{
  "best_metric": 0.6999725103378296,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.6872852233676976,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003436426116838488,
      "grad_norm": 6.875288009643555,
      "learning_rate": 1e-05,
      "loss": 4.5691,
      "step": 1
    },
    {
      "epoch": 0.003436426116838488,
      "eval_loss": 1.1649625301361084,
      "eval_runtime": 56.8704,
      "eval_samples_per_second": 8.616,
      "eval_steps_per_second": 2.163,
      "step": 1
    },
    {
      "epoch": 0.006872852233676976,
      "grad_norm": 10.01762580871582,
      "learning_rate": 2e-05,
      "loss": 4.7735,
      "step": 2
    },
    {
      "epoch": 0.010309278350515464,
      "grad_norm": 8.196460723876953,
      "learning_rate": 3e-05,
      "loss": 5.4559,
      "step": 3
    },
    {
      "epoch": 0.013745704467353952,
      "grad_norm": 7.672198295593262,
      "learning_rate": 4e-05,
      "loss": 4.582,
      "step": 4
    },
    {
      "epoch": 0.01718213058419244,
      "grad_norm": 4.969404220581055,
      "learning_rate": 5e-05,
      "loss": 3.4696,
      "step": 5
    },
    {
      "epoch": 0.020618556701030927,
      "grad_norm": 4.3889851570129395,
      "learning_rate": 6e-05,
      "loss": 3.3103,
      "step": 6
    },
    {
      "epoch": 0.024054982817869417,
      "grad_norm": 4.012556076049805,
      "learning_rate": 7e-05,
      "loss": 3.2335,
      "step": 7
    },
    {
      "epoch": 0.027491408934707903,
      "grad_norm": 3.344139814376831,
      "learning_rate": 8e-05,
      "loss": 3.0476,
      "step": 8
    },
    {
      "epoch": 0.030927835051546393,
      "grad_norm": 3.173506021499634,
      "learning_rate": 9e-05,
      "loss": 3.1844,
      "step": 9
    },
    {
      "epoch": 0.03436426116838488,
      "grad_norm": 2.4493091106414795,
      "learning_rate": 0.0001,
      "loss": 2.782,
      "step": 10
    },
    {
      "epoch": 0.037800687285223365,
      "grad_norm": 3.146341323852539,
      "learning_rate": 9.999316524962345e-05,
      "loss": 3.0659,
      "step": 11
    },
    {
      "epoch": 0.041237113402061855,
      "grad_norm": 2.8679311275482178,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.7422,
      "step": 12
    },
    {
      "epoch": 0.044673539518900345,
      "grad_norm": 2.7918779850006104,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.931,
      "step": 13
    },
    {
      "epoch": 0.048109965635738834,
      "grad_norm": 2.896129846572876,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.8184,
      "step": 14
    },
    {
      "epoch": 0.05154639175257732,
      "grad_norm": 2.6621735095977783,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.6664,
      "step": 15
    },
    {
      "epoch": 0.054982817869415807,
      "grad_norm": 2.8896968364715576,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.7089,
      "step": 16
    },
    {
      "epoch": 0.058419243986254296,
      "grad_norm": 3.0421786308288574,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.7472,
      "step": 17
    },
    {
      "epoch": 0.061855670103092786,
      "grad_norm": 2.725599765777588,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.6994,
      "step": 18
    },
    {
      "epoch": 0.06529209621993128,
      "grad_norm": 2.331462860107422,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.6058,
      "step": 19
    },
    {
      "epoch": 0.06872852233676977,
      "grad_norm": 2.9271135330200195,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.6624,
      "step": 20
    },
    {
      "epoch": 0.07216494845360824,
      "grad_norm": 2.924933433532715,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.5572,
      "step": 21
    },
    {
      "epoch": 0.07560137457044673,
      "grad_norm": 2.7112512588500977,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.6309,
      "step": 22
    },
    {
      "epoch": 0.07903780068728522,
      "grad_norm": 2.8630990982055664,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.5248,
      "step": 23
    },
    {
      "epoch": 0.08247422680412371,
      "grad_norm": 2.5011701583862305,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.6045,
      "step": 24
    },
    {
      "epoch": 0.0859106529209622,
      "grad_norm": 3.070207118988037,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.869,
      "step": 25
    },
    {
      "epoch": 0.08934707903780069,
      "grad_norm": 2.714486837387085,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.8448,
      "step": 26
    },
    {
      "epoch": 0.09278350515463918,
      "grad_norm": 3.2341771125793457,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.6683,
      "step": 27
    },
    {
      "epoch": 0.09621993127147767,
      "grad_norm": 2.5088119506835938,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.6063,
      "step": 28
    },
    {
      "epoch": 0.09965635738831616,
      "grad_norm": 2.819406270980835,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.6812,
      "step": 29
    },
    {
      "epoch": 0.10309278350515463,
      "grad_norm": 2.6176741123199463,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.8282,
      "step": 30
    },
    {
      "epoch": 0.10652920962199312,
      "grad_norm": 2.890087604522705,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.9375,
      "step": 31
    },
    {
      "epoch": 0.10996563573883161,
      "grad_norm": 3.169712781906128,
      "learning_rate": 9.672822322997305e-05,
      "loss": 3.0984,
      "step": 32
    },
    {
      "epoch": 0.1134020618556701,
      "grad_norm": 3.0394792556762695,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.6701,
      "step": 33
    },
    {
      "epoch": 0.11683848797250859,
      "grad_norm": 3.2077884674072266,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.9322,
      "step": 34
    },
    {
      "epoch": 0.12027491408934708,
      "grad_norm": 2.8351423740386963,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.9144,
      "step": 35
    },
    {
      "epoch": 0.12371134020618557,
      "grad_norm": 3.7410457134246826,
      "learning_rate": 9.545032675245813e-05,
      "loss": 3.1276,
      "step": 36
    },
    {
      "epoch": 0.12714776632302405,
      "grad_norm": 3.0850861072540283,
      "learning_rate": 9.509956150664796e-05,
      "loss": 3.2167,
      "step": 37
    },
    {
      "epoch": 0.13058419243986255,
      "grad_norm": 3.5275065898895264,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.67,
      "step": 38
    },
    {
      "epoch": 0.13402061855670103,
      "grad_norm": 3.8704416751861572,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.9465,
      "step": 39
    },
    {
      "epoch": 0.13745704467353953,
      "grad_norm": 4.236602306365967,
      "learning_rate": 9.397368756032445e-05,
      "loss": 3.1699,
      "step": 40
    },
    {
      "epoch": 0.140893470790378,
      "grad_norm": 3.579802989959717,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.5988,
      "step": 41
    },
    {
      "epoch": 0.14432989690721648,
      "grad_norm": 3.929170608520508,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.7835,
      "step": 42
    },
    {
      "epoch": 0.14776632302405499,
      "grad_norm": 4.23232889175415,
      "learning_rate": 9.273963562927695e-05,
      "loss": 3.249,
      "step": 43
    },
    {
      "epoch": 0.15120274914089346,
      "grad_norm": 4.3421454429626465,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.0539,
      "step": 44
    },
    {
      "epoch": 0.15463917525773196,
      "grad_norm": 5.814420223236084,
      "learning_rate": 9.185832391312644e-05,
      "loss": 3.1705,
      "step": 45
    },
    {
      "epoch": 0.15807560137457044,
      "grad_norm": 4.802507400512695,
      "learning_rate": 9.140044155740101e-05,
      "loss": 3.4311,
      "step": 46
    },
    {
      "epoch": 0.16151202749140894,
      "grad_norm": 5.132009029388428,
      "learning_rate": 9.093124073433463e-05,
      "loss": 3.1524,
      "step": 47
    },
    {
      "epoch": 0.16494845360824742,
      "grad_norm": 6.066173553466797,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.9363,
      "step": 48
    },
    {
      "epoch": 0.16838487972508592,
      "grad_norm": 8.14843463897705,
      "learning_rate": 8.995939984474624e-05,
      "loss": 3.9255,
      "step": 49
    },
    {
      "epoch": 0.1718213058419244,
      "grad_norm": 11.949299812316895,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.9737,
      "step": 50
    },
    {
      "epoch": 0.1718213058419244,
      "eval_loss": 0.7489818334579468,
      "eval_runtime": 58.1061,
      "eval_samples_per_second": 8.433,
      "eval_steps_per_second": 2.117,
      "step": 50
    },
    {
      "epoch": 0.17525773195876287,
      "grad_norm": 3.229118824005127,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.9858,
      "step": 51
    },
    {
      "epoch": 0.17869415807560138,
      "grad_norm": 3.4312267303466797,
      "learning_rate": 8.842005554284296e-05,
      "loss": 4.7884,
      "step": 52
    },
    {
      "epoch": 0.18213058419243985,
      "grad_norm": 2.815904140472412,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.1316,
      "step": 53
    },
    {
      "epoch": 0.18556701030927836,
      "grad_norm": 2.907052755355835,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.8479,
      "step": 54
    },
    {
      "epoch": 0.18900343642611683,
      "grad_norm": 2.1939303874969482,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.9194,
      "step": 55
    },
    {
      "epoch": 0.19243986254295534,
      "grad_norm": 2.066952705383301,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.71,
      "step": 56
    },
    {
      "epoch": 0.1958762886597938,
      "grad_norm": 2.301304817199707,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.3676,
      "step": 57
    },
    {
      "epoch": 0.19931271477663232,
      "grad_norm": 3.6853837966918945,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.4181,
      "step": 58
    },
    {
      "epoch": 0.2027491408934708,
      "grad_norm": 2.3933732509613037,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.6629,
      "step": 59
    },
    {
      "epoch": 0.20618556701030927,
      "grad_norm": 2.310511350631714,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.5932,
      "step": 60
    },
    {
      "epoch": 0.20962199312714777,
      "grad_norm": 2.1882212162017822,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.5533,
      "step": 61
    },
    {
      "epoch": 0.21305841924398625,
      "grad_norm": 2.224141836166382,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.6595,
      "step": 62
    },
    {
      "epoch": 0.21649484536082475,
      "grad_norm": 2.0371408462524414,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.4645,
      "step": 63
    },
    {
      "epoch": 0.21993127147766323,
      "grad_norm": 1.9904556274414062,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.815,
      "step": 64
    },
    {
      "epoch": 0.22336769759450173,
      "grad_norm": 1.9936702251434326,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.6361,
      "step": 65
    },
    {
      "epoch": 0.2268041237113402,
      "grad_norm": 2.063617467880249,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.8328,
      "step": 66
    },
    {
      "epoch": 0.23024054982817868,
      "grad_norm": 1.778960108757019,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.5526,
      "step": 67
    },
    {
      "epoch": 0.23367697594501718,
      "grad_norm": 1.8721736669540405,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.2001,
      "step": 68
    },
    {
      "epoch": 0.23711340206185566,
      "grad_norm": 2.1718900203704834,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.6618,
      "step": 69
    },
    {
      "epoch": 0.24054982817869416,
      "grad_norm": 1.905144453048706,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.4549,
      "step": 70
    },
    {
      "epoch": 0.24398625429553264,
      "grad_norm": 2.322458028793335,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.6667,
      "step": 71
    },
    {
      "epoch": 0.24742268041237114,
      "grad_norm": 2.112133741378784,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.4297,
      "step": 72
    },
    {
      "epoch": 0.2508591065292096,
      "grad_norm": 2.050910472869873,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.512,
      "step": 73
    },
    {
      "epoch": 0.2542955326460481,
      "grad_norm": 2.034681558609009,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.4437,
      "step": 74
    },
    {
      "epoch": 0.25773195876288657,
      "grad_norm": 1.984123945236206,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.5732,
      "step": 75
    },
    {
      "epoch": 0.2611683848797251,
      "grad_norm": 2.2462446689605713,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.6211,
      "step": 76
    },
    {
      "epoch": 0.2646048109965636,
      "grad_norm": 2.473987102508545,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.7826,
      "step": 77
    },
    {
      "epoch": 0.26804123711340205,
      "grad_norm": 2.3901572227478027,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.5584,
      "step": 78
    },
    {
      "epoch": 0.27147766323024053,
      "grad_norm": 2.2320001125335693,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.4833,
      "step": 79
    },
    {
      "epoch": 0.27491408934707906,
      "grad_norm": 2.112913131713867,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.4669,
      "step": 80
    },
    {
      "epoch": 0.27835051546391754,
      "grad_norm": 2.4999232292175293,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.633,
      "step": 81
    },
    {
      "epoch": 0.281786941580756,
      "grad_norm": 2.5236027240753174,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.6726,
      "step": 82
    },
    {
      "epoch": 0.2852233676975945,
      "grad_norm": 2.610469341278076,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.6783,
      "step": 83
    },
    {
      "epoch": 0.28865979381443296,
      "grad_norm": 2.888082504272461,
      "learning_rate": 6.701465872208216e-05,
      "loss": 3.0809,
      "step": 84
    },
    {
      "epoch": 0.2920962199312715,
      "grad_norm": 3.045673370361328,
      "learning_rate": 6.623497346023418e-05,
      "loss": 3.2255,
      "step": 85
    },
    {
      "epoch": 0.29553264604810997,
      "grad_norm": 3.2517120838165283,
      "learning_rate": 6.545084971874738e-05,
      "loss": 3.0398,
      "step": 86
    },
    {
      "epoch": 0.29896907216494845,
      "grad_norm": 2.757512331008911,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.72,
      "step": 87
    },
    {
      "epoch": 0.3024054982817869,
      "grad_norm": 2.845644235610962,
      "learning_rate": 6.387014543809223e-05,
      "loss": 3.4635,
      "step": 88
    },
    {
      "epoch": 0.30584192439862545,
      "grad_norm": 2.752856492996216,
      "learning_rate": 6.307399704769099e-05,
      "loss": 3.3043,
      "step": 89
    },
    {
      "epoch": 0.30927835051546393,
      "grad_norm": 3.0697596073150635,
      "learning_rate": 6.227427435703997e-05,
      "loss": 3.2902,
      "step": 90
    },
    {
      "epoch": 0.3127147766323024,
      "grad_norm": 3.290828227996826,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.6147,
      "step": 91
    },
    {
      "epoch": 0.3161512027491409,
      "grad_norm": 3.5331523418426514,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.3361,
      "step": 92
    },
    {
      "epoch": 0.31958762886597936,
      "grad_norm": 3.4285762310028076,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.216,
      "step": 93
    },
    {
      "epoch": 0.3230240549828179,
      "grad_norm": 3.717268466949463,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.1196,
      "step": 94
    },
    {
      "epoch": 0.32646048109965636,
      "grad_norm": 3.960413932800293,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 3.3172,
      "step": 95
    },
    {
      "epoch": 0.32989690721649484,
      "grad_norm": 4.5019850730896,
      "learning_rate": 5.74131823855921e-05,
      "loss": 3.2663,
      "step": 96
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 6.533929347991943,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 3.5329,
      "step": 97
    },
    {
      "epoch": 0.33676975945017185,
      "grad_norm": 5.7289719581604,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.3525,
      "step": 98
    },
    {
      "epoch": 0.3402061855670103,
      "grad_norm": 5.30385684967041,
      "learning_rate": 5.495227651252315e-05,
      "loss": 3.2372,
      "step": 99
    },
    {
      "epoch": 0.3436426116838488,
      "grad_norm": 7.11011266708374,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 4.2534,
      "step": 100
    },
    {
      "epoch": 0.3436426116838488,
      "eval_loss": 0.7277113199234009,
      "eval_runtime": 58.1102,
      "eval_samples_per_second": 8.432,
      "eval_steps_per_second": 2.117,
      "step": 100
    },
    {
      "epoch": 0.3470790378006873,
      "grad_norm": 6.198427200317383,
      "learning_rate": 5.330452921628497e-05,
      "loss": 3.676,
      "step": 101
    },
    {
      "epoch": 0.35051546391752575,
      "grad_norm": 3.831768751144409,
      "learning_rate": 5.247918773366112e-05,
      "loss": 4.3918,
      "step": 102
    },
    {
      "epoch": 0.3539518900343643,
      "grad_norm": 3.025773048400879,
      "learning_rate": 5.165316846586541e-05,
      "loss": 3.9168,
      "step": 103
    },
    {
      "epoch": 0.35738831615120276,
      "grad_norm": 2.913081407546997,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.957,
      "step": 104
    },
    {
      "epoch": 0.36082474226804123,
      "grad_norm": 1.924337387084961,
      "learning_rate": 5e-05,
      "loss": 2.7171,
      "step": 105
    },
    {
      "epoch": 0.3642611683848797,
      "grad_norm": 1.7787362337112427,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.5506,
      "step": 106
    },
    {
      "epoch": 0.36769759450171824,
      "grad_norm": 1.7331814765930176,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.606,
      "step": 107
    },
    {
      "epoch": 0.3711340206185567,
      "grad_norm": 1.8595000505447388,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.4872,
      "step": 108
    },
    {
      "epoch": 0.3745704467353952,
      "grad_norm": 1.7210876941680908,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.4917,
      "step": 109
    },
    {
      "epoch": 0.37800687285223367,
      "grad_norm": 2.297461986541748,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.2728,
      "step": 110
    },
    {
      "epoch": 0.38144329896907214,
      "grad_norm": 2.0554733276367188,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.718,
      "step": 111
    },
    {
      "epoch": 0.3848797250859107,
      "grad_norm": 1.8038687705993652,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.8015,
      "step": 112
    },
    {
      "epoch": 0.38831615120274915,
      "grad_norm": 1.7449924945831299,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.4377,
      "step": 113
    },
    {
      "epoch": 0.3917525773195876,
      "grad_norm": 1.9162296056747437,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.7759,
      "step": 114
    },
    {
      "epoch": 0.3951890034364261,
      "grad_norm": 1.818672776222229,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.5994,
      "step": 115
    },
    {
      "epoch": 0.39862542955326463,
      "grad_norm": 1.7228524684906006,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.3886,
      "step": 116
    },
    {
      "epoch": 0.4020618556701031,
      "grad_norm": 1.7789241075515747,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.5161,
      "step": 117
    },
    {
      "epoch": 0.4054982817869416,
      "grad_norm": 2.012725591659546,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.5251,
      "step": 118
    },
    {
      "epoch": 0.40893470790378006,
      "grad_norm": 2.0002951622009277,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.6226,
      "step": 119
    },
    {
      "epoch": 0.41237113402061853,
      "grad_norm": 1.9130884408950806,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.3705,
      "step": 120
    },
    {
      "epoch": 0.41580756013745707,
      "grad_norm": 1.7260624170303345,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.2924,
      "step": 121
    },
    {
      "epoch": 0.41924398625429554,
      "grad_norm": 1.8843971490859985,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.6148,
      "step": 122
    },
    {
      "epoch": 0.422680412371134,
      "grad_norm": 2.1272003650665283,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.8849,
      "step": 123
    },
    {
      "epoch": 0.4261168384879725,
      "grad_norm": 1.9105393886566162,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.5233,
      "step": 124
    },
    {
      "epoch": 0.42955326460481097,
      "grad_norm": 2.081491231918335,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.6334,
      "step": 125
    },
    {
      "epoch": 0.4329896907216495,
      "grad_norm": 2.01537823677063,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.7079,
      "step": 126
    },
    {
      "epoch": 0.436426116838488,
      "grad_norm": 2.0294597148895264,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.7356,
      "step": 127
    },
    {
      "epoch": 0.43986254295532645,
      "grad_norm": 2.090278387069702,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.5529,
      "step": 128
    },
    {
      "epoch": 0.44329896907216493,
      "grad_norm": 2.0905957221984863,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.6941,
      "step": 129
    },
    {
      "epoch": 0.44673539518900346,
      "grad_norm": 2.0997745990753174,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.6148,
      "step": 130
    },
    {
      "epoch": 0.45017182130584193,
      "grad_norm": 2.496063470840454,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.8272,
      "step": 131
    },
    {
      "epoch": 0.4536082474226804,
      "grad_norm": 2.2960619926452637,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.9424,
      "step": 132
    },
    {
      "epoch": 0.4570446735395189,
      "grad_norm": 2.384944438934326,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.6241,
      "step": 133
    },
    {
      "epoch": 0.46048109965635736,
      "grad_norm": 2.4643306732177734,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.5367,
      "step": 134
    },
    {
      "epoch": 0.4639175257731959,
      "grad_norm": 2.6148459911346436,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.6854,
      "step": 135
    },
    {
      "epoch": 0.46735395189003437,
      "grad_norm": 2.543748140335083,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.7542,
      "step": 136
    },
    {
      "epoch": 0.47079037800687284,
      "grad_norm": 2.4474358558654785,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.8701,
      "step": 137
    },
    {
      "epoch": 0.4742268041237113,
      "grad_norm": 2.7877655029296875,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.1246,
      "step": 138
    },
    {
      "epoch": 0.47766323024054985,
      "grad_norm": 2.8259971141815186,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.8092,
      "step": 139
    },
    {
      "epoch": 0.48109965635738833,
      "grad_norm": 2.7138278484344482,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 3.3688,
      "step": 140
    },
    {
      "epoch": 0.4845360824742268,
      "grad_norm": 2.674522638320923,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.2805,
      "step": 141
    },
    {
      "epoch": 0.4879725085910653,
      "grad_norm": 3.3910646438598633,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.3634,
      "step": 142
    },
    {
      "epoch": 0.49140893470790376,
      "grad_norm": 3.2498741149902344,
      "learning_rate": 2.061073738537635e-05,
      "loss": 3.0114,
      "step": 143
    },
    {
      "epoch": 0.4948453608247423,
      "grad_norm": 3.244091033935547,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 3.0963,
      "step": 144
    },
    {
      "epoch": 0.49828178694158076,
      "grad_norm": 3.6477789878845215,
      "learning_rate": 1.928936436551661e-05,
      "loss": 3.2332,
      "step": 145
    },
    {
      "epoch": 0.5017182130584192,
      "grad_norm": 3.879427194595337,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 3.441,
      "step": 146
    },
    {
      "epoch": 0.5051546391752577,
      "grad_norm": 4.064486503601074,
      "learning_rate": 1.800157297483417e-05,
      "loss": 3.2155,
      "step": 147
    },
    {
      "epoch": 0.5085910652920962,
      "grad_norm": 4.470992088317871,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 3.791,
      "step": 148
    },
    {
      "epoch": 0.5120274914089347,
      "grad_norm": 5.485239028930664,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 3.2473,
      "step": 149
    },
    {
      "epoch": 0.5154639175257731,
      "grad_norm": 7.886268138885498,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.1722,
      "step": 150
    },
    {
      "epoch": 0.5154639175257731,
      "eval_loss": 0.7073755264282227,
      "eval_runtime": 57.6779,
      "eval_samples_per_second": 8.495,
      "eval_steps_per_second": 2.133,
      "step": 150
    },
    {
      "epoch": 0.5189003436426117,
      "grad_norm": 2.143200397491455,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.5219,
      "step": 151
    },
    {
      "epoch": 0.5223367697594502,
      "grad_norm": 4.848270893096924,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 3.958,
      "step": 152
    },
    {
      "epoch": 0.5257731958762887,
      "grad_norm": 2.398545980453491,
      "learning_rate": 1.435357758543015e-05,
      "loss": 3.3358,
      "step": 153
    },
    {
      "epoch": 0.5292096219931272,
      "grad_norm": 2.354947090148926,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 3.3179,
      "step": 154
    },
    {
      "epoch": 0.5326460481099656,
      "grad_norm": 1.799606204032898,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.6117,
      "step": 155
    },
    {
      "epoch": 0.5360824742268041,
      "grad_norm": 1.8230576515197754,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.6321,
      "step": 156
    },
    {
      "epoch": 0.5395189003436426,
      "grad_norm": 1.6931675672531128,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.337,
      "step": 157
    },
    {
      "epoch": 0.5429553264604811,
      "grad_norm": 1.8420969247817993,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.188,
      "step": 158
    },
    {
      "epoch": 0.5463917525773195,
      "grad_norm": 1.7145228385925293,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.5306,
      "step": 159
    },
    {
      "epoch": 0.5498281786941581,
      "grad_norm": 1.7859251499176025,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.3562,
      "step": 160
    },
    {
      "epoch": 0.5532646048109966,
      "grad_norm": 1.8147993087768555,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.4893,
      "step": 161
    },
    {
      "epoch": 0.5567010309278351,
      "grad_norm": 1.7367379665374756,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.6109,
      "step": 162
    },
    {
      "epoch": 0.5601374570446735,
      "grad_norm": 1.6926087141036987,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.3642,
      "step": 163
    },
    {
      "epoch": 0.563573883161512,
      "grad_norm": 1.6831601858139038,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.614,
      "step": 164
    },
    {
      "epoch": 0.5670103092783505,
      "grad_norm": 1.8343570232391357,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.5047,
      "step": 165
    },
    {
      "epoch": 0.570446735395189,
      "grad_norm": 1.8360129594802856,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.8785,
      "step": 166
    },
    {
      "epoch": 0.5738831615120275,
      "grad_norm": 1.619503140449524,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.5612,
      "step": 167
    },
    {
      "epoch": 0.5773195876288659,
      "grad_norm": 1.8024160861968994,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.5232,
      "step": 168
    },
    {
      "epoch": 0.5807560137457045,
      "grad_norm": 1.896394968032837,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.5866,
      "step": 169
    },
    {
      "epoch": 0.584192439862543,
      "grad_norm": 1.9497900009155273,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.4935,
      "step": 170
    },
    {
      "epoch": 0.5876288659793815,
      "grad_norm": 2.036046028137207,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.6188,
      "step": 171
    },
    {
      "epoch": 0.5910652920962199,
      "grad_norm": 1.7684684991836548,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.4919,
      "step": 172
    },
    {
      "epoch": 0.5945017182130584,
      "grad_norm": 1.9862613677978516,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.6625,
      "step": 173
    },
    {
      "epoch": 0.5979381443298969,
      "grad_norm": 1.9642881155014038,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.6936,
      "step": 174
    },
    {
      "epoch": 0.6013745704467354,
      "grad_norm": 1.8774116039276123,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.3646,
      "step": 175
    },
    {
      "epoch": 0.6048109965635738,
      "grad_norm": 1.9103914499282837,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.7065,
      "step": 176
    },
    {
      "epoch": 0.6082474226804123,
      "grad_norm": 2.08084774017334,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.8205,
      "step": 177
    },
    {
      "epoch": 0.6116838487972509,
      "grad_norm": 2.1520981788635254,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.7601,
      "step": 178
    },
    {
      "epoch": 0.6151202749140894,
      "grad_norm": 2.0325026512145996,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.6144,
      "step": 179
    },
    {
      "epoch": 0.6185567010309279,
      "grad_norm": 1.9948471784591675,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.8324,
      "step": 180
    },
    {
      "epoch": 0.6219931271477663,
      "grad_norm": 2.0291099548339844,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5003,
      "step": 181
    },
    {
      "epoch": 0.6254295532646048,
      "grad_norm": 4.525008678436279,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.6767,
      "step": 182
    },
    {
      "epoch": 0.6288659793814433,
      "grad_norm": 2.4280829429626465,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.9752,
      "step": 183
    },
    {
      "epoch": 0.6323024054982818,
      "grad_norm": 2.2994818687438965,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.5894,
      "step": 184
    },
    {
      "epoch": 0.6357388316151202,
      "grad_norm": 2.3175809383392334,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.6399,
      "step": 185
    },
    {
      "epoch": 0.6391752577319587,
      "grad_norm": 2.5007894039154053,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.4445,
      "step": 186
    },
    {
      "epoch": 0.6426116838487973,
      "grad_norm": 2.471940279006958,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.9408,
      "step": 187
    },
    {
      "epoch": 0.6460481099656358,
      "grad_norm": 2.6751317977905273,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.793,
      "step": 188
    },
    {
      "epoch": 0.6494845360824743,
      "grad_norm": 3.116914987564087,
      "learning_rate": 8.247462563808817e-07,
      "loss": 3.3139,
      "step": 189
    },
    {
      "epoch": 0.6529209621993127,
      "grad_norm": 2.9802823066711426,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.9635,
      "step": 190
    },
    {
      "epoch": 0.6563573883161512,
      "grad_norm": 3.1613316535949707,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.3704,
      "step": 191
    },
    {
      "epoch": 0.6597938144329897,
      "grad_norm": 2.9843199253082275,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.7593,
      "step": 192
    },
    {
      "epoch": 0.6632302405498282,
      "grad_norm": 3.2920596599578857,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.8525,
      "step": 193
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 4.103538990020752,
      "learning_rate": 2.458548727494292e-07,
      "loss": 3.2949,
      "step": 194
    },
    {
      "epoch": 0.6701030927835051,
      "grad_norm": 4.331507682800293,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 3.5512,
      "step": 195
    },
    {
      "epoch": 0.6735395189003437,
      "grad_norm": 5.5705389976501465,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 3.3498,
      "step": 196
    },
    {
      "epoch": 0.6769759450171822,
      "grad_norm": 8.418855667114258,
      "learning_rate": 6.150154258476315e-08,
      "loss": 3.6168,
      "step": 197
    },
    {
      "epoch": 0.6804123711340206,
      "grad_norm": 7.6398701667785645,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 3.7834,
      "step": 198
    },
    {
      "epoch": 0.6838487972508591,
      "grad_norm": 9.20870304107666,
      "learning_rate": 6.834750376549792e-09,
      "loss": 3.4641,
      "step": 199
    },
    {
      "epoch": 0.6872852233676976,
      "grad_norm": 9.893965721130371,
      "learning_rate": 0.0,
      "loss": 3.9641,
      "step": 200
    },
    {
      "epoch": 0.6872852233676976,
      "eval_loss": 0.6999725103378296,
      "eval_runtime": 58.1001,
      "eval_samples_per_second": 8.434,
      "eval_steps_per_second": 2.117,
      "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": 5.071188492759859e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}