{
  "best_metric": 0.9678028225898743,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.9060022650056625,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004530011325028313,
      "grad_norm": 2.2205209732055664,
      "learning_rate": 1e-05,
      "loss": 1.6809,
      "step": 1
    },
    {
      "epoch": 0.004530011325028313,
      "eval_loss": 3.339136838912964,
      "eval_runtime": 27.6203,
      "eval_samples_per_second": 13.468,
      "eval_steps_per_second": 3.367,
      "step": 1
    },
    {
      "epoch": 0.009060022650056626,
      "grad_norm": 2.732006788253784,
      "learning_rate": 2e-05,
      "loss": 1.956,
      "step": 2
    },
    {
      "epoch": 0.013590033975084938,
      "grad_norm": 2.8547580242156982,
      "learning_rate": 3e-05,
      "loss": 2.2194,
      "step": 3
    },
    {
      "epoch": 0.01812004530011325,
      "grad_norm": 6.404782772064209,
      "learning_rate": 4e-05,
      "loss": 2.1446,
      "step": 4
    },
    {
      "epoch": 0.022650056625141562,
      "grad_norm": 7.077938079833984,
      "learning_rate": 5e-05,
      "loss": 2.1511,
      "step": 5
    },
    {
      "epoch": 0.027180067950169876,
      "grad_norm": 4.084441661834717,
      "learning_rate": 6e-05,
      "loss": 1.8931,
      "step": 6
    },
    {
      "epoch": 0.031710079275198186,
      "grad_norm": 3.3573696613311768,
      "learning_rate": 7e-05,
      "loss": 1.6399,
      "step": 7
    },
    {
      "epoch": 0.0362400906002265,
      "grad_norm": 2.4906857013702393,
      "learning_rate": 8e-05,
      "loss": 1.5666,
      "step": 8
    },
    {
      "epoch": 0.04077010192525481,
      "grad_norm": 2.2960405349731445,
      "learning_rate": 9e-05,
      "loss": 1.2641,
      "step": 9
    },
    {
      "epoch": 0.045300113250283124,
      "grad_norm": 2.68849778175354,
      "learning_rate": 0.0001,
      "loss": 1.317,
      "step": 10
    },
    {
      "epoch": 0.04983012457531144,
      "grad_norm": 2.1852588653564453,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.3301,
      "step": 11
    },
    {
      "epoch": 0.05436013590033975,
      "grad_norm": 1.943076491355896,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.4835,
      "step": 12
    },
    {
      "epoch": 0.05889014722536806,
      "grad_norm": 2.193492889404297,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.3817,
      "step": 13
    },
    {
      "epoch": 0.06342015855039637,
      "grad_norm": 1.6531401872634888,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.0723,
      "step": 14
    },
    {
      "epoch": 0.06795016987542468,
      "grad_norm": 2.001420259475708,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.2682,
      "step": 15
    },
    {
      "epoch": 0.072480181200453,
      "grad_norm": 1.78839910030365,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.3396,
      "step": 16
    },
    {
      "epoch": 0.07701019252548132,
      "grad_norm": 1.6972793340682983,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.2756,
      "step": 17
    },
    {
      "epoch": 0.08154020385050963,
      "grad_norm": 1.6110180616378784,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.264,
      "step": 18
    },
    {
      "epoch": 0.08607021517553794,
      "grad_norm": 1.5519018173217773,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.2932,
      "step": 19
    },
    {
      "epoch": 0.09060022650056625,
      "grad_norm": 1.9791890382766724,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3924,
      "step": 20
    },
    {
      "epoch": 0.09513023782559456,
      "grad_norm": 1.580385684967041,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4547,
      "step": 21
    },
    {
      "epoch": 0.09966024915062288,
      "grad_norm": 1.379241943359375,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.0998,
      "step": 22
    },
    {
      "epoch": 0.10419026047565119,
      "grad_norm": 1.5138109922409058,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.3034,
      "step": 23
    },
    {
      "epoch": 0.1087202718006795,
      "grad_norm": 1.592476725578308,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.1654,
      "step": 24
    },
    {
      "epoch": 0.11325028312570781,
      "grad_norm": 1.5511529445648193,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.2666,
      "step": 25
    },
    {
      "epoch": 0.11778029445073612,
      "grad_norm": 1.4859206676483154,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.1565,
      "step": 26
    },
    {
      "epoch": 0.12231030577576443,
      "grad_norm": 1.6693722009658813,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.3379,
      "step": 27
    },
    {
      "epoch": 0.12684031710079274,
      "grad_norm": 1.4977747201919556,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.269,
      "step": 28
    },
    {
      "epoch": 0.13137032842582105,
      "grad_norm": 1.4806654453277588,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.3358,
      "step": 29
    },
    {
      "epoch": 0.13590033975084936,
      "grad_norm": 1.6188100576400757,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.2231,
      "step": 30
    },
    {
      "epoch": 0.1404303510758777,
      "grad_norm": 1.5160895586013794,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.2673,
      "step": 31
    },
    {
      "epoch": 0.144960362400906,
      "grad_norm": 1.4920134544372559,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.2324,
      "step": 32
    },
    {
      "epoch": 0.14949037372593432,
      "grad_norm": 1.5743350982666016,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.328,
      "step": 33
    },
    {
      "epoch": 0.15402038505096263,
      "grad_norm": 1.8637919425964355,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.3434,
      "step": 34
    },
    {
      "epoch": 0.15855039637599094,
      "grad_norm": 1.6309973001480103,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.2696,
      "step": 35
    },
    {
      "epoch": 0.16308040770101925,
      "grad_norm": 1.7564451694488525,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.3946,
      "step": 36
    },
    {
      "epoch": 0.16761041902604756,
      "grad_norm": 1.9811019897460938,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.5074,
      "step": 37
    },
    {
      "epoch": 0.17214043035107587,
      "grad_norm": 1.909197211265564,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.3898,
      "step": 38
    },
    {
      "epoch": 0.17667044167610418,
      "grad_norm": 2.2455480098724365,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.4486,
      "step": 39
    },
    {
      "epoch": 0.1812004530011325,
      "grad_norm": 1.9569215774536133,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.3532,
      "step": 40
    },
    {
      "epoch": 0.1857304643261608,
      "grad_norm": 1.9679290056228638,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.348,
      "step": 41
    },
    {
      "epoch": 0.19026047565118911,
      "grad_norm": 2.0249273777008057,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.3808,
      "step": 42
    },
    {
      "epoch": 0.19479048697621745,
      "grad_norm": 2.4989473819732666,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.6074,
      "step": 43
    },
    {
      "epoch": 0.19932049830124576,
      "grad_norm": 2.05576491355896,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.2555,
      "step": 44
    },
    {
      "epoch": 0.20385050962627407,
      "grad_norm": 2.432482957839966,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.5167,
      "step": 45
    },
    {
      "epoch": 0.20838052095130238,
      "grad_norm": 2.685556173324585,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5267,
      "step": 46
    },
    {
      "epoch": 0.2129105322763307,
      "grad_norm": 2.2938554286956787,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.3838,
      "step": 47
    },
    {
      "epoch": 0.217440543601359,
      "grad_norm": 2.5568289756774902,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.4402,
      "step": 48
    },
    {
      "epoch": 0.22197055492638731,
      "grad_norm": 2.9967710971832275,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.5878,
      "step": 49
    },
    {
      "epoch": 0.22650056625141562,
      "grad_norm": 2.4882094860076904,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.3349,
      "step": 50
    },
    {
      "epoch": 0.22650056625141562,
      "eval_loss": 1.367936611175537,
      "eval_runtime": 28.0719,
      "eval_samples_per_second": 13.252,
      "eval_steps_per_second": 3.313,
      "step": 50
    },
    {
      "epoch": 0.23103057757644394,
      "grad_norm": 3.083442211151123,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.9484,
      "step": 51
    },
    {
      "epoch": 0.23556058890147225,
      "grad_norm": 2.4934751987457275,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.1062,
      "step": 52
    },
    {
      "epoch": 0.24009060022650056,
      "grad_norm": 2.051694869995117,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.274,
      "step": 53
    },
    {
      "epoch": 0.24462061155152887,
      "grad_norm": 1.625427484512329,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.0954,
      "step": 54
    },
    {
      "epoch": 0.2491506228765572,
      "grad_norm": 1.1984504461288452,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.9497,
      "step": 55
    },
    {
      "epoch": 0.2536806342015855,
      "grad_norm": 1.2636550664901733,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.8118,
      "step": 56
    },
    {
      "epoch": 0.2582106455266138,
      "grad_norm": 1.199966549873352,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.9799,
      "step": 57
    },
    {
      "epoch": 0.2627406568516421,
      "grad_norm": 1.0819405317306519,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.8596,
      "step": 58
    },
    {
      "epoch": 0.26727066817667045,
      "grad_norm": 1.2944135665893555,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.1552,
      "step": 59
    },
    {
      "epoch": 0.2718006795016987,
      "grad_norm": 1.1227797269821167,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.1225,
      "step": 60
    },
    {
      "epoch": 0.27633069082672707,
      "grad_norm": 1.0874030590057373,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.9369,
      "step": 61
    },
    {
      "epoch": 0.2808607021517554,
      "grad_norm": 1.2239603996276855,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.8961,
      "step": 62
    },
    {
      "epoch": 0.2853907134767837,
      "grad_norm": 1.1895838975906372,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.0421,
      "step": 63
    },
    {
      "epoch": 0.289920724801812,
      "grad_norm": 1.2056409120559692,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.0494,
      "step": 64
    },
    {
      "epoch": 0.2944507361268403,
      "grad_norm": 1.1640489101409912,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.0208,
      "step": 65
    },
    {
      "epoch": 0.29898074745186864,
      "grad_norm": 1.1964879035949707,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.0717,
      "step": 66
    },
    {
      "epoch": 0.3035107587768969,
      "grad_norm": 1.4440969228744507,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.2271,
      "step": 67
    },
    {
      "epoch": 0.30804077010192527,
      "grad_norm": 1.2717689275741577,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.9964,
      "step": 68
    },
    {
      "epoch": 0.31257078142695355,
      "grad_norm": 1.4603880643844604,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.0625,
      "step": 69
    },
    {
      "epoch": 0.3171007927519819,
      "grad_norm": 1.2224987745285034,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.989,
      "step": 70
    },
    {
      "epoch": 0.32163080407701017,
      "grad_norm": 1.2555934190750122,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.0447,
      "step": 71
    },
    {
      "epoch": 0.3261608154020385,
      "grad_norm": 1.3391551971435547,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1567,
      "step": 72
    },
    {
      "epoch": 0.33069082672706684,
      "grad_norm": 1.2767844200134277,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.1225,
      "step": 73
    },
    {
      "epoch": 0.3352208380520951,
      "grad_norm": 1.3020209074020386,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.094,
      "step": 74
    },
    {
      "epoch": 0.33975084937712347,
      "grad_norm": 1.3483256101608276,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.9928,
      "step": 75
    },
    {
      "epoch": 0.34428086070215175,
      "grad_norm": 1.3543988466262817,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.7901,
      "step": 76
    },
    {
      "epoch": 0.3488108720271801,
      "grad_norm": 1.3716752529144287,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.168,
      "step": 77
    },
    {
      "epoch": 0.35334088335220837,
      "grad_norm": 1.3246517181396484,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.9668,
      "step": 78
    },
    {
      "epoch": 0.3578708946772367,
      "grad_norm": 1.6303520202636719,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3676,
      "step": 79
    },
    {
      "epoch": 0.362400906002265,
      "grad_norm": 1.7009350061416626,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3403,
      "step": 80
    },
    {
      "epoch": 0.3669309173272933,
      "grad_norm": 1.724084496498108,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.3636,
      "step": 81
    },
    {
      "epoch": 0.3714609286523216,
      "grad_norm": 1.5446453094482422,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.9936,
      "step": 82
    },
    {
      "epoch": 0.37599093997734995,
      "grad_norm": 1.651163935661316,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2425,
      "step": 83
    },
    {
      "epoch": 0.38052095130237823,
      "grad_norm": 1.673162817955017,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4435,
      "step": 84
    },
    {
      "epoch": 0.38505096262740657,
      "grad_norm": 1.5956557989120483,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.153,
      "step": 85
    },
    {
      "epoch": 0.3895809739524349,
      "grad_norm": 1.5709388256072998,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.2736,
      "step": 86
    },
    {
      "epoch": 0.3941109852774632,
      "grad_norm": 1.7328740358352661,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.2225,
      "step": 87
    },
    {
      "epoch": 0.3986409966024915,
      "grad_norm": 1.8774045705795288,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.2324,
      "step": 88
    },
    {
      "epoch": 0.4031710079275198,
      "grad_norm": 1.797852873802185,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.2112,
      "step": 89
    },
    {
      "epoch": 0.40770101925254815,
      "grad_norm": 1.7267102003097534,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.0336,
      "step": 90
    },
    {
      "epoch": 0.41223103057757643,
      "grad_norm": 1.8740278482437134,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.1558,
      "step": 91
    },
    {
      "epoch": 0.41676104190260477,
      "grad_norm": 1.7628755569458008,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2327,
      "step": 92
    },
    {
      "epoch": 0.42129105322763305,
      "grad_norm": 2.058300256729126,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.3224,
      "step": 93
    },
    {
      "epoch": 0.4258210645526614,
      "grad_norm": 2.2820844650268555,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.5313,
      "step": 94
    },
    {
      "epoch": 0.43035107587768967,
      "grad_norm": 1.8171586990356445,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.1416,
      "step": 95
    },
    {
      "epoch": 0.434881087202718,
      "grad_norm": 2.007176160812378,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.2979,
      "step": 96
    },
    {
      "epoch": 0.43941109852774635,
      "grad_norm": 2.0509390830993652,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.3552,
      "step": 97
    },
    {
      "epoch": 0.44394110985277463,
      "grad_norm": 2.435265302658081,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.425,
      "step": 98
    },
    {
      "epoch": 0.44847112117780297,
      "grad_norm": 2.0283591747283936,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.2923,
      "step": 99
    },
    {
      "epoch": 0.45300113250283125,
      "grad_norm": 2.5698583126068115,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.3523,
      "step": 100
    },
    {
      "epoch": 0.45300113250283125,
      "eval_loss": 1.2774827480316162,
      "eval_runtime": 28.0523,
      "eval_samples_per_second": 13.261,
      "eval_steps_per_second": 3.315,
      "step": 100
    },
    {
      "epoch": 0.4575311438278596,
      "grad_norm": 3.059459924697876,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.0828,
      "step": 101
    },
    {
      "epoch": 0.46206115515288787,
      "grad_norm": 2.08219051361084,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.8233,
      "step": 102
    },
    {
      "epoch": 0.4665911664779162,
      "grad_norm": 2.0053317546844482,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.9212,
      "step": 103
    },
    {
      "epoch": 0.4711211778029445,
      "grad_norm": 1.6165436506271362,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.0142,
      "step": 104
    },
    {
      "epoch": 0.47565118912797283,
      "grad_norm": 1.3095600605010986,
      "learning_rate": 5e-05,
      "loss": 0.8596,
      "step": 105
    },
    {
      "epoch": 0.4801812004530011,
      "grad_norm": 1.4697264432907104,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.9671,
      "step": 106
    },
    {
      "epoch": 0.48471121177802945,
      "grad_norm": 1.2476412057876587,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.8159,
      "step": 107
    },
    {
      "epoch": 0.48924122310305773,
      "grad_norm": 1.1964303255081177,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.8864,
      "step": 108
    },
    {
      "epoch": 0.49377123442808607,
      "grad_norm": 1.3065221309661865,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.8765,
      "step": 109
    },
    {
      "epoch": 0.4983012457531144,
      "grad_norm": 1.1459369659423828,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.8871,
      "step": 110
    },
    {
      "epoch": 0.5028312570781427,
      "grad_norm": 1.342427372932434,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.0379,
      "step": 111
    },
    {
      "epoch": 0.507361268403171,
      "grad_norm": 1.2118736505508423,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.9316,
      "step": 112
    },
    {
      "epoch": 0.5118912797281994,
      "grad_norm": 1.1911784410476685,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.9987,
      "step": 113
    },
    {
      "epoch": 0.5164212910532276,
      "grad_norm": 1.151218056678772,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.8602,
      "step": 114
    },
    {
      "epoch": 0.5209513023782559,
      "grad_norm": 1.2503676414489746,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.0233,
      "step": 115
    },
    {
      "epoch": 0.5254813137032842,
      "grad_norm": 1.1924099922180176,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.949,
      "step": 116
    },
    {
      "epoch": 0.5300113250283126,
      "grad_norm": 1.1373696327209473,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.8729,
      "step": 117
    },
    {
      "epoch": 0.5345413363533409,
      "grad_norm": 1.1316026449203491,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.79,
      "step": 118
    },
    {
      "epoch": 0.5390713476783692,
      "grad_norm": 1.1761401891708374,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.9082,
      "step": 119
    },
    {
      "epoch": 0.5436013590033975,
      "grad_norm": 1.2113300561904907,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.727,
      "step": 120
    },
    {
      "epoch": 0.5481313703284258,
      "grad_norm": 1.1797014474868774,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.8959,
      "step": 121
    },
    {
      "epoch": 0.5526613816534541,
      "grad_norm": 1.1710346937179565,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.8368,
      "step": 122
    },
    {
      "epoch": 0.5571913929784824,
      "grad_norm": 1.244189739227295,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.8148,
      "step": 123
    },
    {
      "epoch": 0.5617214043035108,
      "grad_norm": 1.2178705930709839,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.8379,
      "step": 124
    },
    {
      "epoch": 0.5662514156285391,
      "grad_norm": 1.1625704765319824,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.9036,
      "step": 125
    },
    {
      "epoch": 0.5707814269535674,
      "grad_norm": 1.4666720628738403,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.0237,
      "step": 126
    },
    {
      "epoch": 0.5753114382785957,
      "grad_norm": 1.4170050621032715,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.0469,
      "step": 127
    },
    {
      "epoch": 0.579841449603624,
      "grad_norm": 1.5714566707611084,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.1735,
      "step": 128
    },
    {
      "epoch": 0.5843714609286523,
      "grad_norm": 1.328711986541748,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.9443,
      "step": 129
    },
    {
      "epoch": 0.5889014722536806,
      "grad_norm": 1.5815150737762451,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2009,
      "step": 130
    },
    {
      "epoch": 0.5934314835787089,
      "grad_norm": 1.3264386653900146,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.8751,
      "step": 131
    },
    {
      "epoch": 0.5979614949037373,
      "grad_norm": 1.4832333326339722,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.2158,
      "step": 132
    },
    {
      "epoch": 0.6024915062287656,
      "grad_norm": 1.4787821769714355,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.9265,
      "step": 133
    },
    {
      "epoch": 0.6070215175537939,
      "grad_norm": 1.6038435697555542,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.0933,
      "step": 134
    },
    {
      "epoch": 0.6115515288788222,
      "grad_norm": 1.6053036451339722,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2147,
      "step": 135
    },
    {
      "epoch": 0.6160815402038505,
      "grad_norm": 1.4098361730575562,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.887,
      "step": 136
    },
    {
      "epoch": 0.6206115515288788,
      "grad_norm": 1.9740478992462158,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.4018,
      "step": 137
    },
    {
      "epoch": 0.6251415628539071,
      "grad_norm": 1.8554123640060425,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.5304,
      "step": 138
    },
    {
      "epoch": 0.6296715741789355,
      "grad_norm": 1.898862361907959,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.1713,
      "step": 139
    },
    {
      "epoch": 0.6342015855039638,
      "grad_norm": 1.9535177946090698,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.4764,
      "step": 140
    },
    {
      "epoch": 0.638731596828992,
      "grad_norm": 1.8967586755752563,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.9364,
      "step": 141
    },
    {
      "epoch": 0.6432616081540203,
      "grad_norm": 2.021707534790039,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.2008,
      "step": 142
    },
    {
      "epoch": 0.6477916194790487,
      "grad_norm": 1.9848196506500244,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1432,
      "step": 143
    },
    {
      "epoch": 0.652321630804077,
      "grad_norm": 2.0446388721466064,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.3492,
      "step": 144
    },
    {
      "epoch": 0.6568516421291053,
      "grad_norm": 1.850926160812378,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.2799,
      "step": 145
    },
    {
      "epoch": 0.6613816534541337,
      "grad_norm": 1.6576409339904785,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.0239,
      "step": 146
    },
    {
      "epoch": 0.665911664779162,
      "grad_norm": 2.070793390274048,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.1513,
      "step": 147
    },
    {
      "epoch": 0.6704416761041903,
      "grad_norm": 2.44506573677063,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.3194,
      "step": 148
    },
    {
      "epoch": 0.6749716874292185,
      "grad_norm": 2.0500857830047607,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.3177,
      "step": 149
    },
    {
      "epoch": 0.6795016987542469,
      "grad_norm": 2.224750518798828,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.3438,
      "step": 150
    },
    {
      "epoch": 0.6795016987542469,
      "eval_loss": 1.0172207355499268,
      "eval_runtime": 28.0543,
      "eval_samples_per_second": 13.26,
      "eval_steps_per_second": 3.315,
      "step": 150
    },
    {
      "epoch": 0.6840317100792752,
      "grad_norm": 1.2872601747512817,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.8327,
      "step": 151
    },
    {
      "epoch": 0.6885617214043035,
      "grad_norm": 1.2982555627822876,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.9131,
      "step": 152
    },
    {
      "epoch": 0.6930917327293318,
      "grad_norm": 1.5365681648254395,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.9211,
      "step": 153
    },
    {
      "epoch": 0.6976217440543602,
      "grad_norm": 1.5271148681640625,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.0205,
      "step": 154
    },
    {
      "epoch": 0.7021517553793885,
      "grad_norm": 1.5944899320602417,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.9993,
      "step": 155
    },
    {
      "epoch": 0.7066817667044167,
      "grad_norm": 1.3695976734161377,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.0376,
      "step": 156
    },
    {
      "epoch": 0.711211778029445,
      "grad_norm": 1.1654540300369263,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.8206,
      "step": 157
    },
    {
      "epoch": 0.7157417893544734,
      "grad_norm": 1.1761460304260254,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.8057,
      "step": 158
    },
    {
      "epoch": 0.7202718006795017,
      "grad_norm": 1.0966402292251587,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.7573,
      "step": 159
    },
    {
      "epoch": 0.72480181200453,
      "grad_norm": 1.1546568870544434,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.9628,
      "step": 160
    },
    {
      "epoch": 0.7293318233295584,
      "grad_norm": 1.2177895307540894,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.8587,
      "step": 161
    },
    {
      "epoch": 0.7338618346545867,
      "grad_norm": 1.2533050775527954,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.7749,
      "step": 162
    },
    {
      "epoch": 0.7383918459796149,
      "grad_norm": 1.2467713356018066,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.8342,
      "step": 163
    },
    {
      "epoch": 0.7429218573046432,
      "grad_norm": 1.3294156789779663,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.9558,
      "step": 164
    },
    {
      "epoch": 0.7474518686296716,
      "grad_norm": 1.2277485132217407,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.0999,
      "step": 165
    },
    {
      "epoch": 0.7519818799546999,
      "grad_norm": 1.2494876384735107,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.932,
      "step": 166
    },
    {
      "epoch": 0.7565118912797282,
      "grad_norm": 1.2311893701553345,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.0103,
      "step": 167
    },
    {
      "epoch": 0.7610419026047565,
      "grad_norm": 1.2579753398895264,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.964,
      "step": 168
    },
    {
      "epoch": 0.7655719139297849,
      "grad_norm": 1.1700468063354492,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.6902,
      "step": 169
    },
    {
      "epoch": 0.7701019252548131,
      "grad_norm": 1.2975200414657593,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.8638,
      "step": 170
    },
    {
      "epoch": 0.7746319365798414,
      "grad_norm": 1.137703537940979,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.755,
      "step": 171
    },
    {
      "epoch": 0.7791619479048698,
      "grad_norm": 1.2437477111816406,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.109,
      "step": 172
    },
    {
      "epoch": 0.7836919592298981,
      "grad_norm": 1.2752270698547363,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.8423,
      "step": 173
    },
    {
      "epoch": 0.7882219705549264,
      "grad_norm": 1.1494214534759521,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.7446,
      "step": 174
    },
    {
      "epoch": 0.7927519818799547,
      "grad_norm": 1.3101270198822021,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.9874,
      "step": 175
    },
    {
      "epoch": 0.797281993204983,
      "grad_norm": 1.464393973350525,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.0062,
      "step": 176
    },
    {
      "epoch": 0.8018120045300113,
      "grad_norm": 1.3159586191177368,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.9046,
      "step": 177
    },
    {
      "epoch": 0.8063420158550396,
      "grad_norm": 1.3815009593963623,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.002,
      "step": 178
    },
    {
      "epoch": 0.8108720271800679,
      "grad_norm": 1.341168999671936,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.9808,
      "step": 179
    },
    {
      "epoch": 0.8154020385050963,
      "grad_norm": 1.221179723739624,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.768,
      "step": 180
    },
    {
      "epoch": 0.8199320498301246,
      "grad_norm": 1.3718920946121216,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.9968,
      "step": 181
    },
    {
      "epoch": 0.8244620611551529,
      "grad_norm": 1.5797667503356934,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.0483,
      "step": 182
    },
    {
      "epoch": 0.8289920724801813,
      "grad_norm": 1.5867432355880737,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.0741,
      "step": 183
    },
    {
      "epoch": 0.8335220838052095,
      "grad_norm": 1.5210647583007812,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.104,
      "step": 184
    },
    {
      "epoch": 0.8380520951302378,
      "grad_norm": 1.5403170585632324,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.9884,
      "step": 185
    },
    {
      "epoch": 0.8425821064552661,
      "grad_norm": 1.5561081171035767,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.0043,
      "step": 186
    },
    {
      "epoch": 0.8471121177802945,
      "grad_norm": 1.6020222902297974,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.027,
      "step": 187
    },
    {
      "epoch": 0.8516421291053228,
      "grad_norm": 1.70779550075531,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.0802,
      "step": 188
    },
    {
      "epoch": 0.8561721404303511,
      "grad_norm": 1.6190474033355713,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.0242,
      "step": 189
    },
    {
      "epoch": 0.8607021517553793,
      "grad_norm": 1.9034790992736816,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.128,
      "step": 190
    },
    {
      "epoch": 0.8652321630804077,
      "grad_norm": 1.8138927221298218,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.1939,
      "step": 191
    },
    {
      "epoch": 0.869762174405436,
      "grad_norm": 1.6941299438476562,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.9579,
      "step": 192
    },
    {
      "epoch": 0.8742921857304643,
      "grad_norm": 2.164492130279541,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.3139,
      "step": 193
    },
    {
      "epoch": 0.8788221970554927,
      "grad_norm": 2.3437535762786865,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.1869,
      "step": 194
    },
    {
      "epoch": 0.883352208380521,
      "grad_norm": 2.543923854827881,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.5039,
      "step": 195
    },
    {
      "epoch": 0.8878822197055493,
      "grad_norm": 1.834774374961853,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.1496,
      "step": 196
    },
    {
      "epoch": 0.8924122310305775,
      "grad_norm": 2.0103721618652344,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.2342,
      "step": 197
    },
    {
      "epoch": 0.8969422423556059,
      "grad_norm": 2.2426156997680664,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.3022,
      "step": 198
    },
    {
      "epoch": 0.9014722536806342,
      "grad_norm": 2.0831143856048584,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.0872,
      "step": 199
    },
    {
      "epoch": 0.9060022650056625,
      "grad_norm": 2.114161491394043,
      "learning_rate": 0.0,
      "loss": 1.0237,
      "step": 200
    },
    {
      "epoch": 0.9060022650056625,
      "eval_loss": 0.9678028225898743,
      "eval_runtime": 28.0354,
      "eval_samples_per_second": 13.269,
      "eval_steps_per_second": 3.317,
      "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": 3.017027157491712e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}