{
  "best_metric": 10.34636402130127,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 3.0093457943925235,
  "eval_steps": 50,
  "global_step": 161,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.018691588785046728,
      "grad_norm": 0.058510370552539825,
      "learning_rate": 1e-05,
      "loss": 10.3789,
      "step": 1
    },
    {
      "epoch": 0.018691588785046728,
      "eval_loss": 10.37936019897461,
      "eval_runtime": 0.3544,
      "eval_samples_per_second": 253.969,
      "eval_steps_per_second": 64.903,
      "step": 1
    },
    {
      "epoch": 0.037383177570093455,
      "grad_norm": 0.049845196306705475,
      "learning_rate": 2e-05,
      "loss": 10.3753,
      "step": 2
    },
    {
      "epoch": 0.056074766355140186,
      "grad_norm": 0.052543047815561295,
      "learning_rate": 3e-05,
      "loss": 10.3771,
      "step": 3
    },
    {
      "epoch": 0.07476635514018691,
      "grad_norm": 0.05512682721018791,
      "learning_rate": 4e-05,
      "loss": 10.3784,
      "step": 4
    },
    {
      "epoch": 0.09345794392523364,
      "grad_norm": 0.05461808294057846,
      "learning_rate": 5e-05,
      "loss": 10.3761,
      "step": 5
    },
    {
      "epoch": 0.11214953271028037,
      "grad_norm": 0.05171424150466919,
      "learning_rate": 6e-05,
      "loss": 10.3787,
      "step": 6
    },
    {
      "epoch": 0.1308411214953271,
      "grad_norm": 0.06137857586145401,
      "learning_rate": 7e-05,
      "loss": 10.378,
      "step": 7
    },
    {
      "epoch": 0.14953271028037382,
      "grad_norm": 0.04720870777964592,
      "learning_rate": 8e-05,
      "loss": 10.3785,
      "step": 8
    },
    {
      "epoch": 0.16822429906542055,
      "grad_norm": 0.04900234937667847,
      "learning_rate": 9e-05,
      "loss": 10.3773,
      "step": 9
    },
    {
      "epoch": 0.18691588785046728,
      "grad_norm": 0.05933133885264397,
      "learning_rate": 0.0001,
      "loss": 10.3797,
      "step": 10
    },
    {
      "epoch": 0.205607476635514,
      "grad_norm": 0.060151103883981705,
      "learning_rate": 9.998917893031616e-05,
      "loss": 10.3794,
      "step": 11
    },
    {
      "epoch": 0.22429906542056074,
      "grad_norm": 0.07007813453674316,
      "learning_rate": 9.995672040508655e-05,
      "loss": 10.3706,
      "step": 12
    },
    {
      "epoch": 0.24299065420560748,
      "grad_norm": 0.056078240275382996,
      "learning_rate": 9.990263847374976e-05,
      "loss": 10.3754,
      "step": 13
    },
    {
      "epoch": 0.2616822429906542,
      "grad_norm": 0.06963472813367844,
      "learning_rate": 9.982695654527965e-05,
      "loss": 10.3766,
      "step": 14
    },
    {
      "epoch": 0.2803738317757009,
      "grad_norm": 0.05379687622189522,
      "learning_rate": 9.972970737805311e-05,
      "loss": 10.3752,
      "step": 15
    },
    {
      "epoch": 0.29906542056074764,
      "grad_norm": 0.06033436208963394,
      "learning_rate": 9.961093306567075e-05,
      "loss": 10.3771,
      "step": 16
    },
    {
      "epoch": 0.3177570093457944,
      "grad_norm": 0.07101690769195557,
      "learning_rate": 9.947068501873701e-05,
      "loss": 10.3733,
      "step": 17
    },
    {
      "epoch": 0.3364485981308411,
      "grad_norm": 0.07208770513534546,
      "learning_rate": 9.930902394260747e-05,
      "loss": 10.3727,
      "step": 18
    },
    {
      "epoch": 0.35514018691588783,
      "grad_norm": 0.07476979494094849,
      "learning_rate": 9.912601981111286e-05,
      "loss": 10.3695,
      "step": 19
    },
    {
      "epoch": 0.37383177570093457,
      "grad_norm": 0.07623960822820663,
      "learning_rate": 9.892175183627161e-05,
      "loss": 10.3686,
      "step": 20
    },
    {
      "epoch": 0.3925233644859813,
      "grad_norm": 0.08990193158388138,
      "learning_rate": 9.86963084340033e-05,
      "loss": 10.3769,
      "step": 21
    },
    {
      "epoch": 0.411214953271028,
      "grad_norm": 0.09957633912563324,
      "learning_rate": 9.844978718585855e-05,
      "loss": 10.3772,
      "step": 22
    },
    {
      "epoch": 0.42990654205607476,
      "grad_norm": 0.10019394755363464,
      "learning_rate": 9.818229479678158e-05,
      "loss": 10.3736,
      "step": 23
    },
    {
      "epoch": 0.4485981308411215,
      "grad_norm": 0.12851685285568237,
      "learning_rate": 9.789394704892364e-05,
      "loss": 10.3687,
      "step": 24
    },
    {
      "epoch": 0.4672897196261682,
      "grad_norm": 0.132917121052742,
      "learning_rate": 9.758486875152766e-05,
      "loss": 10.3709,
      "step": 25
    },
    {
      "epoch": 0.48598130841121495,
      "grad_norm": 0.10603531450033188,
      "learning_rate": 9.725519368690538e-05,
      "loss": 10.3686,
      "step": 26
    },
    {
      "epoch": 0.5046728971962616,
      "grad_norm": 0.1067882776260376,
      "learning_rate": 9.690506455253073e-05,
      "loss": 10.3699,
      "step": 27
    },
    {
      "epoch": 0.5233644859813084,
      "grad_norm": 0.11756634712219238,
      "learning_rate": 9.653463289927411e-05,
      "loss": 10.3714,
      "step": 28
    },
    {
      "epoch": 0.5420560747663551,
      "grad_norm": 0.130135178565979,
      "learning_rate": 9.614405906580485e-05,
      "loss": 10.3681,
      "step": 29
    },
    {
      "epoch": 0.5607476635514018,
      "grad_norm": 0.1225796639919281,
      "learning_rate": 9.573351210918974e-05,
      "loss": 10.3688,
      "step": 30
    },
    {
      "epoch": 0.5794392523364486,
      "grad_norm": 0.14659637212753296,
      "learning_rate": 9.53031697317178e-05,
      "loss": 10.3665,
      "step": 31
    },
    {
      "epoch": 0.5981308411214953,
      "grad_norm": 0.122837133705616,
      "learning_rate": 9.485321820398321e-05,
      "loss": 10.3638,
      "step": 32
    },
    {
      "epoch": 0.616822429906542,
      "grad_norm": 0.14036515355110168,
      "learning_rate": 9.438385228425938e-05,
      "loss": 10.3628,
      "step": 33
    },
    {
      "epoch": 0.6355140186915887,
      "grad_norm": 0.1749039888381958,
      "learning_rate": 9.389527513419934e-05,
      "loss": 10.3625,
      "step": 34
    },
    {
      "epoch": 0.6542056074766355,
      "grad_norm": 0.19820836186408997,
      "learning_rate": 9.338769823089853e-05,
      "loss": 10.3652,
      "step": 35
    },
    {
      "epoch": 0.6728971962616822,
      "grad_norm": 0.1706521362066269,
      "learning_rate": 9.286134127535859e-05,
      "loss": 10.3601,
      "step": 36
    },
    {
      "epoch": 0.6915887850467289,
      "grad_norm": 0.17861370742321014,
      "learning_rate": 9.231643209739128e-05,
      "loss": 10.3603,
      "step": 37
    },
    {
      "epoch": 0.7102803738317757,
      "grad_norm": 0.15735653042793274,
      "learning_rate": 9.175320655700406e-05,
      "loss": 10.3561,
      "step": 38
    },
    {
      "epoch": 0.7289719626168224,
      "grad_norm": 0.14328105747699738,
      "learning_rate": 9.117190844230971e-05,
      "loss": 10.3589,
      "step": 39
    },
    {
      "epoch": 0.7476635514018691,
      "grad_norm": 0.16631971299648285,
      "learning_rate": 9.057278936400453e-05,
      "loss": 10.3563,
      "step": 40
    },
    {
      "epoch": 0.7663551401869159,
      "grad_norm": 0.14475564658641815,
      "learning_rate": 8.99561086464603e-05,
      "loss": 10.3583,
      "step": 41
    },
    {
      "epoch": 0.7850467289719626,
      "grad_norm": 0.1279781609773636,
      "learning_rate": 8.93221332154777e-05,
      "loss": 10.3582,
      "step": 42
    },
    {
      "epoch": 0.8037383177570093,
      "grad_norm": 0.1328122615814209,
      "learning_rate": 8.86711374827494e-05,
      "loss": 10.3571,
      "step": 43
    },
    {
      "epoch": 0.822429906542056,
      "grad_norm": 0.12768304347991943,
      "learning_rate": 8.800340322708291e-05,
      "loss": 10.3559,
      "step": 44
    },
    {
      "epoch": 0.8411214953271028,
      "grad_norm": 0.13448527455329895,
      "learning_rate": 8.731921947243469e-05,
      "loss": 10.3548,
      "step": 45
    },
    {
      "epoch": 0.8598130841121495,
      "grad_norm": 0.13614797592163086,
      "learning_rate": 8.661888236280813e-05,
      "loss": 10.355,
      "step": 46
    },
    {
      "epoch": 0.8785046728971962,
      "grad_norm": 0.1163439005613327,
      "learning_rate": 8.590269503406985e-05,
      "loss": 10.3558,
      "step": 47
    },
    {
      "epoch": 0.897196261682243,
      "grad_norm": 0.12675632536411285,
      "learning_rate": 8.517096748273951e-05,
      "loss": 10.352,
      "step": 48
    },
    {
      "epoch": 0.9158878504672897,
      "grad_norm": 0.0987822562456131,
      "learning_rate": 8.442401643181e-05,
      "loss": 10.351,
      "step": 49
    },
    {
      "epoch": 0.9345794392523364,
      "grad_norm": 0.11953787505626678,
      "learning_rate": 8.366216519365621e-05,
      "loss": 10.3503,
      "step": 50
    },
    {
      "epoch": 0.9345794392523364,
      "eval_loss": 10.350443840026855,
      "eval_runtime": 0.3511,
      "eval_samples_per_second": 256.313,
      "eval_steps_per_second": 65.502,
      "step": 50
    },
    {
      "epoch": 0.9532710280373832,
      "grad_norm": 0.10493690520524979,
      "learning_rate": 8.288574353009164e-05,
      "loss": 10.3482,
      "step": 51
    },
    {
      "epoch": 0.9719626168224299,
      "grad_norm": 0.10389183461666107,
      "learning_rate": 8.209508750963328e-05,
      "loss": 10.3519,
      "step": 52
    },
    {
      "epoch": 0.9906542056074766,
      "grad_norm": 0.06221437081694603,
      "learning_rate": 8.129053936203687e-05,
      "loss": 10.3521,
      "step": 53
    },
    {
      "epoch": 1.0093457943925233,
      "grad_norm": 0.194727823138237,
      "learning_rate": 8.047244733016522e-05,
      "loss": 17.1419,
      "step": 54
    },
    {
      "epoch": 1.02803738317757,
      "grad_norm": 0.062127500772476196,
      "learning_rate": 7.964116551925365e-05,
      "loss": 8.9032,
      "step": 55
    },
    {
      "epoch": 1.0467289719626167,
      "grad_norm": 0.06462345272302628,
      "learning_rate": 7.879705374363831e-05,
      "loss": 10.3429,
      "step": 56
    },
    {
      "epoch": 1.0654205607476634,
      "grad_norm": 0.10177793353796005,
      "learning_rate": 7.794047737101297e-05,
      "loss": 10.3216,
      "step": 57
    },
    {
      "epoch": 1.0841121495327102,
      "grad_norm": 0.06885817646980286,
      "learning_rate": 7.707180716428237e-05,
      "loss": 10.265,
      "step": 58
    },
    {
      "epoch": 1.102803738317757,
      "grad_norm": 0.07267408072948456,
      "learning_rate": 7.619141912108008e-05,
      "loss": 10.3714,
      "step": 59
    },
    {
      "epoch": 1.1214953271028036,
      "grad_norm": 0.06198935583233833,
      "learning_rate": 7.529969431102064e-05,
      "loss": 10.3776,
      "step": 60
    },
    {
      "epoch": 1.1401869158878504,
      "grad_norm": 0.06444412469863892,
      "learning_rate": 7.439701871075641e-05,
      "loss": 10.3254,
      "step": 61
    },
    {
      "epoch": 1.158878504672897,
      "grad_norm": 0.08891462534666061,
      "learning_rate": 7.34837830369103e-05,
      "loss": 10.3434,
      "step": 62
    },
    {
      "epoch": 1.1775700934579438,
      "grad_norm": 0.08048012852668762,
      "learning_rate": 7.256038257695687e-05,
      "loss": 10.4076,
      "step": 63
    },
    {
      "epoch": 1.1962616822429906,
      "grad_norm": 0.0884433388710022,
      "learning_rate": 7.162721701812505e-05,
      "loss": 10.4816,
      "step": 64
    },
    {
      "epoch": 1.2149532710280373,
      "grad_norm": 0.062288835644721985,
      "learning_rate": 7.068469027439642e-05,
      "loss": 9.3343,
      "step": 65
    },
    {
      "epoch": 1.233644859813084,
      "grad_norm": 0.07321099936962128,
      "learning_rate": 6.973321031167383e-05,
      "loss": 11.4119,
      "step": 66
    },
    {
      "epoch": 1.2523364485981308,
      "grad_norm": 0.07155217975378036,
      "learning_rate": 6.87731889711965e-05,
      "loss": 11.7486,
      "step": 67
    },
    {
      "epoch": 1.2710280373831775,
      "grad_norm": 0.04724172130227089,
      "learning_rate": 6.780504179127734e-05,
      "loss": 8.78,
      "step": 68
    },
    {
      "epoch": 1.2897196261682242,
      "grad_norm": 0.07420427352190018,
      "learning_rate": 6.682918782744032e-05,
      "loss": 10.3197,
      "step": 69
    },
    {
      "epoch": 1.308411214953271,
      "grad_norm": 0.0748852863907814,
      "learning_rate": 6.584604947103514e-05,
      "loss": 10.3423,
      "step": 70
    },
    {
      "epoch": 1.3271028037383177,
      "grad_norm": 0.05861569568514824,
      "learning_rate": 6.485605226640837e-05,
      "loss": 10.3274,
      "step": 71
    },
    {
      "epoch": 1.3457943925233644,
      "grad_norm": 0.055020853877067566,
      "learning_rate": 6.385962472670953e-05,
      "loss": 10.3534,
      "step": 72
    },
    {
      "epoch": 1.3644859813084111,
      "grad_norm": 0.07727426290512085,
      "learning_rate": 6.28571981484123e-05,
      "loss": 10.3061,
      "step": 73
    },
    {
      "epoch": 1.3831775700934579,
      "grad_norm": 0.0689169391989708,
      "learning_rate": 6.184920642463094e-05,
      "loss": 10.373,
      "step": 74
    },
    {
      "epoch": 1.4018691588785046,
      "grad_norm": 0.04768117889761925,
      "learning_rate": 6.083608585731283e-05,
      "loss": 10.3421,
      "step": 75
    },
    {
      "epoch": 1.4205607476635513,
      "grad_norm": 0.06901471316814423,
      "learning_rate": 5.981827496838822e-05,
      "loss": 10.4413,
      "step": 76
    },
    {
      "epoch": 1.439252336448598,
      "grad_norm": 0.06123807653784752,
      "learning_rate": 5.8796214309959276e-05,
      "loss": 10.4401,
      "step": 77
    },
    {
      "epoch": 1.4579439252336448,
      "grad_norm": 0.06156914308667183,
      "learning_rate": 5.7770346273610254e-05,
      "loss": 9.0744,
      "step": 78
    },
    {
      "epoch": 1.4766355140186915,
      "grad_norm": 0.05391994118690491,
      "learning_rate": 5.674111489892144e-05,
      "loss": 11.5286,
      "step": 79
    },
    {
      "epoch": 1.4953271028037383,
      "grad_norm": 0.05931293964385986,
      "learning_rate": 5.570896568126993e-05,
      "loss": 11.3057,
      "step": 80
    },
    {
      "epoch": 1.514018691588785,
      "grad_norm": 0.05202610790729523,
      "learning_rate": 5.4674345379e-05,
      "loss": 9.4593,
      "step": 81
    },
    {
      "epoch": 1.5327102803738317,
      "grad_norm": 0.06128383055329323,
      "learning_rate": 5.36377018200472e-05,
      "loss": 10.2032,
      "step": 82
    },
    {
      "epoch": 1.5514018691588785,
      "grad_norm": 0.06119455397129059,
      "learning_rate": 5.2599483708099016e-05,
      "loss": 10.3166,
      "step": 83
    },
    {
      "epoch": 1.5700934579439252,
      "grad_norm": 0.06307213008403778,
      "learning_rate": 5.1560140428376956e-05,
      "loss": 10.3933,
      "step": 84
    },
    {
      "epoch": 1.588785046728972,
      "grad_norm": 0.04649597033858299,
      "learning_rate": 5.052012185312322e-05,
      "loss": 10.3412,
      "step": 85
    },
    {
      "epoch": 1.6074766355140186,
      "grad_norm": 0.06939670443534851,
      "learning_rate": 4.9479878146876794e-05,
      "loss": 10.3394,
      "step": 86
    },
    {
      "epoch": 1.6261682242990654,
      "grad_norm": 0.08162733912467957,
      "learning_rate": 4.8439859571623035e-05,
      "loss": 10.3208,
      "step": 87
    },
    {
      "epoch": 1.644859813084112,
      "grad_norm": 0.07205290347337723,
      "learning_rate": 4.740051629190099e-05,
      "loss": 10.3553,
      "step": 88
    },
    {
      "epoch": 1.6635514018691588,
      "grad_norm": 0.06231921538710594,
      "learning_rate": 4.636229817995281e-05,
      "loss": 10.4856,
      "step": 89
    },
    {
      "epoch": 1.6822429906542056,
      "grad_norm": 0.08188804984092712,
      "learning_rate": 4.532565462099999e-05,
      "loss": 10.3032,
      "step": 90
    },
    {
      "epoch": 1.7009345794392523,
      "grad_norm": 0.073176309466362,
      "learning_rate": 4.429103431873009e-05,
      "loss": 10.5851,
      "step": 91
    },
    {
      "epoch": 1.719626168224299,
      "grad_norm": 0.06884468346834183,
      "learning_rate": 4.325888510107856e-05,
      "loss": 9.2705,
      "step": 92
    },
    {
      "epoch": 1.7383177570093458,
      "grad_norm": 0.06942803412675858,
      "learning_rate": 4.2229653726389765e-05,
      "loss": 12.2043,
      "step": 93
    },
    {
      "epoch": 1.7570093457943925,
      "grad_norm": 0.04137524962425232,
      "learning_rate": 4.120378569004074e-05,
      "loss": 9.3463,
      "step": 94
    },
    {
      "epoch": 1.7757009345794392,
      "grad_norm": 0.040095340460538864,
      "learning_rate": 4.0181725031611795e-05,
      "loss": 10.2898,
      "step": 95
    },
    {
      "epoch": 1.794392523364486,
      "grad_norm": 0.05664725601673126,
      "learning_rate": 3.9163914142687184e-05,
      "loss": 10.3072,
      "step": 96
    },
    {
      "epoch": 1.8130841121495327,
      "grad_norm": 0.058860957622528076,
      "learning_rate": 3.8150793575369066e-05,
      "loss": 10.3701,
      "step": 97
    },
    {
      "epoch": 1.8317757009345794,
      "grad_norm": 0.040092833340168,
      "learning_rate": 3.714280185158771e-05,
      "loss": 10.4717,
      "step": 98
    },
    {
      "epoch": 1.8504672897196262,
      "grad_norm": 0.07228094339370728,
      "learning_rate": 3.614037527329048e-05,
      "loss": 10.2426,
      "step": 99
    },
    {
      "epoch": 1.8691588785046729,
      "grad_norm": 0.0683012455701828,
      "learning_rate": 3.514394773359163e-05,
      "loss": 10.3286,
      "step": 100
    },
    {
      "epoch": 1.8691588785046729,
      "eval_loss": 10.3464937210083,
      "eval_runtime": 0.3508,
      "eval_samples_per_second": 256.552,
      "eval_steps_per_second": 65.563,
      "step": 100
    },
    {
      "epoch": 1.8878504672897196,
      "grad_norm": 0.04906328395009041,
      "learning_rate": 3.415395052896487e-05,
      "loss": 10.3188,
      "step": 101
    },
    {
      "epoch": 1.9065420560747663,
      "grad_norm": 0.06512178480625153,
      "learning_rate": 3.31708121725597e-05,
      "loss": 10.31,
      "step": 102
    },
    {
      "epoch": 1.925233644859813,
      "grad_norm": 0.05886486545205116,
      "learning_rate": 3.219495820872265e-05,
      "loss": 10.4184,
      "step": 103
    },
    {
      "epoch": 1.9439252336448598,
      "grad_norm": 0.09364921599626541,
      "learning_rate": 3.122681102880352e-05,
      "loss": 10.342,
      "step": 104
    },
    {
      "epoch": 1.9626168224299065,
      "grad_norm": 0.06608540564775467,
      "learning_rate": 3.0266789688326186e-05,
      "loss": 8.3868,
      "step": 105
    },
    {
      "epoch": 1.9813084112149533,
      "grad_norm": 0.05854523554444313,
      "learning_rate": 2.9315309725603596e-05,
      "loss": 13.3106,
      "step": 106
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.10179751366376877,
      "learning_rate": 2.8372782981874963e-05,
      "loss": 14.4212,
      "step": 107
    },
    {
      "epoch": 2.0186915887850465,
      "grad_norm": 0.056775398552417755,
      "learning_rate": 2.7439617423043145e-05,
      "loss": 10.3451,
      "step": 108
    },
    {
      "epoch": 2.0373831775700935,
      "grad_norm": 0.03595567122101784,
      "learning_rate": 2.6516216963089698e-05,
      "loss": 10.3508,
      "step": 109
    },
    {
      "epoch": 2.05607476635514,
      "grad_norm": 0.051961012184619904,
      "learning_rate": 2.5602981289243578e-05,
      "loss": 10.348,
      "step": 110
    },
    {
      "epoch": 2.074766355140187,
      "grad_norm": 0.05135941877961159,
      "learning_rate": 2.470030568897938e-05,
      "loss": 10.3465,
      "step": 111
    },
    {
      "epoch": 2.0934579439252334,
      "grad_norm": 0.04117996618151665,
      "learning_rate": 2.3808580878919946e-05,
      "loss": 10.3505,
      "step": 112
    },
    {
      "epoch": 2.1121495327102804,
      "grad_norm": 0.049753349274396896,
      "learning_rate": 2.2928192835717644e-05,
      "loss": 10.3479,
      "step": 113
    },
    {
      "epoch": 2.130841121495327,
      "grad_norm": 0.057639382779598236,
      "learning_rate": 2.205952262898704e-05,
      "loss": 10.3513,
      "step": 114
    },
    {
      "epoch": 2.149532710280374,
      "grad_norm": 0.05707767978310585,
      "learning_rate": 2.120294625636171e-05,
      "loss": 10.3491,
      "step": 115
    },
    {
      "epoch": 2.1682242990654204,
      "grad_norm": 0.053419068455696106,
      "learning_rate": 2.0358834480746365e-05,
      "loss": 10.3475,
      "step": 116
    },
    {
      "epoch": 2.1869158878504673,
      "grad_norm": 0.050756603479385376,
      "learning_rate": 1.9527552669834798e-05,
      "loss": 10.3452,
      "step": 117
    },
    {
      "epoch": 2.205607476635514,
      "grad_norm": 0.08589330315589905,
      "learning_rate": 1.8709460637963123e-05,
      "loss": 10.3447,
      "step": 118
    },
    {
      "epoch": 2.2242990654205608,
      "grad_norm": 0.08117268979549408,
      "learning_rate": 1.790491249036672e-05,
      "loss": 10.3421,
      "step": 119
    },
    {
      "epoch": 2.2429906542056073,
      "grad_norm": 0.05790218338370323,
      "learning_rate": 1.711425646990838e-05,
      "loss": 10.3496,
      "step": 120
    },
    {
      "epoch": 2.2616822429906542,
      "grad_norm": 0.05378357321023941,
      "learning_rate": 1.6337834806343783e-05,
      "loss": 10.3466,
      "step": 121
    },
    {
      "epoch": 2.2803738317757007,
      "grad_norm": 0.04171176999807358,
      "learning_rate": 1.557598356819e-05,
      "loss": 10.349,
      "step": 122
    },
    {
      "epoch": 2.2990654205607477,
      "grad_norm": 0.06427237391471863,
      "learning_rate": 1.4829032517260489e-05,
      "loss": 10.349,
      "step": 123
    },
    {
      "epoch": 2.317757009345794,
      "grad_norm": 0.05874955654144287,
      "learning_rate": 1.4097304965930158e-05,
      "loss": 10.3489,
      "step": 124
    },
    {
      "epoch": 2.336448598130841,
      "grad_norm": 0.0420292504131794,
      "learning_rate": 1.3381117637191886e-05,
      "loss": 10.3477,
      "step": 125
    },
    {
      "epoch": 2.3551401869158877,
      "grad_norm": 0.05684233829379082,
      "learning_rate": 1.268078052756531e-05,
      "loss": 10.3465,
      "step": 126
    },
    {
      "epoch": 2.3738317757009346,
      "grad_norm": 0.04188339412212372,
      "learning_rate": 1.199659677291709e-05,
      "loss": 10.3459,
      "step": 127
    },
    {
      "epoch": 2.392523364485981,
      "grad_norm": 0.06711093336343765,
      "learning_rate": 1.132886251725061e-05,
      "loss": 10.3428,
      "step": 128
    },
    {
      "epoch": 2.411214953271028,
      "grad_norm": 0.05786636099219322,
      "learning_rate": 1.0677866784522317e-05,
      "loss": 10.3484,
      "step": 129
    },
    {
      "epoch": 2.4299065420560746,
      "grad_norm": 0.05067261680960655,
      "learning_rate": 1.004389135353972e-05,
      "loss": 10.3458,
      "step": 130
    },
    {
      "epoch": 2.4485981308411215,
      "grad_norm": 0.09212581813335419,
      "learning_rate": 9.427210635995482e-06,
      "loss": 10.3452,
      "step": 131
    },
    {
      "epoch": 2.467289719626168,
      "grad_norm": 0.08457113057374954,
      "learning_rate": 8.828091557690289e-06,
      "loss": 10.3458,
      "step": 132
    },
    {
      "epoch": 2.485981308411215,
      "grad_norm": 0.03952343761920929,
      "learning_rate": 8.246793442995954e-06,
      "loss": 10.3548,
      "step": 133
    },
    {
      "epoch": 2.5046728971962615,
      "grad_norm": 0.03213866800069809,
      "learning_rate": 7.683567902608729e-06,
      "loss": 10.3479,
      "step": 134
    },
    {
      "epoch": 2.5233644859813085,
      "grad_norm": 0.05056702718138695,
      "learning_rate": 7.138658724641417e-06,
      "loss": 10.3496,
      "step": 135
    },
    {
      "epoch": 2.542056074766355,
      "grad_norm": 0.0505887009203434,
      "learning_rate": 6.6123017691014645e-06,
      "loss": 10.35,
      "step": 136
    },
    {
      "epoch": 2.560747663551402,
      "grad_norm": 0.059739403426647186,
      "learning_rate": 6.104724865800665e-06,
      "loss": 10.3482,
      "step": 137
    },
    {
      "epoch": 2.5794392523364484,
      "grad_norm": 0.03824592009186745,
      "learning_rate": 5.616147715740611e-06,
      "loss": 10.3481,
      "step": 138
    },
    {
      "epoch": 2.5981308411214954,
      "grad_norm": 0.05563122779130936,
      "learning_rate": 5.1467817960167975e-06,
      "loss": 10.3471,
      "step": 139
    },
    {
      "epoch": 2.616822429906542,
      "grad_norm": 0.06040153279900551,
      "learning_rate": 4.696830268282204e-06,
      "loss": 10.3479,
      "step": 140
    },
    {
      "epoch": 2.635514018691589,
      "grad_norm": 0.05732399970293045,
      "learning_rate": 4.266487890810256e-06,
      "loss": 10.3474,
      "step": 141
    },
    {
      "epoch": 2.6542056074766354,
      "grad_norm": 0.06715632230043411,
      "learning_rate": 3.855940934195146e-06,
      "loss": 10.3433,
      "step": 142
    },
    {
      "epoch": 2.6728971962616823,
      "grad_norm": 0.0742998942732811,
      "learning_rate": 3.465367100725908e-06,
      "loss": 10.347,
      "step": 143
    },
    {
      "epoch": 2.691588785046729,
      "grad_norm": 0.06412313878536224,
      "learning_rate": 3.0949354474692937e-06,
      "loss": 10.3443,
      "step": 144
    },
    {
      "epoch": 2.710280373831776,
      "grad_norm": 0.060701411217451096,
      "learning_rate": 2.7448063130946224e-06,
      "loss": 10.3424,
      "step": 145
    },
    {
      "epoch": 2.7289719626168223,
      "grad_norm": 0.051176268607378006,
      "learning_rate": 2.4151312484723465e-06,
      "loss": 10.3487,
      "step": 146
    },
    {
      "epoch": 2.7476635514018692,
      "grad_norm": 0.033682528883218765,
      "learning_rate": 2.106052951076365e-06,
      "loss": 10.3463,
      "step": 147
    },
    {
      "epoch": 2.7663551401869158,
      "grad_norm": 0.04690652713179588,
      "learning_rate": 1.8177052032184283e-06,
      "loss": 10.3486,
      "step": 148
    },
    {
      "epoch": 2.7850467289719627,
      "grad_norm": 0.053933314979076385,
      "learning_rate": 1.5502128141414495e-06,
      "loss": 10.3489,
      "step": 149
    },
    {
      "epoch": 2.803738317757009,
      "grad_norm": 0.051178500056266785,
      "learning_rate": 1.303691565996712e-06,
      "loss": 10.3467,
      "step": 150
    },
    {
      "epoch": 2.803738317757009,
      "eval_loss": 10.34636402130127,
      "eval_runtime": 0.3544,
      "eval_samples_per_second": 253.969,
      "eval_steps_per_second": 64.903,
      "step": 150
    },
    {
      "epoch": 2.822429906542056,
      "grad_norm": 0.040263622999191284,
      "learning_rate": 1.0782481637284013e-06,
      "loss": 10.3467,
      "step": 151
    },
    {
      "epoch": 2.8411214953271027,
      "grad_norm": 0.0543273389339447,
      "learning_rate": 8.739801888871469e-07,
      "loss": 10.344,
      "step": 152
    },
    {
      "epoch": 2.8598130841121496,
      "grad_norm": 0.0713372528553009,
      "learning_rate": 6.909760573925561e-07,
      "loss": 10.3462,
      "step": 153
    },
    {
      "epoch": 2.878504672897196,
      "grad_norm": 0.04944860562682152,
      "learning_rate": 5.29314981262985e-07,
      "loss": 10.3462,
      "step": 154
    },
    {
      "epoch": 2.897196261682243,
      "grad_norm": 0.04204186424612999,
      "learning_rate": 3.8906693432924634e-07,
      "loss": 10.3502,
      "step": 155
    },
    {
      "epoch": 2.9158878504672896,
      "grad_norm": 0.034820400178432465,
      "learning_rate": 2.702926219468882e-07,
      "loss": 10.3499,
      "step": 156
    },
    {
      "epoch": 2.9345794392523366,
      "grad_norm": 0.06994479149580002,
      "learning_rate": 1.7304345472035631e-07,
      "loss": 10.3442,
      "step": 157
    },
    {
      "epoch": 2.953271028037383,
      "grad_norm": 0.06652852892875671,
      "learning_rate": 9.73615262502503e-08,
      "loss": 10.3429,
      "step": 158
    },
    {
      "epoch": 2.97196261682243,
      "grad_norm": 0.041439808905124664,
      "learning_rate": 4.3279594913447905e-08,
      "loss": 10.3469,
      "step": 159
    },
    {
      "epoch": 2.9906542056074765,
      "grad_norm": 0.05341007933020592,
      "learning_rate": 1.082106968385288e-08,
      "loss": 10.3456,
      "step": 160
    },
    {
      "epoch": 3.0093457943925235,
      "grad_norm": 0.09586185216903687,
      "learning_rate": 0.0,
      "loss": 16.8503,
      "step": 161
    }
  ],
  "logging_steps": 1,
  "max_steps": 161,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "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": 34706199281664.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}