{
  "best_metric": 1.2237151861190796,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.27155465037338766,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013577732518669382,
      "grad_norm": 0.2727883458137512,
      "learning_rate": 6e-06,
      "loss": 1.3028,
      "step": 1
    },
    {
      "epoch": 0.0013577732518669382,
      "eval_loss": 1.8691651821136475,
      "eval_runtime": 20.576,
      "eval_samples_per_second": 60.313,
      "eval_steps_per_second": 15.115,
      "step": 1
    },
    {
      "epoch": 0.0027155465037338763,
      "grad_norm": 0.291520893573761,
      "learning_rate": 1.2e-05,
      "loss": 1.515,
      "step": 2
    },
    {
      "epoch": 0.004073319755600814,
      "grad_norm": 0.3538097143173218,
      "learning_rate": 1.8e-05,
      "loss": 1.3986,
      "step": 3
    },
    {
      "epoch": 0.005431093007467753,
      "grad_norm": 0.34487512707710266,
      "learning_rate": 2.4e-05,
      "loss": 1.4966,
      "step": 4
    },
    {
      "epoch": 0.006788866259334691,
      "grad_norm": 0.38797444105148315,
      "learning_rate": 3e-05,
      "loss": 1.4055,
      "step": 5
    },
    {
      "epoch": 0.008146639511201629,
      "grad_norm": 0.4047252833843231,
      "learning_rate": 3.6e-05,
      "loss": 1.4603,
      "step": 6
    },
    {
      "epoch": 0.009504412763068567,
      "grad_norm": 0.3205205798149109,
      "learning_rate": 4.2e-05,
      "loss": 1.4481,
      "step": 7
    },
    {
      "epoch": 0.010862186014935505,
      "grad_norm": 0.48629483580589294,
      "learning_rate": 4.8e-05,
      "loss": 1.4678,
      "step": 8
    },
    {
      "epoch": 0.012219959266802444,
      "grad_norm": 0.4313899278640747,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 1.5263,
      "step": 9
    },
    {
      "epoch": 0.013577732518669382,
      "grad_norm": 0.404436856508255,
      "learning_rate": 6e-05,
      "loss": 1.5395,
      "step": 10
    },
    {
      "epoch": 0.01493550577053632,
      "grad_norm": 0.35745665431022644,
      "learning_rate": 5.999589914977407e-05,
      "loss": 1.547,
      "step": 11
    },
    {
      "epoch": 0.016293279022403257,
      "grad_norm": 0.3784649968147278,
      "learning_rate": 5.998359772022778e-05,
      "loss": 1.3765,
      "step": 12
    },
    {
      "epoch": 0.017651052274270197,
      "grad_norm": 0.38371866941452026,
      "learning_rate": 5.996309907444915e-05,
      "loss": 1.4725,
      "step": 13
    },
    {
      "epoch": 0.019008825526137134,
      "grad_norm": 0.4023575782775879,
      "learning_rate": 5.9934408816563236e-05,
      "loss": 1.5672,
      "step": 14
    },
    {
      "epoch": 0.020366598778004074,
      "grad_norm": 0.4079117774963379,
      "learning_rate": 5.98975347902001e-05,
      "loss": 1.4821,
      "step": 15
    },
    {
      "epoch": 0.02172437202987101,
      "grad_norm": 0.3644375801086426,
      "learning_rate": 5.9852487076350345e-05,
      "loss": 1.4558,
      "step": 16
    },
    {
      "epoch": 0.02308214528173795,
      "grad_norm": 0.3798045516014099,
      "learning_rate": 5.979927799060915e-05,
      "loss": 1.3945,
      "step": 17
    },
    {
      "epoch": 0.024439918533604887,
      "grad_norm": 0.3799417018890381,
      "learning_rate": 5.9737922079809257e-05,
      "loss": 1.5371,
      "step": 18
    },
    {
      "epoch": 0.025797691785471828,
      "grad_norm": 0.45709484815597534,
      "learning_rate": 5.9668436118044054e-05,
      "loss": 1.5058,
      "step": 19
    },
    {
      "epoch": 0.027155465037338764,
      "grad_norm": 0.4341858923435211,
      "learning_rate": 5.959083910208167e-05,
      "loss": 1.4192,
      "step": 20
    },
    {
      "epoch": 0.028513238289205704,
      "grad_norm": 0.37348461151123047,
      "learning_rate": 5.9505152246171474e-05,
      "loss": 1.3679,
      "step": 21
    },
    {
      "epoch": 0.02987101154107264,
      "grad_norm": 0.4123372435569763,
      "learning_rate": 5.941139897624428e-05,
      "loss": 1.4606,
      "step": 22
    },
    {
      "epoch": 0.031228784792939578,
      "grad_norm": 0.41863560676574707,
      "learning_rate": 5.9309604923507984e-05,
      "loss": 1.5268,
      "step": 23
    },
    {
      "epoch": 0.032586558044806514,
      "grad_norm": 0.3617679476737976,
      "learning_rate": 5.9199797917440176e-05,
      "loss": 1.3856,
      "step": 24
    },
    {
      "epoch": 0.03394433129667346,
      "grad_norm": 0.4142565131187439,
      "learning_rate": 5.908200797817991e-05,
      "loss": 1.5046,
      "step": 25
    },
    {
      "epoch": 0.035302104548540394,
      "grad_norm": 0.45500728487968445,
      "learning_rate": 5.895626730832046e-05,
      "loss": 1.5162,
      "step": 26
    },
    {
      "epoch": 0.03665987780040733,
      "grad_norm": 0.3940930664539337,
      "learning_rate": 5.882261028410545e-05,
      "loss": 1.4056,
      "step": 27
    },
    {
      "epoch": 0.03801765105227427,
      "grad_norm": 0.37967368960380554,
      "learning_rate": 5.8681073446030734e-05,
      "loss": 1.4119,
      "step": 28
    },
    {
      "epoch": 0.03937542430414121,
      "grad_norm": 0.4179513454437256,
      "learning_rate": 5.853169548885461e-05,
      "loss": 1.3789,
      "step": 29
    },
    {
      "epoch": 0.04073319755600815,
      "grad_norm": 0.4345241189002991,
      "learning_rate": 5.8374517251019035e-05,
      "loss": 1.3115,
      "step": 30
    },
    {
      "epoch": 0.042090970807875085,
      "grad_norm": 0.44251716136932373,
      "learning_rate": 5.820958170348484e-05,
      "loss": 1.343,
      "step": 31
    },
    {
      "epoch": 0.04344874405974202,
      "grad_norm": 0.44365394115448,
      "learning_rate": 5.8036933937983825e-05,
      "loss": 1.4196,
      "step": 32
    },
    {
      "epoch": 0.04480651731160896,
      "grad_norm": 0.4314393401145935,
      "learning_rate": 5.7856621154691217e-05,
      "loss": 1.3913,
      "step": 33
    },
    {
      "epoch": 0.0461642905634759,
      "grad_norm": 0.4089583456516266,
      "learning_rate": 5.766869264932154e-05,
      "loss": 1.4025,
      "step": 34
    },
    {
      "epoch": 0.04752206381534284,
      "grad_norm": 0.40149539709091187,
      "learning_rate": 5.747319979965172e-05,
      "loss": 1.406,
      "step": 35
    },
    {
      "epoch": 0.048879837067209775,
      "grad_norm": 0.4526827931404114,
      "learning_rate": 5.727019605147488e-05,
      "loss": 1.4015,
      "step": 36
    },
    {
      "epoch": 0.05023761031907671,
      "grad_norm": 0.46217799186706543,
      "learning_rate": 5.7059736903988775e-05,
      "loss": 1.4696,
      "step": 37
    },
    {
      "epoch": 0.051595383570943655,
      "grad_norm": 0.43967458605766296,
      "learning_rate": 5.684187989462291e-05,
      "loss": 1.348,
      "step": 38
    },
    {
      "epoch": 0.05295315682281059,
      "grad_norm": 0.44179558753967285,
      "learning_rate": 5.661668458330836e-05,
      "loss": 1.3125,
      "step": 39
    },
    {
      "epoch": 0.05431093007467753,
      "grad_norm": 0.5129258632659912,
      "learning_rate": 5.638421253619467e-05,
      "loss": 1.3217,
      "step": 40
    },
    {
      "epoch": 0.055668703326544465,
      "grad_norm": 0.5877251625061035,
      "learning_rate": 5.614452730881832e-05,
      "loss": 1.4882,
      "step": 41
    },
    {
      "epoch": 0.05702647657841141,
      "grad_norm": 0.48830196261405945,
      "learning_rate": 5.589769442872722e-05,
      "loss": 1.3058,
      "step": 42
    },
    {
      "epoch": 0.058384249830278345,
      "grad_norm": 0.5610761046409607,
      "learning_rate": 5.5643781377566175e-05,
      "loss": 1.2949,
      "step": 43
    },
    {
      "epoch": 0.05974202308214528,
      "grad_norm": 0.5211129188537598,
      "learning_rate": 5.538285757262806e-05,
      "loss": 1.2745,
      "step": 44
    },
    {
      "epoch": 0.06109979633401222,
      "grad_norm": 0.625769317150116,
      "learning_rate": 5.5114994347875856e-05,
      "loss": 1.3896,
      "step": 45
    },
    {
      "epoch": 0.062457569585879155,
      "grad_norm": 0.607923686504364,
      "learning_rate": 5.48402649344406e-05,
      "loss": 1.2187,
      "step": 46
    },
    {
      "epoch": 0.06381534283774609,
      "grad_norm": 0.8620524406433105,
      "learning_rate": 5.455874444060078e-05,
      "loss": 1.5842,
      "step": 47
    },
    {
      "epoch": 0.06517311608961303,
      "grad_norm": 0.8019170165061951,
      "learning_rate": 5.427050983124843e-05,
      "loss": 1.5473,
      "step": 48
    },
    {
      "epoch": 0.06653088934147998,
      "grad_norm": 1.2089159488677979,
      "learning_rate": 5.397563990684774e-05,
      "loss": 1.5974,
      "step": 49
    },
    {
      "epoch": 0.06788866259334692,
      "grad_norm": 1.9010008573532104,
      "learning_rate": 5.367421528189181e-05,
      "loss": 1.7379,
      "step": 50
    },
    {
      "epoch": 0.06788866259334692,
      "eval_loss": 1.3871045112609863,
      "eval_runtime": 20.6029,
      "eval_samples_per_second": 60.234,
      "eval_steps_per_second": 15.095,
      "step": 50
    },
    {
      "epoch": 0.06924643584521385,
      "grad_norm": 0.6502735018730164,
      "learning_rate": 5.336631836286338e-05,
      "loss": 1.15,
      "step": 51
    },
    {
      "epoch": 0.07060420909708079,
      "grad_norm": 0.7638380527496338,
      "learning_rate": 5.3052033325705774e-05,
      "loss": 1.3236,
      "step": 52
    },
    {
      "epoch": 0.07196198234894773,
      "grad_norm": 0.8170233964920044,
      "learning_rate": 5.2731446092810044e-05,
      "loss": 1.4022,
      "step": 53
    },
    {
      "epoch": 0.07331975560081466,
      "grad_norm": 0.8679696917533875,
      "learning_rate": 5.240464430952462e-05,
      "loss": 1.4187,
      "step": 54
    },
    {
      "epoch": 0.0746775288526816,
      "grad_norm": 0.7302316427230835,
      "learning_rate": 5.207171732019395e-05,
      "loss": 1.4995,
      "step": 55
    },
    {
      "epoch": 0.07603530210454854,
      "grad_norm": 0.6195611953735352,
      "learning_rate": 5.1732756143732675e-05,
      "loss": 1.386,
      "step": 56
    },
    {
      "epoch": 0.07739307535641547,
      "grad_norm": 0.6950957775115967,
      "learning_rate": 5.1387853448741916e-05,
      "loss": 1.3669,
      "step": 57
    },
    {
      "epoch": 0.07875084860828242,
      "grad_norm": 0.6076563000679016,
      "learning_rate": 5.103710352817465e-05,
      "loss": 1.3245,
      "step": 58
    },
    {
      "epoch": 0.08010862186014936,
      "grad_norm": 0.5559670329093933,
      "learning_rate": 5.068060227355698e-05,
      "loss": 1.3563,
      "step": 59
    },
    {
      "epoch": 0.0814663951120163,
      "grad_norm": 0.43826499581336975,
      "learning_rate": 5.0318447148772234e-05,
      "loss": 1.2848,
      "step": 60
    },
    {
      "epoch": 0.08282416836388323,
      "grad_norm": 0.4111815094947815,
      "learning_rate": 4.995073716341545e-05,
      "loss": 1.2948,
      "step": 61
    },
    {
      "epoch": 0.08418194161575017,
      "grad_norm": 0.38872504234313965,
      "learning_rate": 4.957757284572506e-05,
      "loss": 1.2717,
      "step": 62
    },
    {
      "epoch": 0.0855397148676171,
      "grad_norm": 0.3297795057296753,
      "learning_rate": 4.91990562150995e-05,
      "loss": 1.2787,
      "step": 63
    },
    {
      "epoch": 0.08689748811948404,
      "grad_norm": 0.3442164659500122,
      "learning_rate": 4.881529075420611e-05,
      "loss": 1.2691,
      "step": 64
    },
    {
      "epoch": 0.08825526137135098,
      "grad_norm": 0.337697833776474,
      "learning_rate": 4.8426381380690036e-05,
      "loss": 1.2559,
      "step": 65
    },
    {
      "epoch": 0.08961303462321792,
      "grad_norm": 0.377407968044281,
      "learning_rate": 4.8032434418490753e-05,
      "loss": 1.1306,
      "step": 66
    },
    {
      "epoch": 0.09097080787508487,
      "grad_norm": 0.37138831615448,
      "learning_rate": 4.7633557568774194e-05,
      "loss": 1.3517,
      "step": 67
    },
    {
      "epoch": 0.0923285811269518,
      "grad_norm": 0.3734515309333801,
      "learning_rate": 4.722985988048831e-05,
      "loss": 1.241,
      "step": 68
    },
    {
      "epoch": 0.09368635437881874,
      "grad_norm": 0.40681424736976624,
      "learning_rate": 4.6821451720550184e-05,
      "loss": 1.3452,
      "step": 69
    },
    {
      "epoch": 0.09504412763068568,
      "grad_norm": 0.3799170255661011,
      "learning_rate": 4.640844474367282e-05,
      "loss": 1.2281,
      "step": 70
    },
    {
      "epoch": 0.09640190088255261,
      "grad_norm": 0.40699976682662964,
      "learning_rate": 4.5990951861839815e-05,
      "loss": 1.2165,
      "step": 71
    },
    {
      "epoch": 0.09775967413441955,
      "grad_norm": 0.4101755917072296,
      "learning_rate": 4.5569087213436455e-05,
      "loss": 1.2346,
      "step": 72
    },
    {
      "epoch": 0.09911744738628649,
      "grad_norm": 0.41790157556533813,
      "learning_rate": 4.514296613204532e-05,
      "loss": 1.3324,
      "step": 73
    },
    {
      "epoch": 0.10047522063815342,
      "grad_norm": 0.4398515224456787,
      "learning_rate": 4.471270511491525e-05,
      "loss": 1.2324,
      "step": 74
    },
    {
      "epoch": 0.10183299389002037,
      "grad_norm": 0.41198304295539856,
      "learning_rate": 4.427842179111221e-05,
      "loss": 1.1749,
      "step": 75
    },
    {
      "epoch": 0.10319076714188731,
      "grad_norm": 0.40906915068626404,
      "learning_rate": 4.3840234889360634e-05,
      "loss": 1.21,
      "step": 76
    },
    {
      "epoch": 0.10454854039375425,
      "grad_norm": 0.38414862751960754,
      "learning_rate": 4.33982642055842e-05,
      "loss": 1.2519,
      "step": 77
    },
    {
      "epoch": 0.10590631364562118,
      "grad_norm": 0.3740120828151703,
      "learning_rate": 4.2952630570154785e-05,
      "loss": 1.2096,
      "step": 78
    },
    {
      "epoch": 0.10726408689748812,
      "grad_norm": 0.3862381875514984,
      "learning_rate": 4.250345581485871e-05,
      "loss": 1.2626,
      "step": 79
    },
    {
      "epoch": 0.10862186014935506,
      "grad_norm": 0.3637017011642456,
      "learning_rate": 4.205086273958909e-05,
      "loss": 1.2496,
      "step": 80
    },
    {
      "epoch": 0.109979633401222,
      "grad_norm": 0.38009828329086304,
      "learning_rate": 4.1594975078773565e-05,
      "loss": 1.2312,
      "step": 81
    },
    {
      "epoch": 0.11133740665308893,
      "grad_norm": 0.353546142578125,
      "learning_rate": 4.113591746754662e-05,
      "loss": 1.2196,
      "step": 82
    },
    {
      "epoch": 0.11269517990495587,
      "grad_norm": 0.4014517068862915,
      "learning_rate": 4.06738154076755e-05,
      "loss": 1.2508,
      "step": 83
    },
    {
      "epoch": 0.11405295315682282,
      "grad_norm": 0.4051145315170288,
      "learning_rate": 4.020879523324929e-05,
      "loss": 1.2438,
      "step": 84
    },
    {
      "epoch": 0.11541072640868975,
      "grad_norm": 0.39561352133750916,
      "learning_rate": 3.974098407614051e-05,
      "loss": 1.2708,
      "step": 85
    },
    {
      "epoch": 0.11676849966055669,
      "grad_norm": 0.4071887135505676,
      "learning_rate": 3.927050983124842e-05,
      "loss": 1.2017,
      "step": 86
    },
    {
      "epoch": 0.11812627291242363,
      "grad_norm": 0.4768187403678894,
      "learning_rate": 3.8797501121533946e-05,
      "loss": 1.2136,
      "step": 87
    },
    {
      "epoch": 0.11948404616429056,
      "grad_norm": 0.4878613352775574,
      "learning_rate": 3.832208726285534e-05,
      "loss": 1.1696,
      "step": 88
    },
    {
      "epoch": 0.1208418194161575,
      "grad_norm": 0.4963112473487854,
      "learning_rate": 3.784439822861459e-05,
      "loss": 1.1474,
      "step": 89
    },
    {
      "epoch": 0.12219959266802444,
      "grad_norm": 0.5566866993904114,
      "learning_rate": 3.7364564614223976e-05,
      "loss": 1.1513,
      "step": 90
    },
    {
      "epoch": 0.12355736591989137,
      "grad_norm": 0.49696850776672363,
      "learning_rate": 3.688271760140255e-05,
      "loss": 1.0694,
      "step": 91
    },
    {
      "epoch": 0.12491513917175831,
      "grad_norm": 0.4537845551967621,
      "learning_rate": 3.6398988922312406e-05,
      "loss": 1.2268,
      "step": 92
    },
    {
      "epoch": 0.12627291242362526,
      "grad_norm": 0.5615867376327515,
      "learning_rate": 3.591351082354441e-05,
      "loss": 1.209,
      "step": 93
    },
    {
      "epoch": 0.12763068567549218,
      "grad_norm": 0.5191481113433838,
      "learning_rate": 3.54264160299633e-05,
      "loss": 0.9851,
      "step": 94
    },
    {
      "epoch": 0.12898845892735913,
      "grad_norm": 0.5748217701911926,
      "learning_rate": 3.493783770842202e-05,
      "loss": 1.1675,
      "step": 95
    },
    {
      "epoch": 0.13034623217922606,
      "grad_norm": 0.5830937623977661,
      "learning_rate": 3.444790943135526e-05,
      "loss": 1.1482,
      "step": 96
    },
    {
      "epoch": 0.131704005431093,
      "grad_norm": 0.7683277130126953,
      "learning_rate": 3.3956765140262074e-05,
      "loss": 1.1567,
      "step": 97
    },
    {
      "epoch": 0.13306177868295996,
      "grad_norm": 0.9153833985328674,
      "learning_rate": 3.346453910908759e-05,
      "loss": 1.0613,
      "step": 98
    },
    {
      "epoch": 0.13441955193482688,
      "grad_norm": 1.457146167755127,
      "learning_rate": 3.297136590751389e-05,
      "loss": 1.4215,
      "step": 99
    },
    {
      "epoch": 0.13577732518669383,
      "grad_norm": 2.4654388427734375,
      "learning_rate": 3.247738036416998e-05,
      "loss": 1.3325,
      "step": 100
    },
    {
      "epoch": 0.13577732518669383,
      "eval_loss": 1.2617239952087402,
      "eval_runtime": 20.5894,
      "eval_samples_per_second": 60.274,
      "eval_steps_per_second": 15.105,
      "step": 100
    },
    {
      "epoch": 0.13713509843856075,
      "grad_norm": 0.3299471437931061,
      "learning_rate": 3.1982717529770985e-05,
      "loss": 1.043,
      "step": 101
    },
    {
      "epoch": 0.1384928716904277,
      "grad_norm": 0.5103358626365662,
      "learning_rate": 3.148751264019667e-05,
      "loss": 1.0821,
      "step": 102
    },
    {
      "epoch": 0.13985064494229463,
      "grad_norm": 0.38934123516082764,
      "learning_rate": 3.099190107951924e-05,
      "loss": 1.1851,
      "step": 103
    },
    {
      "epoch": 0.14120841819416158,
      "grad_norm": 0.4618616998195648,
      "learning_rate": 3.049601834299076e-05,
      "loss": 1.1477,
      "step": 104
    },
    {
      "epoch": 0.1425661914460285,
      "grad_norm": 0.4667770564556122,
      "learning_rate": 3e-05,
      "loss": 1.3379,
      "step": 105
    },
    {
      "epoch": 0.14392396469789545,
      "grad_norm": 0.5612397789955139,
      "learning_rate": 2.9503981657009246e-05,
      "loss": 1.2474,
      "step": 106
    },
    {
      "epoch": 0.1452817379497624,
      "grad_norm": 0.4959563612937927,
      "learning_rate": 2.9008098920480752e-05,
      "loss": 1.2996,
      "step": 107
    },
    {
      "epoch": 0.14663951120162932,
      "grad_norm": 0.46022874116897583,
      "learning_rate": 2.851248735980333e-05,
      "loss": 1.2365,
      "step": 108
    },
    {
      "epoch": 0.14799728445349628,
      "grad_norm": 0.4637479782104492,
      "learning_rate": 2.801728247022902e-05,
      "loss": 1.1274,
      "step": 109
    },
    {
      "epoch": 0.1493550577053632,
      "grad_norm": 0.46539270877838135,
      "learning_rate": 2.7522619635830034e-05,
      "loss": 1.2835,
      "step": 110
    },
    {
      "epoch": 0.15071283095723015,
      "grad_norm": 0.44558578729629517,
      "learning_rate": 2.702863409248612e-05,
      "loss": 1.1671,
      "step": 111
    },
    {
      "epoch": 0.15207060420909707,
      "grad_norm": 0.4058595299720764,
      "learning_rate": 2.6535460890912416e-05,
      "loss": 1.3161,
      "step": 112
    },
    {
      "epoch": 0.15342837746096402,
      "grad_norm": 0.41494250297546387,
      "learning_rate": 2.604323485973793e-05,
      "loss": 1.2028,
      "step": 113
    },
    {
      "epoch": 0.15478615071283094,
      "grad_norm": 0.3981316387653351,
      "learning_rate": 2.555209056864474e-05,
      "loss": 1.3,
      "step": 114
    },
    {
      "epoch": 0.1561439239646979,
      "grad_norm": 0.3824096620082855,
      "learning_rate": 2.5062162291577978e-05,
      "loss": 1.3645,
      "step": 115
    },
    {
      "epoch": 0.15750169721656485,
      "grad_norm": 0.37386175990104675,
      "learning_rate": 2.4573583970036712e-05,
      "loss": 1.1997,
      "step": 116
    },
    {
      "epoch": 0.15885947046843177,
      "grad_norm": 0.3558681607246399,
      "learning_rate": 2.4086489176455595e-05,
      "loss": 1.1774,
      "step": 117
    },
    {
      "epoch": 0.16021724372029872,
      "grad_norm": 0.3079434037208557,
      "learning_rate": 2.36010110776876e-05,
      "loss": 1.1493,
      "step": 118
    },
    {
      "epoch": 0.16157501697216564,
      "grad_norm": 0.3555084764957428,
      "learning_rate": 2.3117282398597456e-05,
      "loss": 1.2321,
      "step": 119
    },
    {
      "epoch": 0.1629327902240326,
      "grad_norm": 0.3722546398639679,
      "learning_rate": 2.263543538577603e-05,
      "loss": 1.2395,
      "step": 120
    },
    {
      "epoch": 0.16429056347589951,
      "grad_norm": 0.34470927715301514,
      "learning_rate": 2.215560177138541e-05,
      "loss": 1.3016,
      "step": 121
    },
    {
      "epoch": 0.16564833672776647,
      "grad_norm": 0.36612775921821594,
      "learning_rate": 2.167791273714467e-05,
      "loss": 1.2412,
      "step": 122
    },
    {
      "epoch": 0.1670061099796334,
      "grad_norm": 0.4002249538898468,
      "learning_rate": 2.1202498878466062e-05,
      "loss": 1.2357,
      "step": 123
    },
    {
      "epoch": 0.16836388323150034,
      "grad_norm": 0.34691083431243896,
      "learning_rate": 2.072949016875158e-05,
      "loss": 1.2007,
      "step": 124
    },
    {
      "epoch": 0.1697216564833673,
      "grad_norm": 0.36753562092781067,
      "learning_rate": 2.0259015923859498e-05,
      "loss": 1.2795,
      "step": 125
    },
    {
      "epoch": 0.1710794297352342,
      "grad_norm": 0.35715538263320923,
      "learning_rate": 1.979120476675071e-05,
      "loss": 1.2341,
      "step": 126
    },
    {
      "epoch": 0.17243720298710116,
      "grad_norm": 0.3618859350681305,
      "learning_rate": 1.9326184592324503e-05,
      "loss": 1.3268,
      "step": 127
    },
    {
      "epoch": 0.17379497623896809,
      "grad_norm": 0.3898789584636688,
      "learning_rate": 1.8864082532453373e-05,
      "loss": 1.2401,
      "step": 128
    },
    {
      "epoch": 0.17515274949083504,
      "grad_norm": 0.38298866152763367,
      "learning_rate": 1.840502492122644e-05,
      "loss": 1.1924,
      "step": 129
    },
    {
      "epoch": 0.17651052274270196,
      "grad_norm": 0.41282638907432556,
      "learning_rate": 1.7949137260410924e-05,
      "loss": 1.299,
      "step": 130
    },
    {
      "epoch": 0.1778682959945689,
      "grad_norm": 0.40525734424591064,
      "learning_rate": 1.7496544185141295e-05,
      "loss": 1.1921,
      "step": 131
    },
    {
      "epoch": 0.17922606924643583,
      "grad_norm": 0.3740069568157196,
      "learning_rate": 1.7047369429845216e-05,
      "loss": 1.1422,
      "step": 132
    },
    {
      "epoch": 0.18058384249830278,
      "grad_norm": 0.4477040469646454,
      "learning_rate": 1.6601735794415806e-05,
      "loss": 1.273,
      "step": 133
    },
    {
      "epoch": 0.18194161575016973,
      "grad_norm": 0.4326741695404053,
      "learning_rate": 1.615976511063937e-05,
      "loss": 1.1538,
      "step": 134
    },
    {
      "epoch": 0.18329938900203666,
      "grad_norm": 0.40829864144325256,
      "learning_rate": 1.5721578208887793e-05,
      "loss": 1.1364,
      "step": 135
    },
    {
      "epoch": 0.1846571622539036,
      "grad_norm": 0.48507779836654663,
      "learning_rate": 1.5287294885084766e-05,
      "loss": 1.1943,
      "step": 136
    },
    {
      "epoch": 0.18601493550577053,
      "grad_norm": 0.4629991054534912,
      "learning_rate": 1.4857033867954697e-05,
      "loss": 1.3347,
      "step": 137
    },
    {
      "epoch": 0.18737270875763748,
      "grad_norm": 0.448877215385437,
      "learning_rate": 1.4430912786563554e-05,
      "loss": 1.2268,
      "step": 138
    },
    {
      "epoch": 0.1887304820095044,
      "grad_norm": 0.4393776059150696,
      "learning_rate": 1.4009048138160195e-05,
      "loss": 1.1314,
      "step": 139
    },
    {
      "epoch": 0.19008825526137135,
      "grad_norm": 0.4872407913208008,
      "learning_rate": 1.3591555256327199e-05,
      "loss": 1.1743,
      "step": 140
    },
    {
      "epoch": 0.19144602851323828,
      "grad_norm": 0.5232310891151428,
      "learning_rate": 1.3178548279449822e-05,
      "loss": 1.1972,
      "step": 141
    },
    {
      "epoch": 0.19280380176510523,
      "grad_norm": 0.5275864005088806,
      "learning_rate": 1.2770140119511693e-05,
      "loss": 1.2252,
      "step": 142
    },
    {
      "epoch": 0.19416157501697218,
      "grad_norm": 0.6005786657333374,
      "learning_rate": 1.2366442431225809e-05,
      "loss": 1.2634,
      "step": 143
    },
    {
      "epoch": 0.1955193482688391,
      "grad_norm": 0.6898880004882812,
      "learning_rate": 1.1967565581509248e-05,
      "loss": 1.2417,
      "step": 144
    },
    {
      "epoch": 0.19687712152070605,
      "grad_norm": 0.711155116558075,
      "learning_rate": 1.1573618619309965e-05,
      "loss": 1.1956,
      "step": 145
    },
    {
      "epoch": 0.19823489477257297,
      "grad_norm": 0.6882777810096741,
      "learning_rate": 1.1184709245793889e-05,
      "loss": 1.0373,
      "step": 146
    },
    {
      "epoch": 0.19959266802443992,
      "grad_norm": 0.9063619375228882,
      "learning_rate": 1.0800943784900502e-05,
      "loss": 1.1443,
      "step": 147
    },
    {
      "epoch": 0.20095044127630685,
      "grad_norm": 1.1058382987976074,
      "learning_rate": 1.042242715427494e-05,
      "loss": 1.2277,
      "step": 148
    },
    {
      "epoch": 0.2023082145281738,
      "grad_norm": 1.3979732990264893,
      "learning_rate": 1.004926283658455e-05,
      "loss": 1.072,
      "step": 149
    },
    {
      "epoch": 0.20366598778004075,
      "grad_norm": 2.6124637126922607,
      "learning_rate": 9.681552851227774e-06,
      "loss": 1.6226,
      "step": 150
    },
    {
      "epoch": 0.20366598778004075,
      "eval_loss": 1.230950117111206,
      "eval_runtime": 20.5956,
      "eval_samples_per_second": 60.256,
      "eval_steps_per_second": 15.1,
      "step": 150
    },
    {
      "epoch": 0.20502376103190767,
      "grad_norm": 0.30644547939300537,
      "learning_rate": 9.319397726443026e-06,
      "loss": 1.0185,
      "step": 151
    },
    {
      "epoch": 0.20638153428377462,
      "grad_norm": 0.3141520023345947,
      "learning_rate": 8.962896471825342e-06,
      "loss": 1.205,
      "step": 152
    },
    {
      "epoch": 0.20773930753564154,
      "grad_norm": 0.3089138865470886,
      "learning_rate": 8.61214655125809e-06,
      "loss": 1.2672,
      "step": 153
    },
    {
      "epoch": 0.2090970807875085,
      "grad_norm": 0.3366032838821411,
      "learning_rate": 8.267243856267331e-06,
      "loss": 1.2912,
      "step": 154
    },
    {
      "epoch": 0.21045485403937542,
      "grad_norm": 0.3448525071144104,
      "learning_rate": 7.928282679806052e-06,
      "loss": 1.162,
      "step": 155
    },
    {
      "epoch": 0.21181262729124237,
      "grad_norm": 0.33545318245887756,
      "learning_rate": 7.595355690475393e-06,
      "loss": 1.1831,
      "step": 156
    },
    {
      "epoch": 0.2131704005431093,
      "grad_norm": 0.37803956866264343,
      "learning_rate": 7.268553907189964e-06,
      "loss": 1.2012,
      "step": 157
    },
    {
      "epoch": 0.21452817379497624,
      "grad_norm": 0.387054979801178,
      "learning_rate": 6.947966674294236e-06,
      "loss": 1.1716,
      "step": 158
    },
    {
      "epoch": 0.2158859470468432,
      "grad_norm": 0.3598926365375519,
      "learning_rate": 6.6336816371366305e-06,
      "loss": 1.3053,
      "step": 159
    },
    {
      "epoch": 0.2172437202987101,
      "grad_norm": 0.3465365171432495,
      "learning_rate": 6.325784718108196e-06,
      "loss": 1.2507,
      "step": 160
    },
    {
      "epoch": 0.21860149355057706,
      "grad_norm": 0.34380465745925903,
      "learning_rate": 6.0243600931522595e-06,
      "loss": 1.1837,
      "step": 161
    },
    {
      "epoch": 0.219959266802444,
      "grad_norm": 0.38984525203704834,
      "learning_rate": 5.72949016875158e-06,
      "loss": 1.1786,
      "step": 162
    },
    {
      "epoch": 0.22131704005431094,
      "grad_norm": 0.32796481251716614,
      "learning_rate": 5.44125555939923e-06,
      "loss": 1.1675,
      "step": 163
    },
    {
      "epoch": 0.22267481330617786,
      "grad_norm": 0.3847575783729553,
      "learning_rate": 5.159735065559399e-06,
      "loss": 1.1848,
      "step": 164
    },
    {
      "epoch": 0.2240325865580448,
      "grad_norm": 0.3389010429382324,
      "learning_rate": 4.885005652124144e-06,
      "loss": 1.2208,
      "step": 165
    },
    {
      "epoch": 0.22539035980991173,
      "grad_norm": 0.3421236574649811,
      "learning_rate": 4.617142427371934e-06,
      "loss": 1.2395,
      "step": 166
    },
    {
      "epoch": 0.22674813306177868,
      "grad_norm": 0.40454038977622986,
      "learning_rate": 4.3562186224338265e-06,
      "loss": 1.1845,
      "step": 167
    },
    {
      "epoch": 0.22810590631364563,
      "grad_norm": 0.38095664978027344,
      "learning_rate": 4.102305571272783e-06,
      "loss": 1.2256,
      "step": 168
    },
    {
      "epoch": 0.22946367956551256,
      "grad_norm": 0.39606228470802307,
      "learning_rate": 3.855472691181678e-06,
      "loss": 1.2515,
      "step": 169
    },
    {
      "epoch": 0.2308214528173795,
      "grad_norm": 0.40278103947639465,
      "learning_rate": 3.615787463805331e-06,
      "loss": 1.1997,
      "step": 170
    },
    {
      "epoch": 0.23217922606924643,
      "grad_norm": 0.38342180848121643,
      "learning_rate": 3.383315416691646e-06,
      "loss": 1.2968,
      "step": 171
    },
    {
      "epoch": 0.23353699932111338,
      "grad_norm": 0.3293931484222412,
      "learning_rate": 3.158120105377096e-06,
      "loss": 1.105,
      "step": 172
    },
    {
      "epoch": 0.2348947725729803,
      "grad_norm": 0.3917769491672516,
      "learning_rate": 2.940263096011233e-06,
      "loss": 1.2725,
      "step": 173
    },
    {
      "epoch": 0.23625254582484725,
      "grad_norm": 0.40385058522224426,
      "learning_rate": 2.729803948525125e-06,
      "loss": 1.268,
      "step": 174
    },
    {
      "epoch": 0.23761031907671418,
      "grad_norm": 0.3761354088783264,
      "learning_rate": 2.526800200348275e-06,
      "loss": 1.2043,
      "step": 175
    },
    {
      "epoch": 0.23896809232858113,
      "grad_norm": 0.35785752534866333,
      "learning_rate": 2.3313073506784575e-06,
      "loss": 1.2083,
      "step": 176
    },
    {
      "epoch": 0.24032586558044808,
      "grad_norm": 0.3641122281551361,
      "learning_rate": 2.143378845308791e-06,
      "loss": 1.1328,
      "step": 177
    },
    {
      "epoch": 0.241683638832315,
      "grad_norm": 0.3801313042640686,
      "learning_rate": 1.9630660620161777e-06,
      "loss": 1.196,
      "step": 178
    },
    {
      "epoch": 0.24304141208418195,
      "grad_norm": 0.3835360109806061,
      "learning_rate": 1.790418296515165e-06,
      "loss": 1.1675,
      "step": 179
    },
    {
      "epoch": 0.24439918533604887,
      "grad_norm": 0.39598432183265686,
      "learning_rate": 1.625482748980961e-06,
      "loss": 1.1781,
      "step": 180
    },
    {
      "epoch": 0.24575695858791582,
      "grad_norm": 0.43931299448013306,
      "learning_rate": 1.4683045111453942e-06,
      "loss": 1.3284,
      "step": 181
    },
    {
      "epoch": 0.24711473183978275,
      "grad_norm": 0.38031482696533203,
      "learning_rate": 1.3189265539692707e-06,
      "loss": 1.1546,
      "step": 182
    },
    {
      "epoch": 0.2484725050916497,
      "grad_norm": 0.43264663219451904,
      "learning_rate": 1.1773897158945557e-06,
      "loss": 1.1991,
      "step": 183
    },
    {
      "epoch": 0.24983027834351662,
      "grad_norm": 0.39598092436790466,
      "learning_rate": 1.0437326916795432e-06,
      "loss": 1.217,
      "step": 184
    },
    {
      "epoch": 0.25118805159538354,
      "grad_norm": 0.434064120054245,
      "learning_rate": 9.179920218200888e-07,
      "loss": 1.1195,
      "step": 185
    },
    {
      "epoch": 0.2525458248472505,
      "grad_norm": 0.4763726592063904,
      "learning_rate": 8.002020825598277e-07,
      "loss": 1.2145,
      "step": 186
    },
    {
      "epoch": 0.25390359809911744,
      "grad_norm": 0.45298972725868225,
      "learning_rate": 6.90395076492022e-07,
      "loss": 1.2858,
      "step": 187
    },
    {
      "epoch": 0.25526137135098437,
      "grad_norm": 0.5360013842582703,
      "learning_rate": 5.886010237557194e-07,
      "loss": 1.1961,
      "step": 188
    },
    {
      "epoch": 0.25661914460285135,
      "grad_norm": 0.44087594747543335,
      "learning_rate": 4.94847753828529e-07,
      "loss": 1.0941,
      "step": 189
    },
    {
      "epoch": 0.25797691785471827,
      "grad_norm": 0.5056028962135315,
      "learning_rate": 4.091608979183303e-07,
      "loss": 1.1411,
      "step": 190
    },
    {
      "epoch": 0.2593346911065852,
      "grad_norm": 0.5390208959579468,
      "learning_rate": 3.315638819559452e-07,
      "loss": 1.0857,
      "step": 191
    },
    {
      "epoch": 0.2606924643584521,
      "grad_norm": 0.5694705843925476,
      "learning_rate": 2.6207792019074414e-07,
      "loss": 1.1992,
      "step": 192
    },
    {
      "epoch": 0.2620502376103191,
      "grad_norm": 0.5326172709465027,
      "learning_rate": 2.0072200939085573e-07,
      "loss": 1.2654,
      "step": 193
    },
    {
      "epoch": 0.263408010862186,
      "grad_norm": 0.6019105315208435,
      "learning_rate": 1.475129236496575e-07,
      "loss": 1.2256,
      "step": 194
    },
    {
      "epoch": 0.26476578411405294,
      "grad_norm": 0.6420196890830994,
      "learning_rate": 1.0246520979990459e-07,
      "loss": 1.1443,
      "step": 195
    },
    {
      "epoch": 0.2661235573659199,
      "grad_norm": 0.6914029121398926,
      "learning_rate": 6.559118343676396e-08,
      "loss": 1.1849,
      "step": 196
    },
    {
      "epoch": 0.26748133061778684,
      "grad_norm": 0.8857573866844177,
      "learning_rate": 3.690092555085789e-08,
      "loss": 1.239,
      "step": 197
    },
    {
      "epoch": 0.26883910386965376,
      "grad_norm": 1.1338528394699097,
      "learning_rate": 1.640227977221853e-08,
      "loss": 1.3823,
      "step": 198
    },
    {
      "epoch": 0.2701968771215207,
      "grad_norm": 1.0841175317764282,
      "learning_rate": 4.1008502259298755e-09,
      "loss": 1.2757,
      "step": 199
    },
    {
      "epoch": 0.27155465037338766,
      "grad_norm": 2.116804599761963,
      "learning_rate": 0.0,
      "loss": 1.257,
      "step": 200
    },
    {
      "epoch": 0.27155465037338766,
      "eval_loss": 1.2237151861190796,
      "eval_runtime": 20.6081,
      "eval_samples_per_second": 60.219,
      "eval_steps_per_second": 15.091,
      "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": 4,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.375493127929856e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}