{
  "best_metric": 0.5420661568641663,
  "best_model_checkpoint": "miner_id_24/checkpoint-100",
  "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": 75.5278549194336,
      "learning_rate": 1e-05,
      "loss": 4.7111,
      "step": 1
    },
    {
      "epoch": 0.018691588785046728,
      "eval_loss": 1.2935967445373535,
      "eval_runtime": 6.6536,
      "eval_samples_per_second": 13.527,
      "eval_steps_per_second": 3.457,
      "step": 1
    },
    {
      "epoch": 0.037383177570093455,
      "grad_norm": 79.87003326416016,
      "learning_rate": 2e-05,
      "loss": 4.8689,
      "step": 2
    },
    {
      "epoch": 0.056074766355140186,
      "grad_norm": 12.065667152404785,
      "learning_rate": 3e-05,
      "loss": 4.8761,
      "step": 3
    },
    {
      "epoch": 0.07476635514018691,
      "grad_norm": 8.754213333129883,
      "learning_rate": 4e-05,
      "loss": 4.4877,
      "step": 4
    },
    {
      "epoch": 0.09345794392523364,
      "grad_norm": 8.01810073852539,
      "learning_rate": 5e-05,
      "loss": 4.4319,
      "step": 5
    },
    {
      "epoch": 0.11214953271028037,
      "grad_norm": 7.005212783813477,
      "learning_rate": 6e-05,
      "loss": 4.2899,
      "step": 6
    },
    {
      "epoch": 0.1308411214953271,
      "grad_norm": 6.277080059051514,
      "learning_rate": 7e-05,
      "loss": 4.1717,
      "step": 7
    },
    {
      "epoch": 0.14953271028037382,
      "grad_norm": 6.010244369506836,
      "learning_rate": 8e-05,
      "loss": 3.5584,
      "step": 8
    },
    {
      "epoch": 0.16822429906542055,
      "grad_norm": 6.130393028259277,
      "learning_rate": 9e-05,
      "loss": 3.5363,
      "step": 9
    },
    {
      "epoch": 0.18691588785046728,
      "grad_norm": 7.306447982788086,
      "learning_rate": 0.0001,
      "loss": 3.5089,
      "step": 10
    },
    {
      "epoch": 0.205607476635514,
      "grad_norm": 5.845422267913818,
      "learning_rate": 9.998917893031616e-05,
      "loss": 3.3278,
      "step": 11
    },
    {
      "epoch": 0.22429906542056074,
      "grad_norm": 6.0055341720581055,
      "learning_rate": 9.995672040508655e-05,
      "loss": 3.2595,
      "step": 12
    },
    {
      "epoch": 0.24299065420560748,
      "grad_norm": 8.127114295959473,
      "learning_rate": 9.990263847374976e-05,
      "loss": 3.3545,
      "step": 13
    },
    {
      "epoch": 0.2616822429906542,
      "grad_norm": 5.668106555938721,
      "learning_rate": 9.982695654527965e-05,
      "loss": 3.7328,
      "step": 14
    },
    {
      "epoch": 0.2803738317757009,
      "grad_norm": 4.873712062835693,
      "learning_rate": 9.972970737805311e-05,
      "loss": 3.3928,
      "step": 15
    },
    {
      "epoch": 0.29906542056074764,
      "grad_norm": 4.584672927856445,
      "learning_rate": 9.961093306567075e-05,
      "loss": 3.478,
      "step": 16
    },
    {
      "epoch": 0.3177570093457944,
      "grad_norm": 4.411548614501953,
      "learning_rate": 9.947068501873701e-05,
      "loss": 3.2009,
      "step": 17
    },
    {
      "epoch": 0.3364485981308411,
      "grad_norm": 4.52484655380249,
      "learning_rate": 9.930902394260747e-05,
      "loss": 3.3146,
      "step": 18
    },
    {
      "epoch": 0.35514018691588783,
      "grad_norm": 4.530684471130371,
      "learning_rate": 9.912601981111286e-05,
      "loss": 2.9815,
      "step": 19
    },
    {
      "epoch": 0.37383177570093457,
      "grad_norm": 4.421339988708496,
      "learning_rate": 9.892175183627161e-05,
      "loss": 3.0047,
      "step": 20
    },
    {
      "epoch": 0.3925233644859813,
      "grad_norm": 3.9043452739715576,
      "learning_rate": 9.86963084340033e-05,
      "loss": 2.7392,
      "step": 21
    },
    {
      "epoch": 0.411214953271028,
      "grad_norm": 4.088406562805176,
      "learning_rate": 9.844978718585855e-05,
      "loss": 2.695,
      "step": 22
    },
    {
      "epoch": 0.42990654205607476,
      "grad_norm": 3.8909029960632324,
      "learning_rate": 9.818229479678158e-05,
      "loss": 2.4581,
      "step": 23
    },
    {
      "epoch": 0.4485981308411215,
      "grad_norm": 4.095215797424316,
      "learning_rate": 9.789394704892364e-05,
      "loss": 2.4724,
      "step": 24
    },
    {
      "epoch": 0.4672897196261682,
      "grad_norm": 4.23715877532959,
      "learning_rate": 9.758486875152766e-05,
      "loss": 2.542,
      "step": 25
    },
    {
      "epoch": 0.48598130841121495,
      "grad_norm": 5.026467800140381,
      "learning_rate": 9.725519368690538e-05,
      "loss": 2.4877,
      "step": 26
    },
    {
      "epoch": 0.5046728971962616,
      "grad_norm": 4.204124450683594,
      "learning_rate": 9.690506455253073e-05,
      "loss": 3.535,
      "step": 27
    },
    {
      "epoch": 0.5233644859813084,
      "grad_norm": 3.596137285232544,
      "learning_rate": 9.653463289927411e-05,
      "loss": 3.1903,
      "step": 28
    },
    {
      "epoch": 0.5420560747663551,
      "grad_norm": 3.3169479370117188,
      "learning_rate": 9.614405906580485e-05,
      "loss": 3.0368,
      "step": 29
    },
    {
      "epoch": 0.5607476635514018,
      "grad_norm": 3.6391289234161377,
      "learning_rate": 9.573351210918974e-05,
      "loss": 2.9024,
      "step": 30
    },
    {
      "epoch": 0.5794392523364486,
      "grad_norm": 3.591444253921509,
      "learning_rate": 9.53031697317178e-05,
      "loss": 2.6325,
      "step": 31
    },
    {
      "epoch": 0.5981308411214953,
      "grad_norm": 3.2920684814453125,
      "learning_rate": 9.485321820398321e-05,
      "loss": 2.7544,
      "step": 32
    },
    {
      "epoch": 0.616822429906542,
      "grad_norm": 3.3061318397521973,
      "learning_rate": 9.438385228425938e-05,
      "loss": 2.4373,
      "step": 33
    },
    {
      "epoch": 0.6355140186915887,
      "grad_norm": 3.2665421962738037,
      "learning_rate": 9.389527513419934e-05,
      "loss": 2.3135,
      "step": 34
    },
    {
      "epoch": 0.6542056074766355,
      "grad_norm": 3.2658181190490723,
      "learning_rate": 9.338769823089853e-05,
      "loss": 2.4305,
      "step": 35
    },
    {
      "epoch": 0.6728971962616822,
      "grad_norm": 3.3264851570129395,
      "learning_rate": 9.286134127535859e-05,
      "loss": 2.2953,
      "step": 36
    },
    {
      "epoch": 0.6915887850467289,
      "grad_norm": 3.2159934043884277,
      "learning_rate": 9.231643209739128e-05,
      "loss": 2.1619,
      "step": 37
    },
    {
      "epoch": 0.7102803738317757,
      "grad_norm": 3.685558319091797,
      "learning_rate": 9.175320655700406e-05,
      "loss": 2.3576,
      "step": 38
    },
    {
      "epoch": 0.7289719626168224,
      "grad_norm": 4.613719940185547,
      "learning_rate": 9.117190844230971e-05,
      "loss": 2.3371,
      "step": 39
    },
    {
      "epoch": 0.7476635514018691,
      "grad_norm": 3.3902971744537354,
      "learning_rate": 9.057278936400453e-05,
      "loss": 3.2807,
      "step": 40
    },
    {
      "epoch": 0.7663551401869159,
      "grad_norm": 3.2625577449798584,
      "learning_rate": 8.99561086464603e-05,
      "loss": 3.0179,
      "step": 41
    },
    {
      "epoch": 0.7850467289719626,
      "grad_norm": 2.896350860595703,
      "learning_rate": 8.93221332154777e-05,
      "loss": 2.4378,
      "step": 42
    },
    {
      "epoch": 0.8037383177570093,
      "grad_norm": 2.859814405441284,
      "learning_rate": 8.86711374827494e-05,
      "loss": 2.6171,
      "step": 43
    },
    {
      "epoch": 0.822429906542056,
      "grad_norm": 3.141897678375244,
      "learning_rate": 8.800340322708291e-05,
      "loss": 2.5338,
      "step": 44
    },
    {
      "epoch": 0.8411214953271028,
      "grad_norm": 3.143669605255127,
      "learning_rate": 8.731921947243469e-05,
      "loss": 2.2142,
      "step": 45
    },
    {
      "epoch": 0.8598130841121495,
      "grad_norm": 3.0150790214538574,
      "learning_rate": 8.661888236280813e-05,
      "loss": 2.2935,
      "step": 46
    },
    {
      "epoch": 0.8785046728971962,
      "grad_norm": 3.0681867599487305,
      "learning_rate": 8.590269503406985e-05,
      "loss": 2.0896,
      "step": 47
    },
    {
      "epoch": 0.897196261682243,
      "grad_norm": 3.0204715728759766,
      "learning_rate": 8.517096748273951e-05,
      "loss": 2.0929,
      "step": 48
    },
    {
      "epoch": 0.9158878504672897,
      "grad_norm": 3.0762898921966553,
      "learning_rate": 8.442401643181e-05,
      "loss": 2.2656,
      "step": 49
    },
    {
      "epoch": 0.9345794392523364,
      "grad_norm": 3.108924627304077,
      "learning_rate": 8.366216519365621e-05,
      "loss": 2.019,
      "step": 50
    },
    {
      "epoch": 0.9345794392523364,
      "eval_loss": 0.6054360866546631,
      "eval_runtime": 6.8114,
      "eval_samples_per_second": 13.213,
      "eval_steps_per_second": 3.377,
      "step": 50
    },
    {
      "epoch": 0.9532710280373832,
      "grad_norm": 3.5540387630462646,
      "learning_rate": 8.288574353009164e-05,
      "loss": 2.3411,
      "step": 51
    },
    {
      "epoch": 0.9719626168224299,
      "grad_norm": 4.078913688659668,
      "learning_rate": 8.209508750963328e-05,
      "loss": 2.4101,
      "step": 52
    },
    {
      "epoch": 0.9906542056074766,
      "grad_norm": 3.0565555095672607,
      "learning_rate": 8.129053936203687e-05,
      "loss": 2.5263,
      "step": 53
    },
    {
      "epoch": 1.0093457943925233,
      "grad_norm": 2.9273529052734375,
      "learning_rate": 8.047244733016522e-05,
      "loss": 2.2407,
      "step": 54
    },
    {
      "epoch": 1.02803738317757,
      "grad_norm": 2.5409481525421143,
      "learning_rate": 7.964116551925365e-05,
      "loss": 2.4278,
      "step": 55
    },
    {
      "epoch": 1.0467289719626167,
      "grad_norm": 2.4449081420898438,
      "learning_rate": 7.879705374363831e-05,
      "loss": 2.2068,
      "step": 56
    },
    {
      "epoch": 1.0654205607476634,
      "grad_norm": 2.635206699371338,
      "learning_rate": 7.794047737101297e-05,
      "loss": 2.1125,
      "step": 57
    },
    {
      "epoch": 1.0841121495327102,
      "grad_norm": 2.6412863731384277,
      "learning_rate": 7.707180716428237e-05,
      "loss": 1.9887,
      "step": 58
    },
    {
      "epoch": 1.102803738317757,
      "grad_norm": 2.732677459716797,
      "learning_rate": 7.619141912108008e-05,
      "loss": 1.8207,
      "step": 59
    },
    {
      "epoch": 1.1214953271028036,
      "grad_norm": 2.849564552307129,
      "learning_rate": 7.529969431102064e-05,
      "loss": 1.737,
      "step": 60
    },
    {
      "epoch": 1.1401869158878504,
      "grad_norm": 2.92907452583313,
      "learning_rate": 7.439701871075641e-05,
      "loss": 1.6603,
      "step": 61
    },
    {
      "epoch": 1.158878504672897,
      "grad_norm": 2.9295947551727295,
      "learning_rate": 7.34837830369103e-05,
      "loss": 1.5171,
      "step": 62
    },
    {
      "epoch": 1.1775700934579438,
      "grad_norm": 3.098073959350586,
      "learning_rate": 7.256038257695687e-05,
      "loss": 1.7382,
      "step": 63
    },
    {
      "epoch": 1.1962616822429906,
      "grad_norm": 2.990051031112671,
      "learning_rate": 7.162721701812505e-05,
      "loss": 1.3463,
      "step": 64
    },
    {
      "epoch": 1.2149532710280373,
      "grad_norm": 3.1232120990753174,
      "learning_rate": 7.068469027439642e-05,
      "loss": 1.4849,
      "step": 65
    },
    {
      "epoch": 1.233644859813084,
      "grad_norm": 3.6227595806121826,
      "learning_rate": 6.973321031167383e-05,
      "loss": 1.4202,
      "step": 66
    },
    {
      "epoch": 1.2523364485981308,
      "grad_norm": 3.465322971343994,
      "learning_rate": 6.87731889711965e-05,
      "loss": 2.0408,
      "step": 67
    },
    {
      "epoch": 1.2710280373831775,
      "grad_norm": 3.2346737384796143,
      "learning_rate": 6.780504179127734e-05,
      "loss": 2.587,
      "step": 68
    },
    {
      "epoch": 1.2897196261682242,
      "grad_norm": 2.9583311080932617,
      "learning_rate": 6.682918782744032e-05,
      "loss": 2.0695,
      "step": 69
    },
    {
      "epoch": 1.308411214953271,
      "grad_norm": 2.7404375076293945,
      "learning_rate": 6.584604947103514e-05,
      "loss": 1.9229,
      "step": 70
    },
    {
      "epoch": 1.3271028037383177,
      "grad_norm": 2.6932313442230225,
      "learning_rate": 6.485605226640837e-05,
      "loss": 1.8492,
      "step": 71
    },
    {
      "epoch": 1.3457943925233644,
      "grad_norm": 2.6280977725982666,
      "learning_rate": 6.385962472670953e-05,
      "loss": 1.7129,
      "step": 72
    },
    {
      "epoch": 1.3644859813084111,
      "grad_norm": 2.8215901851654053,
      "learning_rate": 6.28571981484123e-05,
      "loss": 1.897,
      "step": 73
    },
    {
      "epoch": 1.3831775700934579,
      "grad_norm": 2.6610543727874756,
      "learning_rate": 6.184920642463094e-05,
      "loss": 1.5173,
      "step": 74
    },
    {
      "epoch": 1.4018691588785046,
      "grad_norm": 2.7060275077819824,
      "learning_rate": 6.083608585731283e-05,
      "loss": 1.4563,
      "step": 75
    },
    {
      "epoch": 1.4205607476635513,
      "grad_norm": 2.716724395751953,
      "learning_rate": 5.981827496838822e-05,
      "loss": 1.4614,
      "step": 76
    },
    {
      "epoch": 1.439252336448598,
      "grad_norm": 2.8248190879821777,
      "learning_rate": 5.8796214309959276e-05,
      "loss": 1.3249,
      "step": 77
    },
    {
      "epoch": 1.4579439252336448,
      "grad_norm": 3.035578489303589,
      "learning_rate": 5.7770346273610254e-05,
      "loss": 1.4241,
      "step": 78
    },
    {
      "epoch": 1.4766355140186915,
      "grad_norm": 3.185750722885132,
      "learning_rate": 5.674111489892144e-05,
      "loss": 1.5109,
      "step": 79
    },
    {
      "epoch": 1.4953271028037383,
      "grad_norm": 3.3218882083892822,
      "learning_rate": 5.570896568126993e-05,
      "loss": 2.0867,
      "step": 80
    },
    {
      "epoch": 1.514018691588785,
      "grad_norm": 3.4760122299194336,
      "learning_rate": 5.4674345379e-05,
      "loss": 2.5535,
      "step": 81
    },
    {
      "epoch": 1.5327102803738317,
      "grad_norm": 3.2637925148010254,
      "learning_rate": 5.36377018200472e-05,
      "loss": 2.2514,
      "step": 82
    },
    {
      "epoch": 1.5514018691588785,
      "grad_norm": 2.8867454528808594,
      "learning_rate": 5.2599483708099016e-05,
      "loss": 1.9343,
      "step": 83
    },
    {
      "epoch": 1.5700934579439252,
      "grad_norm": 2.9041574001312256,
      "learning_rate": 5.1560140428376956e-05,
      "loss": 1.8784,
      "step": 84
    },
    {
      "epoch": 1.588785046728972,
      "grad_norm": 2.917921543121338,
      "learning_rate": 5.052012185312322e-05,
      "loss": 1.9778,
      "step": 85
    },
    {
      "epoch": 1.6074766355140186,
      "grad_norm": 2.7021610736846924,
      "learning_rate": 4.9479878146876794e-05,
      "loss": 1.6593,
      "step": 86
    },
    {
      "epoch": 1.6261682242990654,
      "grad_norm": 2.481092691421509,
      "learning_rate": 4.8439859571623035e-05,
      "loss": 1.4812,
      "step": 87
    },
    {
      "epoch": 1.644859813084112,
      "grad_norm": 2.678561210632324,
      "learning_rate": 4.740051629190099e-05,
      "loss": 1.4928,
      "step": 88
    },
    {
      "epoch": 1.6635514018691588,
      "grad_norm": 2.7361412048339844,
      "learning_rate": 4.636229817995281e-05,
      "loss": 1.5058,
      "step": 89
    },
    {
      "epoch": 1.6822429906542056,
      "grad_norm": 2.873293876647949,
      "learning_rate": 4.532565462099999e-05,
      "loss": 1.4377,
      "step": 90
    },
    {
      "epoch": 1.7009345794392523,
      "grad_norm": 2.7621207237243652,
      "learning_rate": 4.429103431873009e-05,
      "loss": 1.2078,
      "step": 91
    },
    {
      "epoch": 1.719626168224299,
      "grad_norm": 3.2851028442382812,
      "learning_rate": 4.325888510107856e-05,
      "loss": 1.4654,
      "step": 92
    },
    {
      "epoch": 1.7383177570093458,
      "grad_norm": 3.0891690254211426,
      "learning_rate": 4.2229653726389765e-05,
      "loss": 1.7659,
      "step": 93
    },
    {
      "epoch": 1.7570093457943925,
      "grad_norm": 2.995526075363159,
      "learning_rate": 4.120378569004074e-05,
      "loss": 2.2523,
      "step": 94
    },
    {
      "epoch": 1.7757009345794392,
      "grad_norm": 3.3323326110839844,
      "learning_rate": 4.0181725031611795e-05,
      "loss": 2.39,
      "step": 95
    },
    {
      "epoch": 1.794392523364486,
      "grad_norm": 2.783883810043335,
      "learning_rate": 3.9163914142687184e-05,
      "loss": 1.7915,
      "step": 96
    },
    {
      "epoch": 1.8130841121495327,
      "grad_norm": 2.8621714115142822,
      "learning_rate": 3.8150793575369066e-05,
      "loss": 1.7143,
      "step": 97
    },
    {
      "epoch": 1.8317757009345794,
      "grad_norm": 2.7393438816070557,
      "learning_rate": 3.714280185158771e-05,
      "loss": 1.6266,
      "step": 98
    },
    {
      "epoch": 1.8504672897196262,
      "grad_norm": 2.8787941932678223,
      "learning_rate": 3.614037527329048e-05,
      "loss": 1.7711,
      "step": 99
    },
    {
      "epoch": 1.8691588785046729,
      "grad_norm": 2.7294249534606934,
      "learning_rate": 3.514394773359163e-05,
      "loss": 1.4201,
      "step": 100
    },
    {
      "epoch": 1.8691588785046729,
      "eval_loss": 0.5420661568641663,
      "eval_runtime": 6.8129,
      "eval_samples_per_second": 13.21,
      "eval_steps_per_second": 3.376,
      "step": 100
    },
    {
      "epoch": 1.8878504672897196,
      "grad_norm": 2.7175283432006836,
      "learning_rate": 3.415395052896487e-05,
      "loss": 1.325,
      "step": 101
    },
    {
      "epoch": 1.9065420560747663,
      "grad_norm": 2.6674723625183105,
      "learning_rate": 3.31708121725597e-05,
      "loss": 1.429,
      "step": 102
    },
    {
      "epoch": 1.925233644859813,
      "grad_norm": 2.736255407333374,
      "learning_rate": 3.219495820872265e-05,
      "loss": 1.4033,
      "step": 103
    },
    {
      "epoch": 1.9439252336448598,
      "grad_norm": 3.0553455352783203,
      "learning_rate": 3.122681102880352e-05,
      "loss": 1.3932,
      "step": 104
    },
    {
      "epoch": 1.9626168224299065,
      "grad_norm": 3.263404369354248,
      "learning_rate": 3.0266789688326186e-05,
      "loss": 1.401,
      "step": 105
    },
    {
      "epoch": 1.9813084112149533,
      "grad_norm": 3.210214614868164,
      "learning_rate": 2.9315309725603596e-05,
      "loss": 1.725,
      "step": 106
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.974839210510254,
      "learning_rate": 2.8372782981874963e-05,
      "loss": 1.4878,
      "step": 107
    },
    {
      "epoch": 2.0186915887850465,
      "grad_norm": 2.363774538040161,
      "learning_rate": 2.7439617423043145e-05,
      "loss": 1.9556,
      "step": 108
    },
    {
      "epoch": 2.0373831775700935,
      "grad_norm": 2.4537739753723145,
      "learning_rate": 2.6516216963089698e-05,
      "loss": 1.7125,
      "step": 109
    },
    {
      "epoch": 2.05607476635514,
      "grad_norm": 2.437591791152954,
      "learning_rate": 2.5602981289243578e-05,
      "loss": 1.4821,
      "step": 110
    },
    {
      "epoch": 2.074766355140187,
      "grad_norm": 2.365839719772339,
      "learning_rate": 2.470030568897938e-05,
      "loss": 1.267,
      "step": 111
    },
    {
      "epoch": 2.0934579439252334,
      "grad_norm": 2.4470767974853516,
      "learning_rate": 2.3808580878919946e-05,
      "loss": 1.158,
      "step": 112
    },
    {
      "epoch": 2.1121495327102804,
      "grad_norm": 2.586884021759033,
      "learning_rate": 2.2928192835717644e-05,
      "loss": 1.1941,
      "step": 113
    },
    {
      "epoch": 2.130841121495327,
      "grad_norm": 2.675332546234131,
      "learning_rate": 2.205952262898704e-05,
      "loss": 1.2099,
      "step": 114
    },
    {
      "epoch": 2.149532710280374,
      "grad_norm": 2.671232223510742,
      "learning_rate": 2.120294625636171e-05,
      "loss": 0.9791,
      "step": 115
    },
    {
      "epoch": 2.1682242990654204,
      "grad_norm": 2.761700391769409,
      "learning_rate": 2.0358834480746365e-05,
      "loss": 0.9811,
      "step": 116
    },
    {
      "epoch": 2.1869158878504673,
      "grad_norm": 2.856433868408203,
      "learning_rate": 1.9527552669834798e-05,
      "loss": 0.9574,
      "step": 117
    },
    {
      "epoch": 2.205607476635514,
      "grad_norm": 2.6864655017852783,
      "learning_rate": 1.8709460637963123e-05,
      "loss": 0.7894,
      "step": 118
    },
    {
      "epoch": 2.2242990654205608,
      "grad_norm": 2.9782700538635254,
      "learning_rate": 1.790491249036672e-05,
      "loss": 0.8351,
      "step": 119
    },
    {
      "epoch": 2.2429906542056073,
      "grad_norm": 3.188793659210205,
      "learning_rate": 1.711425646990838e-05,
      "loss": 0.861,
      "step": 120
    },
    {
      "epoch": 2.2616822429906542,
      "grad_norm": 3.1211624145507812,
      "learning_rate": 1.6337834806343783e-05,
      "loss": 1.9863,
      "step": 121
    },
    {
      "epoch": 2.2803738317757007,
      "grad_norm": 3.115051031112671,
      "learning_rate": 1.557598356819e-05,
      "loss": 1.5083,
      "step": 122
    },
    {
      "epoch": 2.2990654205607477,
      "grad_norm": 2.92958927154541,
      "learning_rate": 1.4829032517260489e-05,
      "loss": 1.284,
      "step": 123
    },
    {
      "epoch": 2.317757009345794,
      "grad_norm": 3.037280797958374,
      "learning_rate": 1.4097304965930158e-05,
      "loss": 1.2021,
      "step": 124
    },
    {
      "epoch": 2.336448598130841,
      "grad_norm": 2.9107697010040283,
      "learning_rate": 1.3381117637191886e-05,
      "loss": 1.0792,
      "step": 125
    },
    {
      "epoch": 2.3551401869158877,
      "grad_norm": 2.8054959774017334,
      "learning_rate": 1.268078052756531e-05,
      "loss": 1.0345,
      "step": 126
    },
    {
      "epoch": 2.3738317757009346,
      "grad_norm": 2.7768619060516357,
      "learning_rate": 1.199659677291709e-05,
      "loss": 0.9557,
      "step": 127
    },
    {
      "epoch": 2.392523364485981,
      "grad_norm": 2.694972038269043,
      "learning_rate": 1.132886251725061e-05,
      "loss": 0.8214,
      "step": 128
    },
    {
      "epoch": 2.411214953271028,
      "grad_norm": 2.806529998779297,
      "learning_rate": 1.0677866784522317e-05,
      "loss": 0.8501,
      "step": 129
    },
    {
      "epoch": 2.4299065420560746,
      "grad_norm": 2.850677490234375,
      "learning_rate": 1.004389135353972e-05,
      "loss": 0.9375,
      "step": 130
    },
    {
      "epoch": 2.4485981308411215,
      "grad_norm": 3.034752130508423,
      "learning_rate": 9.427210635995482e-06,
      "loss": 0.883,
      "step": 131
    },
    {
      "epoch": 2.467289719626168,
      "grad_norm": 3.2129502296447754,
      "learning_rate": 8.828091557690289e-06,
      "loss": 0.937,
      "step": 132
    },
    {
      "epoch": 2.485981308411215,
      "grad_norm": 3.229306936264038,
      "learning_rate": 8.246793442995954e-06,
      "loss": 0.6623,
      "step": 133
    },
    {
      "epoch": 2.5046728971962615,
      "grad_norm": 2.887923002243042,
      "learning_rate": 7.683567902608729e-06,
      "loss": 1.779,
      "step": 134
    },
    {
      "epoch": 2.5233644859813085,
      "grad_norm": 2.9174389839172363,
      "learning_rate": 7.138658724641417e-06,
      "loss": 1.5203,
      "step": 135
    },
    {
      "epoch": 2.542056074766355,
      "grad_norm": 2.7020833492279053,
      "learning_rate": 6.6123017691014645e-06,
      "loss": 1.1735,
      "step": 136
    },
    {
      "epoch": 2.560747663551402,
      "grad_norm": 2.930314779281616,
      "learning_rate": 6.104724865800665e-06,
      "loss": 1.304,
      "step": 137
    },
    {
      "epoch": 2.5794392523364484,
      "grad_norm": 2.734633684158325,
      "learning_rate": 5.616147715740611e-06,
      "loss": 1.0521,
      "step": 138
    },
    {
      "epoch": 2.5981308411214954,
      "grad_norm": 2.814070463180542,
      "learning_rate": 5.1467817960167975e-06,
      "loss": 1.1024,
      "step": 139
    },
    {
      "epoch": 2.616822429906542,
      "grad_norm": 2.812589406967163,
      "learning_rate": 4.696830268282204e-06,
      "loss": 0.9961,
      "step": 140
    },
    {
      "epoch": 2.635514018691589,
      "grad_norm": 2.73301100730896,
      "learning_rate": 4.266487890810256e-06,
      "loss": 0.8994,
      "step": 141
    },
    {
      "epoch": 2.6542056074766354,
      "grad_norm": 2.720045328140259,
      "learning_rate": 3.855940934195146e-06,
      "loss": 0.8717,
      "step": 142
    },
    {
      "epoch": 2.6728971962616823,
      "grad_norm": 2.7447330951690674,
      "learning_rate": 3.465367100725908e-06,
      "loss": 0.8762,
      "step": 143
    },
    {
      "epoch": 2.691588785046729,
      "grad_norm": 2.8292927742004395,
      "learning_rate": 3.0949354474692937e-06,
      "loss": 0.7848,
      "step": 144
    },
    {
      "epoch": 2.710280373831776,
      "grad_norm": 3.1395366191864014,
      "learning_rate": 2.7448063130946224e-06,
      "loss": 0.8932,
      "step": 145
    },
    {
      "epoch": 2.7289719626168223,
      "grad_norm": 3.0559487342834473,
      "learning_rate": 2.4151312484723465e-06,
      "loss": 0.7451,
      "step": 146
    },
    {
      "epoch": 2.7476635514018692,
      "grad_norm": 2.800384759902954,
      "learning_rate": 2.106052951076365e-06,
      "loss": 1.7696,
      "step": 147
    },
    {
      "epoch": 2.7663551401869158,
      "grad_norm": 3.025876760482788,
      "learning_rate": 1.8177052032184283e-06,
      "loss": 1.6944,
      "step": 148
    },
    {
      "epoch": 2.7850467289719627,
      "grad_norm": 2.750108242034912,
      "learning_rate": 1.5502128141414495e-06,
      "loss": 1.2296,
      "step": 149
    },
    {
      "epoch": 2.803738317757009,
      "grad_norm": 2.793736219406128,
      "learning_rate": 1.303691565996712e-06,
      "loss": 1.2055,
      "step": 150
    },
    {
      "epoch": 2.803738317757009,
      "eval_loss": 0.5477179884910583,
      "eval_runtime": 6.8079,
      "eval_samples_per_second": 13.22,
      "eval_steps_per_second": 3.378,
      "step": 150
    },
    {
      "epoch": 2.822429906542056,
      "grad_norm": 2.9440102577209473,
      "learning_rate": 1.0782481637284013e-06,
      "loss": 1.2952,
      "step": 151
    },
    {
      "epoch": 2.8411214953271027,
      "grad_norm": 2.854104518890381,
      "learning_rate": 8.739801888871469e-07,
      "loss": 1.117,
      "step": 152
    },
    {
      "epoch": 2.8598130841121496,
      "grad_norm": 2.865081310272217,
      "learning_rate": 6.909760573925561e-07,
      "loss": 1.0454,
      "step": 153
    },
    {
      "epoch": 2.878504672897196,
      "grad_norm": 2.6739699840545654,
      "learning_rate": 5.29314981262985e-07,
      "loss": 0.8551,
      "step": 154
    },
    {
      "epoch": 2.897196261682243,
      "grad_norm": 2.7586936950683594,
      "learning_rate": 3.8906693432924634e-07,
      "loss": 0.8919,
      "step": 155
    },
    {
      "epoch": 2.9158878504672896,
      "grad_norm": 2.8043391704559326,
      "learning_rate": 2.702926219468882e-07,
      "loss": 0.8729,
      "step": 156
    },
    {
      "epoch": 2.9345794392523366,
      "grad_norm": 2.6751370429992676,
      "learning_rate": 1.7304345472035631e-07,
      "loss": 0.7368,
      "step": 157
    },
    {
      "epoch": 2.953271028037383,
      "grad_norm": 2.866875171661377,
      "learning_rate": 9.73615262502503e-08,
      "loss": 0.7555,
      "step": 158
    },
    {
      "epoch": 2.97196261682243,
      "grad_norm": 3.274054527282715,
      "learning_rate": 4.3279594913447905e-08,
      "loss": 0.7321,
      "step": 159
    },
    {
      "epoch": 2.9906542056074765,
      "grad_norm": 2.7593677043914795,
      "learning_rate": 1.082106968385288e-08,
      "loss": 1.1862,
      "step": 160
    },
    {
      "epoch": 3.0093457943925235,
      "grad_norm": 2.860818386077881,
      "learning_rate": 0.0,
      "loss": 1.2629,
      "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": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.327271547254866e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}