{
  "best_metric": 0.4373587667942047,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 1.834862385321101,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.009174311926605505,
      "grad_norm": 0.22757866978645325,
      "learning_rate": 1e-05,
      "loss": 0.7218,
      "step": 1
    },
    {
      "epoch": 0.009174311926605505,
      "eval_loss": 0.9211078882217407,
      "eval_runtime": 5.5968,
      "eval_samples_per_second": 32.876,
      "eval_steps_per_second": 8.219,
      "step": 1
    },
    {
      "epoch": 0.01834862385321101,
      "grad_norm": 0.2518715262413025,
      "learning_rate": 2e-05,
      "loss": 0.3937,
      "step": 2
    },
    {
      "epoch": 0.027522935779816515,
      "grad_norm": 0.25987499952316284,
      "learning_rate": 3e-05,
      "loss": 0.4462,
      "step": 3
    },
    {
      "epoch": 0.03669724770642202,
      "grad_norm": 0.2748376429080963,
      "learning_rate": 4e-05,
      "loss": 0.6877,
      "step": 4
    },
    {
      "epoch": 0.045871559633027525,
      "grad_norm": 0.26689621806144714,
      "learning_rate": 5e-05,
      "loss": 0.6018,
      "step": 5
    },
    {
      "epoch": 0.05504587155963303,
      "grad_norm": 0.3114973306655884,
      "learning_rate": 6e-05,
      "loss": 0.4678,
      "step": 6
    },
    {
      "epoch": 0.06422018348623854,
      "grad_norm": 0.3408745229244232,
      "learning_rate": 7e-05,
      "loss": 0.5996,
      "step": 7
    },
    {
      "epoch": 0.07339449541284404,
      "grad_norm": 0.28301388025283813,
      "learning_rate": 8e-05,
      "loss": 0.6158,
      "step": 8
    },
    {
      "epoch": 0.08256880733944955,
      "grad_norm": 0.24982784688472748,
      "learning_rate": 9e-05,
      "loss": 0.5184,
      "step": 9
    },
    {
      "epoch": 0.09174311926605505,
      "grad_norm": 0.27339503169059753,
      "learning_rate": 0.0001,
      "loss": 0.707,
      "step": 10
    },
    {
      "epoch": 0.10091743119266056,
      "grad_norm": 0.23815961182117462,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.4169,
      "step": 11
    },
    {
      "epoch": 0.11009174311926606,
      "grad_norm": 0.2776719331741333,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.5363,
      "step": 12
    },
    {
      "epoch": 0.11926605504587157,
      "grad_norm": 0.34082862734794617,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.6406,
      "step": 13
    },
    {
      "epoch": 0.12844036697247707,
      "grad_norm": 0.3164188265800476,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.5925,
      "step": 14
    },
    {
      "epoch": 0.13761467889908258,
      "grad_norm": 0.38142895698547363,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.6991,
      "step": 15
    },
    {
      "epoch": 0.14678899082568808,
      "grad_norm": 0.35001686215400696,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.6228,
      "step": 16
    },
    {
      "epoch": 0.1559633027522936,
      "grad_norm": 0.3453150689601898,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.6417,
      "step": 17
    },
    {
      "epoch": 0.1651376146788991,
      "grad_norm": 0.3127257823944092,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.7347,
      "step": 18
    },
    {
      "epoch": 0.1743119266055046,
      "grad_norm": 0.3859875500202179,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.7505,
      "step": 19
    },
    {
      "epoch": 0.1834862385321101,
      "grad_norm": 0.36082231998443604,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.7524,
      "step": 20
    },
    {
      "epoch": 0.1926605504587156,
      "grad_norm": 0.3412916958332062,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.6174,
      "step": 21
    },
    {
      "epoch": 0.2018348623853211,
      "grad_norm": 0.38019198179244995,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.7653,
      "step": 22
    },
    {
      "epoch": 0.21100917431192662,
      "grad_norm": 0.3949272632598877,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.7468,
      "step": 23
    },
    {
      "epoch": 0.22018348623853212,
      "grad_norm": 0.4098750054836273,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.5949,
      "step": 24
    },
    {
      "epoch": 0.22935779816513763,
      "grad_norm": 0.48608002066612244,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.7669,
      "step": 25
    },
    {
      "epoch": 0.23853211009174313,
      "grad_norm": 0.5924227833747864,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.8667,
      "step": 26
    },
    {
      "epoch": 0.24770642201834864,
      "grad_norm": 0.952046275138855,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.3204,
      "step": 27
    },
    {
      "epoch": 0.25688073394495414,
      "grad_norm": 0.28786131739616394,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.3764,
      "step": 28
    },
    {
      "epoch": 0.26605504587155965,
      "grad_norm": 0.2809140384197235,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.26,
      "step": 29
    },
    {
      "epoch": 0.27522935779816515,
      "grad_norm": 0.2796563506126404,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.2351,
      "step": 30
    },
    {
      "epoch": 0.28440366972477066,
      "grad_norm": 0.30367258191108704,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.4879,
      "step": 31
    },
    {
      "epoch": 0.29357798165137616,
      "grad_norm": 0.27832701802253723,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.4503,
      "step": 32
    },
    {
      "epoch": 0.30275229357798167,
      "grad_norm": 0.22784149646759033,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.4026,
      "step": 33
    },
    {
      "epoch": 0.3119266055045872,
      "grad_norm": 0.23567678034305573,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.4635,
      "step": 34
    },
    {
      "epoch": 0.3211009174311927,
      "grad_norm": 0.21042758226394653,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.4569,
      "step": 35
    },
    {
      "epoch": 0.3302752293577982,
      "grad_norm": 0.22146302461624146,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.485,
      "step": 36
    },
    {
      "epoch": 0.3394495412844037,
      "grad_norm": 0.24271990358829498,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.5733,
      "step": 37
    },
    {
      "epoch": 0.3486238532110092,
      "grad_norm": 0.21694903075695038,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.4542,
      "step": 38
    },
    {
      "epoch": 0.3577981651376147,
      "grad_norm": 0.2296721190214157,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.3965,
      "step": 39
    },
    {
      "epoch": 0.3669724770642202,
      "grad_norm": 0.30406907200813293,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.6953,
      "step": 40
    },
    {
      "epoch": 0.3761467889908257,
      "grad_norm": 0.2594863176345825,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.5206,
      "step": 41
    },
    {
      "epoch": 0.3853211009174312,
      "grad_norm": 0.23742105066776276,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.4841,
      "step": 42
    },
    {
      "epoch": 0.3944954128440367,
      "grad_norm": 0.2492372840642929,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.5908,
      "step": 43
    },
    {
      "epoch": 0.4036697247706422,
      "grad_norm": 0.25272437930107117,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.5104,
      "step": 44
    },
    {
      "epoch": 0.41284403669724773,
      "grad_norm": 0.2697179615497589,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.5563,
      "step": 45
    },
    {
      "epoch": 0.42201834862385323,
      "grad_norm": 0.2949214577674866,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.6487,
      "step": 46
    },
    {
      "epoch": 0.43119266055045874,
      "grad_norm": 0.3049638569355011,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.6187,
      "step": 47
    },
    {
      "epoch": 0.44036697247706424,
      "grad_norm": 0.28287872672080994,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.5873,
      "step": 48
    },
    {
      "epoch": 0.44954128440366975,
      "grad_norm": 0.3109414875507355,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.5381,
      "step": 49
    },
    {
      "epoch": 0.45871559633027525,
      "grad_norm": 0.2838362753391266,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.4779,
      "step": 50
    },
    {
      "epoch": 0.45871559633027525,
      "eval_loss": 0.5077647566795349,
      "eval_runtime": 5.656,
      "eval_samples_per_second": 32.532,
      "eval_steps_per_second": 8.133,
      "step": 50
    },
    {
      "epoch": 0.46788990825688076,
      "grad_norm": 0.4119388461112976,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.7617,
      "step": 51
    },
    {
      "epoch": 0.47706422018348627,
      "grad_norm": 0.39679720997810364,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.6846,
      "step": 52
    },
    {
      "epoch": 0.48623853211009177,
      "grad_norm": 0.48571300506591797,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6158,
      "step": 53
    },
    {
      "epoch": 0.4954128440366973,
      "grad_norm": 0.7497296333312988,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.8934,
      "step": 54
    },
    {
      "epoch": 0.5045871559633027,
      "grad_norm": 0.12586839497089386,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.2073,
      "step": 55
    },
    {
      "epoch": 0.5137614678899083,
      "grad_norm": 0.16683803498744965,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.369,
      "step": 56
    },
    {
      "epoch": 0.5229357798165137,
      "grad_norm": 0.17478333413600922,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.257,
      "step": 57
    },
    {
      "epoch": 0.5321100917431193,
      "grad_norm": 0.2005760818719864,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3284,
      "step": 58
    },
    {
      "epoch": 0.5412844036697247,
      "grad_norm": 0.2104957401752472,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.2932,
      "step": 59
    },
    {
      "epoch": 0.5504587155963303,
      "grad_norm": 0.19943290948867798,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.4085,
      "step": 60
    },
    {
      "epoch": 0.5596330275229358,
      "grad_norm": 0.20082347095012665,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.3372,
      "step": 61
    },
    {
      "epoch": 0.5688073394495413,
      "grad_norm": 0.20336656272411346,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.451,
      "step": 62
    },
    {
      "epoch": 0.5779816513761468,
      "grad_norm": 0.18236014246940613,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.2637,
      "step": 63
    },
    {
      "epoch": 0.5871559633027523,
      "grad_norm": 0.18370752036571503,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.3246,
      "step": 64
    },
    {
      "epoch": 0.5963302752293578,
      "grad_norm": 0.21893881261348724,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.4081,
      "step": 65
    },
    {
      "epoch": 0.6055045871559633,
      "grad_norm": 0.218050017952919,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.4276,
      "step": 66
    },
    {
      "epoch": 0.6146788990825688,
      "grad_norm": 0.2780652642250061,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.5171,
      "step": 67
    },
    {
      "epoch": 0.6238532110091743,
      "grad_norm": 0.2711992561817169,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.6147,
      "step": 68
    },
    {
      "epoch": 0.6330275229357798,
      "grad_norm": 0.3022063672542572,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.6623,
      "step": 69
    },
    {
      "epoch": 0.6422018348623854,
      "grad_norm": 0.3054291009902954,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.6667,
      "step": 70
    },
    {
      "epoch": 0.6513761467889908,
      "grad_norm": 0.21913567185401917,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.3601,
      "step": 71
    },
    {
      "epoch": 0.6605504587155964,
      "grad_norm": 0.324846088886261,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.5895,
      "step": 72
    },
    {
      "epoch": 0.6697247706422018,
      "grad_norm": 0.26592978835105896,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.4553,
      "step": 73
    },
    {
      "epoch": 0.6788990825688074,
      "grad_norm": 0.3039897084236145,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.5398,
      "step": 74
    },
    {
      "epoch": 0.6880733944954128,
      "grad_norm": 0.3944902718067169,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5678,
      "step": 75
    },
    {
      "epoch": 0.6972477064220184,
      "grad_norm": 0.289384663105011,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.4565,
      "step": 76
    },
    {
      "epoch": 0.7064220183486238,
      "grad_norm": 0.34469839930534363,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.6127,
      "step": 77
    },
    {
      "epoch": 0.7155963302752294,
      "grad_norm": 0.35868868231773376,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.7055,
      "step": 78
    },
    {
      "epoch": 0.7247706422018348,
      "grad_norm": 0.38293564319610596,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.6233,
      "step": 79
    },
    {
      "epoch": 0.7339449541284404,
      "grad_norm": 0.44443824887275696,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.658,
      "step": 80
    },
    {
      "epoch": 0.7431192660550459,
      "grad_norm": 0.5801151990890503,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.6653,
      "step": 81
    },
    {
      "epoch": 0.7522935779816514,
      "grad_norm": 0.13994742929935455,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.2437,
      "step": 82
    },
    {
      "epoch": 0.7614678899082569,
      "grad_norm": 0.14491501450538635,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.2894,
      "step": 83
    },
    {
      "epoch": 0.7706422018348624,
      "grad_norm": 0.17788182199001312,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.4203,
      "step": 84
    },
    {
      "epoch": 0.7798165137614679,
      "grad_norm": 0.17706838250160217,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3385,
      "step": 85
    },
    {
      "epoch": 0.7889908256880734,
      "grad_norm": 0.2025514394044876,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.405,
      "step": 86
    },
    {
      "epoch": 0.7981651376146789,
      "grad_norm": 0.1984269618988037,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.4032,
      "step": 87
    },
    {
      "epoch": 0.8073394495412844,
      "grad_norm": 0.2340664267539978,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.6006,
      "step": 88
    },
    {
      "epoch": 0.8165137614678899,
      "grad_norm": 0.2235722541809082,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.5356,
      "step": 89
    },
    {
      "epoch": 0.8256880733944955,
      "grad_norm": 0.23755811154842377,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.5478,
      "step": 90
    },
    {
      "epoch": 0.8348623853211009,
      "grad_norm": 0.23930624127388,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.4736,
      "step": 91
    },
    {
      "epoch": 0.8440366972477065,
      "grad_norm": 0.2998720109462738,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.6657,
      "step": 92
    },
    {
      "epoch": 0.8532110091743119,
      "grad_norm": 0.23668353259563446,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.4412,
      "step": 93
    },
    {
      "epoch": 0.8623853211009175,
      "grad_norm": 0.22916096448898315,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.4792,
      "step": 94
    },
    {
      "epoch": 0.8715596330275229,
      "grad_norm": 0.2859799861907959,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.5534,
      "step": 95
    },
    {
      "epoch": 0.8807339449541285,
      "grad_norm": 0.2234007865190506,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.3116,
      "step": 96
    },
    {
      "epoch": 0.8899082568807339,
      "grad_norm": 0.24725241959095,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.4434,
      "step": 97
    },
    {
      "epoch": 0.8990825688073395,
      "grad_norm": 0.33359816670417786,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.6264,
      "step": 98
    },
    {
      "epoch": 0.908256880733945,
      "grad_norm": 0.3186333179473877,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.5801,
      "step": 99
    },
    {
      "epoch": 0.9174311926605505,
      "grad_norm": 0.29427671432495117,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.4406,
      "step": 100
    },
    {
      "epoch": 0.9174311926605505,
      "eval_loss": 0.4615384042263031,
      "eval_runtime": 5.6472,
      "eval_samples_per_second": 32.582,
      "eval_steps_per_second": 8.146,
      "step": 100
    },
    {
      "epoch": 0.926605504587156,
      "grad_norm": 0.30921870470046997,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.5243,
      "step": 101
    },
    {
      "epoch": 0.9357798165137615,
      "grad_norm": 0.2794451117515564,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.4482,
      "step": 102
    },
    {
      "epoch": 0.944954128440367,
      "grad_norm": 0.34087374806404114,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.5345,
      "step": 103
    },
    {
      "epoch": 0.9541284403669725,
      "grad_norm": 0.31846287846565247,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.4961,
      "step": 104
    },
    {
      "epoch": 0.963302752293578,
      "grad_norm": 0.3624880611896515,
      "learning_rate": 5e-05,
      "loss": 0.5455,
      "step": 105
    },
    {
      "epoch": 0.9724770642201835,
      "grad_norm": 0.44983717799186707,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.6828,
      "step": 106
    },
    {
      "epoch": 0.981651376146789,
      "grad_norm": 0.4052683413028717,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.4228,
      "step": 107
    },
    {
      "epoch": 0.9908256880733946,
      "grad_norm": 0.8898103833198547,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.9861,
      "step": 108
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.25735190510749817,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.5298,
      "step": 109
    },
    {
      "epoch": 1.0091743119266054,
      "grad_norm": 0.128637433052063,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.1975,
      "step": 110
    },
    {
      "epoch": 1.018348623853211,
      "grad_norm": 0.14077633619308472,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2318,
      "step": 111
    },
    {
      "epoch": 1.0275229357798166,
      "grad_norm": 0.13651268184185028,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.1996,
      "step": 112
    },
    {
      "epoch": 1.036697247706422,
      "grad_norm": 0.18093420565128326,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3846,
      "step": 113
    },
    {
      "epoch": 1.0458715596330275,
      "grad_norm": 0.2242155224084854,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.4636,
      "step": 114
    },
    {
      "epoch": 1.0550458715596331,
      "grad_norm": 0.18750467896461487,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2834,
      "step": 115
    },
    {
      "epoch": 1.0642201834862386,
      "grad_norm": 0.2178245633840561,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.4311,
      "step": 116
    },
    {
      "epoch": 1.073394495412844,
      "grad_norm": 0.17102479934692383,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.2747,
      "step": 117
    },
    {
      "epoch": 1.0825688073394495,
      "grad_norm": 0.21090644598007202,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.3743,
      "step": 118
    },
    {
      "epoch": 1.091743119266055,
      "grad_norm": 0.24766762554645538,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.4842,
      "step": 119
    },
    {
      "epoch": 1.1009174311926606,
      "grad_norm": 0.2183620184659958,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.2963,
      "step": 120
    },
    {
      "epoch": 1.110091743119266,
      "grad_norm": 0.24873808026313782,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.2973,
      "step": 121
    },
    {
      "epoch": 1.1192660550458715,
      "grad_norm": 0.2532205581665039,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.4489,
      "step": 122
    },
    {
      "epoch": 1.1284403669724772,
      "grad_norm": 0.28706297278404236,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.4766,
      "step": 123
    },
    {
      "epoch": 1.1376146788990826,
      "grad_norm": 0.29905185103416443,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.5887,
      "step": 124
    },
    {
      "epoch": 1.146788990825688,
      "grad_norm": 0.30878540873527527,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.5662,
      "step": 125
    },
    {
      "epoch": 1.1559633027522935,
      "grad_norm": 0.27856945991516113,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.4332,
      "step": 126
    },
    {
      "epoch": 1.165137614678899,
      "grad_norm": 0.23180490732192993,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3203,
      "step": 127
    },
    {
      "epoch": 1.1743119266055047,
      "grad_norm": 0.2905110716819763,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.4472,
      "step": 128
    },
    {
      "epoch": 1.18348623853211,
      "grad_norm": 0.3204229772090912,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.4192,
      "step": 129
    },
    {
      "epoch": 1.1926605504587156,
      "grad_norm": 0.3300638794898987,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.4646,
      "step": 130
    },
    {
      "epoch": 1.2018348623853212,
      "grad_norm": 0.33702805638313293,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.4715,
      "step": 131
    },
    {
      "epoch": 1.2110091743119267,
      "grad_norm": 0.3043479323387146,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.4083,
      "step": 132
    },
    {
      "epoch": 1.2201834862385321,
      "grad_norm": 0.37676098942756653,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.5549,
      "step": 133
    },
    {
      "epoch": 1.2293577981651376,
      "grad_norm": 0.3981700539588928,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.5209,
      "step": 134
    },
    {
      "epoch": 1.238532110091743,
      "grad_norm": 0.42665356397628784,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.4003,
      "step": 135
    },
    {
      "epoch": 1.2477064220183487,
      "grad_norm": 0.5916845798492432,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.5795,
      "step": 136
    },
    {
      "epoch": 1.2568807339449541,
      "grad_norm": 0.1779555082321167,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.4313,
      "step": 137
    },
    {
      "epoch": 1.2660550458715596,
      "grad_norm": 0.18440808355808258,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.3235,
      "step": 138
    },
    {
      "epoch": 1.2752293577981653,
      "grad_norm": 0.2160906344652176,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3691,
      "step": 139
    },
    {
      "epoch": 1.2844036697247707,
      "grad_norm": 0.19731925427913666,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.3363,
      "step": 140
    },
    {
      "epoch": 1.2935779816513762,
      "grad_norm": 0.23964619636535645,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.4056,
      "step": 141
    },
    {
      "epoch": 1.3027522935779816,
      "grad_norm": 0.21658800542354584,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2884,
      "step": 142
    },
    {
      "epoch": 1.311926605504587,
      "grad_norm": 0.216840922832489,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.2469,
      "step": 143
    },
    {
      "epoch": 1.3211009174311927,
      "grad_norm": 0.28055331110954285,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.4354,
      "step": 144
    },
    {
      "epoch": 1.3302752293577982,
      "grad_norm": 0.20430664718151093,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.2323,
      "step": 145
    },
    {
      "epoch": 1.3394495412844036,
      "grad_norm": 0.27874380350112915,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.423,
      "step": 146
    },
    {
      "epoch": 1.3486238532110093,
      "grad_norm": 0.2390122264623642,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.3875,
      "step": 147
    },
    {
      "epoch": 1.3577981651376148,
      "grad_norm": 0.36405956745147705,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.513,
      "step": 148
    },
    {
      "epoch": 1.3669724770642202,
      "grad_norm": 0.2961496412754059,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.4645,
      "step": 149
    },
    {
      "epoch": 1.3761467889908257,
      "grad_norm": 0.29616445302963257,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.4593,
      "step": 150
    },
    {
      "epoch": 1.3761467889908257,
      "eval_loss": 0.4373587667942047,
      "eval_runtime": 5.654,
      "eval_samples_per_second": 32.543,
      "eval_steps_per_second": 8.136,
      "step": 150
    },
    {
      "epoch": 1.385321100917431,
      "grad_norm": 0.2940848767757416,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.4251,
      "step": 151
    },
    {
      "epoch": 1.3944954128440368,
      "grad_norm": 0.2853677570819855,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.4258,
      "step": 152
    },
    {
      "epoch": 1.4036697247706422,
      "grad_norm": 0.37697380781173706,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.3823,
      "step": 153
    },
    {
      "epoch": 1.4128440366972477,
      "grad_norm": 0.3618301451206207,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.5075,
      "step": 154
    },
    {
      "epoch": 1.4220183486238533,
      "grad_norm": 0.29573801159858704,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.3739,
      "step": 155
    },
    {
      "epoch": 1.4311926605504588,
      "grad_norm": 0.31452590227127075,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.3457,
      "step": 156
    },
    {
      "epoch": 1.4403669724770642,
      "grad_norm": 0.3553619086742401,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3682,
      "step": 157
    },
    {
      "epoch": 1.4495412844036697,
      "grad_norm": 0.361484169960022,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.3775,
      "step": 158
    },
    {
      "epoch": 1.4587155963302751,
      "grad_norm": 0.4299015700817108,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.5613,
      "step": 159
    },
    {
      "epoch": 1.4678899082568808,
      "grad_norm": 0.42649099230766296,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.4361,
      "step": 160
    },
    {
      "epoch": 1.4770642201834863,
      "grad_norm": 0.4709986448287964,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.4462,
      "step": 161
    },
    {
      "epoch": 1.4862385321100917,
      "grad_norm": 0.4364341199398041,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.4034,
      "step": 162
    },
    {
      "epoch": 1.4954128440366974,
      "grad_norm": 0.6107885241508484,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.5582,
      "step": 163
    },
    {
      "epoch": 1.5045871559633026,
      "grad_norm": 0.17299024760723114,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.3185,
      "step": 164
    },
    {
      "epoch": 1.5137614678899083,
      "grad_norm": 0.15328259766101837,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.1588,
      "step": 165
    },
    {
      "epoch": 1.5229357798165137,
      "grad_norm": 0.15708227455615997,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.1638,
      "step": 166
    },
    {
      "epoch": 1.5321100917431192,
      "grad_norm": 0.2163887321949005,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.3577,
      "step": 167
    },
    {
      "epoch": 1.5412844036697249,
      "grad_norm": 0.2087908536195755,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2963,
      "step": 168
    },
    {
      "epoch": 1.5504587155963303,
      "grad_norm": 0.21406951546669006,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2773,
      "step": 169
    },
    {
      "epoch": 1.5596330275229358,
      "grad_norm": 0.2188555896282196,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.3109,
      "step": 170
    },
    {
      "epoch": 1.5688073394495414,
      "grad_norm": 0.2507384121417999,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.3785,
      "step": 171
    },
    {
      "epoch": 1.5779816513761467,
      "grad_norm": 0.266482412815094,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.4062,
      "step": 172
    },
    {
      "epoch": 1.5871559633027523,
      "grad_norm": 0.2641501724720001,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.4341,
      "step": 173
    },
    {
      "epoch": 1.5963302752293578,
      "grad_norm": 0.2912276089191437,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.4272,
      "step": 174
    },
    {
      "epoch": 1.6055045871559632,
      "grad_norm": 0.2779712378978729,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3962,
      "step": 175
    },
    {
      "epoch": 1.614678899082569,
      "grad_norm": 0.2843896746635437,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.3575,
      "step": 176
    },
    {
      "epoch": 1.6238532110091743,
      "grad_norm": 0.29254084825515747,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.433,
      "step": 177
    },
    {
      "epoch": 1.6330275229357798,
      "grad_norm": 0.30045565962791443,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.3434,
      "step": 178
    },
    {
      "epoch": 1.6422018348623855,
      "grad_norm": 0.2814767360687256,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3407,
      "step": 179
    },
    {
      "epoch": 1.6513761467889907,
      "grad_norm": 0.3373102843761444,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.4427,
      "step": 180
    },
    {
      "epoch": 1.6605504587155964,
      "grad_norm": 0.33493924140930176,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.3176,
      "step": 181
    },
    {
      "epoch": 1.6697247706422018,
      "grad_norm": 0.35189321637153625,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.4488,
      "step": 182
    },
    {
      "epoch": 1.6788990825688073,
      "grad_norm": 0.42480501532554626,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.605,
      "step": 183
    },
    {
      "epoch": 1.688073394495413,
      "grad_norm": 0.37318694591522217,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.4053,
      "step": 184
    },
    {
      "epoch": 1.6972477064220184,
      "grad_norm": 0.44527217745780945,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.5155,
      "step": 185
    },
    {
      "epoch": 1.7064220183486238,
      "grad_norm": 0.3411549925804138,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.3661,
      "step": 186
    },
    {
      "epoch": 1.7155963302752295,
      "grad_norm": 0.4443856179714203,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.5133,
      "step": 187
    },
    {
      "epoch": 1.7247706422018347,
      "grad_norm": 0.49494215846061707,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.5058,
      "step": 188
    },
    {
      "epoch": 1.7339449541284404,
      "grad_norm": 0.6086474657058716,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.5155,
      "step": 189
    },
    {
      "epoch": 1.7431192660550459,
      "grad_norm": 0.9181317090988159,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.6246,
      "step": 190
    },
    {
      "epoch": 1.7522935779816513,
      "grad_norm": 0.17362545430660248,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.273,
      "step": 191
    },
    {
      "epoch": 1.761467889908257,
      "grad_norm": 0.15214654803276062,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.1905,
      "step": 192
    },
    {
      "epoch": 1.7706422018348624,
      "grad_norm": 0.1746424287557602,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2505,
      "step": 193
    },
    {
      "epoch": 1.7798165137614679,
      "grad_norm": 0.2676239311695099,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.4329,
      "step": 194
    },
    {
      "epoch": 1.7889908256880735,
      "grad_norm": 0.17189502716064453,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.2094,
      "step": 195
    },
    {
      "epoch": 1.7981651376146788,
      "grad_norm": 0.20616373419761658,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.2695,
      "step": 196
    },
    {
      "epoch": 1.8073394495412844,
      "grad_norm": 0.27885767817497253,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.4735,
      "step": 197
    },
    {
      "epoch": 1.81651376146789,
      "grad_norm": 0.26438018679618835,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.398,
      "step": 198
    },
    {
      "epoch": 1.8256880733944953,
      "grad_norm": 0.28252848982810974,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.4276,
      "step": 199
    },
    {
      "epoch": 1.834862385321101,
      "grad_norm": 0.25258275866508484,
      "learning_rate": 0.0,
      "loss": 0.3277,
      "step": 200
    },
    {
      "epoch": 1.834862385321101,
      "eval_loss": 0.43834975361824036,
      "eval_runtime": 5.653,
      "eval_samples_per_second": 32.549,
      "eval_steps_per_second": 8.137,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.61808020979712e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}