236816,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.7329,
      "step": 61
    },
    {
      "epoch": 0.027883966719136496,
      "grad_norm": 3.9255106449127197,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.7193,
      "step": 62
    },
    {
      "epoch": 0.028333708117832247,
      "grad_norm": 2.566946268081665,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.6186,
      "step": 63
    },
    {
      "epoch": 0.028783449516527997,
      "grad_norm": 1.653268575668335,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.623,
      "step": 64
    },
    {
      "epoch": 0.029233190915223747,
      "grad_norm": 1.484372615814209,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.3574,
      "step": 65
    },
    {
      "epoch": 0.029682932313919497,
      "grad_norm": 1.5897136926651,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.4067,
      "step": 66
    },
    {
      "epoch": 0.030132673712615247,
      "grad_norm": 1.4185423851013184,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.4376,
      "step": 67
    },
    {
      "epoch": 0.030582415111310997,
      "grad_norm": 1.5517725944519043,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3866,
      "step": 68
    },
    {
      "epoch": 0.031032156510006747,
      "grad_norm": 1.1931599378585815,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.2473,
      "step": 69
    },
    {
      "epoch": 0.0314818979087025,
      "grad_norm": 1.46664559841156,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.3175,
      "step": 70
    },
    {
      "epoch": 0.031931639307398244,
      "grad_norm": 1.9563831090927124,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.5112,
      "step": 71
    },
    {
      "epoch": 0.032381380706094,
      "grad_norm": 7.301497459411621,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.2561,
      "step": 72
    },
    {
      "epoch": 0.032831122104789744,
      "grad_norm": 1.067612886428833,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.3224,
      "step": 73
    },
    {
      "epoch": 0.0332808635034855,
      "grad_norm": 1.5555410385131836,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.3207,
      "step": 74
    },
    {
      "epoch": 0.033730604902181244,
      "grad_norm": 2.963475465774536,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.6059,
      "step": 75
    },
    {
      "epoch": 0.034180346300877,
      "grad_norm": 1.4874687194824219,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.2906,
      "step": 76
    },
    {
      "epoch": 0.034630087699572744,
      "grad_norm": 1.4899744987487793,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.2471,
      "step": 77
    },
    {
      "epoch": 0.0350798290982685,
      "grad_norm": 1.3694696426391602,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4091,
      "step": 78
    },
    {
      "epoch": 0.035529570496964244,
      "grad_norm": 1.13101327419281,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.2343,
      "step": 79
    },
    {
      "epoch": 0.03597931189566,
      "grad_norm": 1.5319714546203613,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.3799,
      "step": 80
    },
    {
      "epoch": 0.036429053294355744,
      "grad_norm": 2.903934955596924,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.9646,
      "step": 81
    },
    {
      "epoch": 0.0368787946930515,
      "grad_norm": 2.913566827774048,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.6251,
      "step": 82
    },
    {
      "epoch": 0.037328536091747244,
      "grad_norm": 2.2844438552856445,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.6425,
      "step": 83
    },
    {
      "epoch": 0.037778277490443,
      "grad_norm": 1.5728940963745117,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.4093,
      "step": 84
    },
    {
      "epoch": 0.038228018889138744,
      "grad_norm": 2.174161911010742,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.2731,
      "step": 85
    },
    {
      "epoch": 0.0386777602878345,
      "grad_norm": 1.046677827835083,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.2023,
      "step": 86
    },
    {
      "epoch": 0.039127501686530244,
      "grad_norm": 0.5291465520858765,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.1288,
      "step": 87
    },
    {
      "epoch": 0.039577243085226,
      "grad_norm": 0.8379286527633667,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3177,
      "step": 88
    },
    {
      "epoch": 0.040026984483921745,
      "grad_norm": 1.1625365018844604,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.2223,
      "step": 89
    },
    {
      "epoch": 0.0404767258826175,
      "grad_norm": 4.300884246826172,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.8295,
      "step": 90
    },
    {
      "epoch": 0.040926467281313245,
      "grad_norm": 3.1497511863708496,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.5312,
      "step": 91
    },
    {
      "epoch": 0.041376208680009,
      "grad_norm": 0.6804894208908081,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.1431,
      "step": 92
    },
    {
      "epoch": 0.041825950078704745,
      "grad_norm": 0.6079594492912292,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.1529,
      "step": 93
    },
    {
      "epoch": 0.04227569147740049,
      "grad_norm": 0.5444039702415466,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.1474,
      "step": 94
    },
    {
      "epoch": 0.042725432876096245,
      "grad_norm": 0.46234166622161865,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.0676,
      "step": 95
    },
    {
      "epoch": 0.04317517427479199,
      "grad_norm": 0.3226686120033264,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.11,
      "step": 96
    },
    {
      "epoch": 0.043624915673487745,
      "grad_norm": 0.41360148787498474,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.1298,
      "step": 97
    },
    {
      "epoch": 0.04407465707218349,
      "grad_norm": 1.2676596641540527,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3512,
      "step": 98
    },
    {
      "epoch": 0.044524398470879245,
      "grad_norm": 0.8815937638282776,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.3051,
      "step": 99
    },
    {
      "epoch": 0.04497413986957499,
      "grad_norm": 0.6634875535964966,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3279,
      "step": 100
    },
    {
      "epoch": 0.04497413986957499,
      "eval_loss": 0.15176475048065186,
      "eval_runtime": 203.3086,
      "eval_samples_per_second": 18.42,
      "eval_steps_per_second": 4.609,
      "step": 100
    },
    {
      "epoch": 0.045423881268270745,
      "grad_norm": 4.503838539123535,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.8085,
      "step": 101
    },
    {
      "epoch": 0.04587362266696649,
      "grad_norm": 5.392167568206787,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.1901,
      "step": 102
    },
    {
      "epoch": 0.046323364065662245,
      "grad_norm": 4.531358242034912,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.4164,
      "step": 103
    },
    {
      "epoch": 0.04677310546435799,
      "grad_norm": 5.333924770355225,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.1685,
      "step": 104
    },
    {
      "epoch": 0.047222846863053745,
      "grad_norm": 3.8220691680908203,
      "learning_rate": 5e-05,
      "loss": 1.5173,
      "step": 105
    },
    {
      "epoch": 0.04767258826174949,
      "grad_norm": 4.193398952484131,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2325,
      "step": 106
    },
    {
      "epoch": 0.048122329660445246,
      "grad_norm": 5.377170562744141,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.8986,
      "step": 107
    },
    {
      "epoch": 0.04857207105914099,
      "grad_norm": 1.7722346782684326,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.4836,
      "step": 108
    },
    {
      "epoch": 0.049021812457836746,
      "grad_norm": 2.2386422157287598,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4334,
      "step": 109
    },
    {
      "epoch": 0.04947155385653249,
      "grad_norm": 1.7240759134292603,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.5815,
      "step": 110
    },
    {
      "epoch": 0.049921295255228246,
      "grad_norm": 1.6704418659210205,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.7404,
      "step": 111
    },
    {
      "epoch": 0.05037103665392399,
      "grad_norm": 1.263595700263977,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.5995,
      "step": 112
    },
    {
      "epoch": 0.050820778052619746,
      "grad_norm": 1.5287951231002808,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.6653,
      "step": 113
    },
    {
      "epoch": 0.05127051945131549,
      "grad_norm": 1.0167535543441772,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3116,
      "step": 114
    },
    {
      "epoch": 0.051720260850011246,
      "grad_norm": 0.8350121378898621,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2246,
      "step": 115
    },
    {
      "epoch": 0.05217000224870699,
      "grad_norm": 1.3227272033691406,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3421,
      "step": 116
    },
    {
      "epoch": 0.052619743647402746,
      "grad_norm": 1.902307391166687,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3858,
      "step": 117
    },
    {
      "epoch": 0.05306948504609849,
      "grad_norm": 1.223228096961975,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.3105,
      "step": 118
    },
    {
      "epoch": 0.053519226444794246,
      "grad_norm": 1.2240155935287476,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.2104,
      "step": 119
    },
    {
      "epoch": 0.05396896784348999,
      "grad_norm": 1.3407142162322998,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.1675,
      "step": 120
    },
    {
      "epoch": 0.054418709242185746,
      "grad_norm": 1.5004065036773682,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.2572,
      "step": 121
    },
    {
      "epoch": 0.05486845064088149,
      "grad_norm": 1.446515440940857,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3286,
      "step": 122
    },
    {
      "epoch": 0.055318192039577246,
      "grad_norm": 1.0073150396347046,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.2843,
      "step": 123
    },
    {
      "epoch": 0.05576793343827299,
      "grad_norm": 1.3104205131530762,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.3014,
      "step": 124
    },
    {
      "epoch": 0.05621767483696874,
      "grad_norm": 1.247299075126648,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.2103,
      "step": 125
    },
    {
      "epoch": 0.05666741623566449,
      "grad_norm": 1.8214682340621948,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.2691,
      "step": 126
    },
    {
      "epoch": 0.05711715763436024,
      "grad_norm": 2.4894745349884033,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.2715,
      "step": 127
    },
    {
      "epoch": 0.05756689903305599,
      "grad_norm": 1.0628043413162231,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.1814,
      "step": 128
    },
    {
      "epoch": 0.05801664043175174,
      "grad_norm": 1.6424274444580078,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.2328,
      "step": 129
    },
    {
      "epoch": 0.05846638183044749,
      "grad_norm": 0.9742172956466675,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3133,
      "step": 130
    },
    {
      "epoch": 0.05891612322914324,
      "grad_norm": 1.1638513803482056,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.2381,
      "step": 131
    },
    {
      "epoch": 0.05936586462783899,
      "grad_norm": 1.194196343421936,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.3567,
      "step": 132
    },
    {
      "epoch": 0.05981560602653474,
      "grad_norm": 1.968641757965088,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.4313,
      "step": 133
    },
    {
      "epoch": 0.060265347425230494,
      "grad_norm": 2.274766445159912,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.3296,
      "step": 134
    },
    {
      "epoch": 0.06071508882392624,
      "grad_norm": 1.1454808712005615,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.203,
      "step": 135
    },
    {
      "epoch": 0.061164830222621994,
      "grad_norm": 0.920929491519928,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.2365,
      "step": 136
    },
    {
      "epoch": 0.06161457162131774,
      "grad_norm": 0.7115468382835388,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.0966,
      "step": 137
    },
    {
      "epoch": 0.062064313020013494,
      "grad_norm": 0.4595581591129303,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.1102,
      "step": 138
    },
    {
      "epoch": 0.06251405441870925,
      "grad_norm": 0.9952919483184814,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0689,
      "step": 139
    },
    {
      "epoch": 0.062963795817405,
      "grad_norm": 0.9480682015419006,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.2832,
      "step": 140
    },
    {
      "epoch": 0.06341353721610074,
      "grad_norm": 1.8338645696640015,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.1448,
      "step": 141
    },
    {
      "epoch": 0.06386327861479649,
      "grad_norm": 0.7585236430168152,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.1645,
      "step": 142
    },
    {
      "epoch": 0.06431302001349225,
      "grad_norm": 0.6739196181297302,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.2236,
      "step": 143
    },
    {
      "epoch": 0.064762761412188,
      "grad_norm": 0.7191562056541443,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.1577,
      "step": 144
    },
    {
      "epoch": 0.06521250281088374,
      "grad_norm": 1.1242461204528809,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.2121,
      "step": 145
    },
    {
      "epoch": 0.06566224420957949,
      "grad_norm": 0.4112837016582489,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.1017,
      "step": 146
    },
    {
      "epoch": 0.06611198560827525,
      "grad_norm": 0.629940927028656,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.1428,
      "step": 147
    },
    {
      "epoch": 0.066561727006971,
      "grad_norm": 1.1342787742614746,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2692,
      "step": 148
    },
    {
      "epoch": 0.06701146840566674,
      "grad_norm": 0.9170231223106384,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.2879,
      "step": 149
    },
    {
      "epoch": 0.06746120980436249,
      "grad_norm": 0.6451442837715149,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.3306,
      "step": 150
    },
    {
      "epoch": 0.06746120980436249,
      "eval_loss": 0.10282353311777115,
      "eval_runtime": 203.4856,
      "eval_samples_per_second": 18.404,
      "eval_steps_per_second": 4.605,
      "step": 150
    },
    {
      "epoch": 0.06791095120305825,
      "grad_norm": 5.447854042053223,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.5053,
      "step": 151
    },
    {
      "epoch": 0.068360692601754,
      "grad_norm": 4.3016276359558105,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.455,
      "step": 152
    },
    {
      "epoch": 0.06881043400044974,
      "grad_norm": 4.338517665863037,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.8149,
      "step": 153
    },
    {
      "epoch": 0.06926017539914549,
      "grad_norm": 3.9996535778045654,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.6023,
      "step": 154
    },
    {
      "epoch": 0.06970991679784125,
      "grad_norm": 3.978335380554199,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.6419,
      "step": 155
    },
    {
      "epoch": 0.070159658196537,
      "grad_norm": 2.7861545085906982,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.4984,
      "step": 156
    },
    {
      "epoch": 0.07060939959523274,
      "grad_norm": 1.541221261024475,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.4129,
      "step": 157
    },
    {
      "epoch": 0.07105914099392849,
      "grad_norm": 1.5657017230987549,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.2656,
      "step": 158
    },
    {
      "epoch": 0.07150888239262423,
      "grad_norm": 1.461788296699524,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.2147,
      "step": 159
    },
    {
      "epoch": 0.07195862379132,
      "grad_norm": 2.2769362926483154,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.3122,
      "step": 160
    },
    {
      "epoch": 0.07240836519001574,
      "grad_norm": 1.3771231174468994,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.3311,
      "step": 161
    },
    {
      "epoch": 0.07285810658871149,
      "grad_norm": 1.3016321659088135,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2202,
      "step": 162
    },
    {
      "epoch": 0.07330784798740723,
      "grad_norm": 2.679879665374756,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0122,
      "step": 163
    },
    {
      "epoch": 0.073757589386103,
      "grad_norm": 0.7211549282073975,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2187,
      "step": 164
    },
    {
      "epoch": 0.07420733078479874,
      "grad_norm": 0.8960670232772827,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.2479,
      "step": 165
    },
    {
      "epoch": 0.07465707218349449,
      "grad_norm": 0.5109663605690002,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.1192,
      "step": 166
    },
    {
      "epoch": 0.07510681358219023,
      "grad_norm": 1.322703242301941,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.4005,
      "step": 167
    },
    {
      "epoch": 0.075556554980886,
      "grad_norm": 1.3286811113357544,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2547,
      "step": 168
    },
    {
      "epoch": 0.07600629637958174,
      "grad_norm": 1.4223722219467163,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.3935,
      "step": 169
    },
    {
      "epoch": 0.07645603777827749,
      "grad_norm": 0.7864736914634705,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.1529,
      "step": 170
    },
    {
      "epoch": 0.07690577917697324,
      "grad_norm": 2.2625200748443604,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3137,
      "step": 171
    },
    {
      "epoch": 0.077355520575669,
      "grad_norm": 3.223945140838623,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.4011,
      "step": 172
    },
    {
      "epoch": 0.07780526197436474,
      "grad_norm": 1.4796351194381714,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.3865,
      "step": 173
    },
    {
      "epoch": 0.07825500337306049,
      "grad_norm": 0.8055946826934814,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.2541,
      "step": 174
    },
    {
      "epoch": 0.07870474477175624,
      "grad_norm": 1.4209413528442383,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.2835,
      "step": 175
    },
    {
      "epoch": 0.079154486170452,
      "grad_norm": 0.9519951939582825,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.1982,
      "step": 176
    },
    {
      "epoch": 0.07960422756914774,
      "grad_norm": 1.743548035621643,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.4277,
      "step": 177
    },
    {
      "epoch": 0.08005396896784349,
      "grad_norm": 1.4887621402740479,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.23,
      "step": 178
    },
    {
      "epoch": 0.08050371036653924,
      "grad_norm": 0.45318371057510376,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0581,
      "step": 179
    },
    {
      "epoch": 0.080953451765235,
      "grad_norm": 0.478161096572876,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.1342,
      "step": 180
    },
    {
      "epoch": 0.08140319316393074,
      "grad_norm": 0.6548275351524353,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.1667,
      "step": 181
    },
    {
      "epoch": 0.08185293456262649,
      "grad_norm": 0.7000573873519897,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.2032,
      "step": 182
    },
    {
      "epoch": 0.08230267596132224,
      "grad_norm": 1.8243629932403564,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.2544,
      "step": 183
    },
    {
      "epoch": 0.082752417360018,
      "grad_norm": 1.8137884140014648,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.2308,
      "step": 184
    },
    {
      "epoch": 0.08320215875871374,
      "grad_norm": 1.4033957719802856,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.1467,
      "step": 185
    },
    {
      "epoch": 0.08365190015740949,
      "grad_norm": 0.8043022155761719,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0863,
      "step": 186
    },
    {
      "epoch": 0.08410164155610524,
      "grad_norm": 1.1089059114456177,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.23,
      "step": 187
    },
    {
      "epoch": 0.08455138295480098,
      "grad_norm": 1.123526930809021,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2615,
      "step": 188
    },
    {
      "epoch": 0.08500112435349674,
      "grad_norm": 0.5019885301589966,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.1144,
      "step": 189
    },
    {
      "epoch": 0.08545086575219249,
      "grad_norm": 0.5750763416290283,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.1098,
      "step": 190
    },
    {
      "epoch": 0.08590060715088824,
      "grad_norm": 1.3271559476852417,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.1464,
      "step": 191
    },
    {
      "epoch": 0.08635034854958398,
      "grad_norm": 0.5027560591697693,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.1223,
      "step": 192
    },
    {
      "epoch": 0.08680008994827974,
      "grad_norm": 0.5951797962188721,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.1446,
      "step": 193
    },
    {
      "epoch": 0.08724983134697549,
      "grad_norm": 0.3374186158180237,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0873,
      "step": 194
    },
    {
      "epoch": 0.08769957274567124,
      "grad_norm": 0.7564144730567932,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.1846,
      "step": 195
    },
    {
      "epoch": 0.08814931414436698,
      "grad_norm": 0.5421136021614075,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.1527,
      "step": 196
    },
    {
      "epoch": 0.08859905554306274,
      "grad_norm": 0.6341227293014526,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.2451,
      "step": 197
    },
    {
      "epoch": 0.08904879694175849,
      "grad_norm": 0.8133850693702698,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.1658,
      "step": 198
    },
    {
      "epoch": 0.08949853834045424,
      "grad_norm": 0.9790681004524231,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2589,
      "step": 199
    },
    {
      "epoch": 0.08994827973914998,
      "grad_norm": 1.0550408363342285,
      "learning_rate": 0.0,
      "loss": 0.2961,
      "step": 200
    },
    {
      "epoch": 0.08994827973914998,
      "eval_loss": 0.08815223723649979,
      "eval_runtime": 203.3042,
      "eval_samples_per_second": 18.421,
      "eval_steps_per_second": 4.609,
      "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
}