{
  "best_metric": 6.854687690734863,
  "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": 0.2938673198223114,
      "learning_rate": 1e-05,
      "loss": 6.9441,
      "step": 1
    },
    {
      "epoch": 0.018691588785046728,
      "eval_loss": 6.942073345184326,
      "eval_runtime": 0.1271,
      "eval_samples_per_second": 707.983,
      "eval_steps_per_second": 180.929,
      "step": 1
    },
    {
      "epoch": 0.037383177570093455,
      "grad_norm": 0.3331543803215027,
      "learning_rate": 2e-05,
      "loss": 6.9483,
      "step": 2
    },
    {
      "epoch": 0.056074766355140186,
      "grad_norm": 0.36663663387298584,
      "learning_rate": 3e-05,
      "loss": 6.945,
      "step": 3
    },
    {
      "epoch": 0.07476635514018691,
      "grad_norm": 0.30907315015792847,
      "learning_rate": 4e-05,
      "loss": 6.9434,
      "step": 4
    },
    {
      "epoch": 0.09345794392523364,
      "grad_norm": 0.355998694896698,
      "learning_rate": 5e-05,
      "loss": 6.9413,
      "step": 5
    },
    {
      "epoch": 0.11214953271028037,
      "grad_norm": 0.40504294633865356,
      "learning_rate": 6e-05,
      "loss": 6.9436,
      "step": 6
    },
    {
      "epoch": 0.1308411214953271,
      "grad_norm": 0.27534759044647217,
      "learning_rate": 7e-05,
      "loss": 6.9404,
      "step": 7
    },
    {
      "epoch": 0.14953271028037382,
      "grad_norm": 0.19393278658390045,
      "learning_rate": 8e-05,
      "loss": 6.935,
      "step": 8
    },
    {
      "epoch": 0.16822429906542055,
      "grad_norm": 0.23395928740501404,
      "learning_rate": 9e-05,
      "loss": 6.9348,
      "step": 9
    },
    {
      "epoch": 0.18691588785046728,
      "grad_norm": 0.210158571600914,
      "learning_rate": 0.0001,
      "loss": 6.9368,
      "step": 10
    },
    {
      "epoch": 0.205607476635514,
      "grad_norm": 0.2233603298664093,
      "learning_rate": 9.998917893031616e-05,
      "loss": 6.9362,
      "step": 11
    },
    {
      "epoch": 0.22429906542056074,
      "grad_norm": 0.22712969779968262,
      "learning_rate": 9.995672040508655e-05,
      "loss": 6.9324,
      "step": 12
    },
    {
      "epoch": 0.24299065420560748,
      "grad_norm": 0.26222285628318787,
      "learning_rate": 9.990263847374976e-05,
      "loss": 6.927,
      "step": 13
    },
    {
      "epoch": 0.2616822429906542,
      "grad_norm": 0.27522316575050354,
      "learning_rate": 9.982695654527965e-05,
      "loss": 6.9338,
      "step": 14
    },
    {
      "epoch": 0.2803738317757009,
      "grad_norm": 0.33045950531959534,
      "learning_rate": 9.972970737805311e-05,
      "loss": 6.9289,
      "step": 15
    },
    {
      "epoch": 0.29906542056074764,
      "grad_norm": 0.3802172839641571,
      "learning_rate": 9.961093306567075e-05,
      "loss": 6.9237,
      "step": 16
    },
    {
      "epoch": 0.3177570093457944,
      "grad_norm": 0.38254377245903015,
      "learning_rate": 9.947068501873701e-05,
      "loss": 6.9202,
      "step": 17
    },
    {
      "epoch": 0.3364485981308411,
      "grad_norm": 0.3661504089832306,
      "learning_rate": 9.930902394260747e-05,
      "loss": 6.9184,
      "step": 18
    },
    {
      "epoch": 0.35514018691588783,
      "grad_norm": 0.3992804288864136,
      "learning_rate": 9.912601981111286e-05,
      "loss": 6.9178,
      "step": 19
    },
    {
      "epoch": 0.37383177570093457,
      "grad_norm": 0.3236754834651947,
      "learning_rate": 9.892175183627161e-05,
      "loss": 6.9215,
      "step": 20
    },
    {
      "epoch": 0.3925233644859813,
      "grad_norm": 0.2195035219192505,
      "learning_rate": 9.86963084340033e-05,
      "loss": 6.9283,
      "step": 21
    },
    {
      "epoch": 0.411214953271028,
      "grad_norm": 0.2215508669614792,
      "learning_rate": 9.844978718585855e-05,
      "loss": 6.9281,
      "step": 22
    },
    {
      "epoch": 0.42990654205607476,
      "grad_norm": 0.21688078343868256,
      "learning_rate": 9.818229479678158e-05,
      "loss": 6.9253,
      "step": 23
    },
    {
      "epoch": 0.4485981308411215,
      "grad_norm": 0.21395814418792725,
      "learning_rate": 9.789394704892364e-05,
      "loss": 6.928,
      "step": 24
    },
    {
      "epoch": 0.4672897196261682,
      "grad_norm": 0.23305416107177734,
      "learning_rate": 9.758486875152766e-05,
      "loss": 6.9286,
      "step": 25
    },
    {
      "epoch": 0.48598130841121495,
      "grad_norm": 0.2735269069671631,
      "learning_rate": 9.725519368690538e-05,
      "loss": 6.9236,
      "step": 26
    },
    {
      "epoch": 0.5046728971962616,
      "grad_norm": 0.30058524012565613,
      "learning_rate": 9.690506455253073e-05,
      "loss": 6.914,
      "step": 27
    },
    {
      "epoch": 0.5233644859813084,
      "grad_norm": 0.3103170096874237,
      "learning_rate": 9.653463289927411e-05,
      "loss": 6.9108,
      "step": 28
    },
    {
      "epoch": 0.5420560747663551,
      "grad_norm": 0.389228492975235,
      "learning_rate": 9.614405906580485e-05,
      "loss": 6.8998,
      "step": 29
    },
    {
      "epoch": 0.5607476635514018,
      "grad_norm": 0.3750993609428406,
      "learning_rate": 9.573351210918974e-05,
      "loss": 6.8996,
      "step": 30
    },
    {
      "epoch": 0.5794392523364486,
      "grad_norm": 0.38516804575920105,
      "learning_rate": 9.53031697317178e-05,
      "loss": 6.8939,
      "step": 31
    },
    {
      "epoch": 0.5981308411214953,
      "grad_norm": 0.44313955307006836,
      "learning_rate": 9.485321820398321e-05,
      "loss": 6.8893,
      "step": 32
    },
    {
      "epoch": 0.616822429906542,
      "grad_norm": 0.3667280972003937,
      "learning_rate": 9.438385228425938e-05,
      "loss": 6.8969,
      "step": 33
    },
    {
      "epoch": 0.6355140186915887,
      "grad_norm": 0.25418129563331604,
      "learning_rate": 9.389527513419934e-05,
      "loss": 6.9136,
      "step": 34
    },
    {
      "epoch": 0.6542056074766355,
      "grad_norm": 0.23331645131111145,
      "learning_rate": 9.338769823089853e-05,
      "loss": 6.9183,
      "step": 35
    },
    {
      "epoch": 0.6728971962616822,
      "grad_norm": 0.24229319393634796,
      "learning_rate": 9.286134127535859e-05,
      "loss": 6.9153,
      "step": 36
    },
    {
      "epoch": 0.6915887850467289,
      "grad_norm": 0.24221663177013397,
      "learning_rate": 9.231643209739128e-05,
      "loss": 6.9107,
      "step": 37
    },
    {
      "epoch": 0.7102803738317757,
      "grad_norm": 0.26956361532211304,
      "learning_rate": 9.175320655700406e-05,
      "loss": 6.9103,
      "step": 38
    },
    {
      "epoch": 0.7289719626168224,
      "grad_norm": 0.3278099298477173,
      "learning_rate": 9.117190844230971e-05,
      "loss": 6.9031,
      "step": 39
    },
    {
      "epoch": 0.7476635514018691,
      "grad_norm": 0.25633710622787476,
      "learning_rate": 9.057278936400453e-05,
      "loss": 6.8918,
      "step": 40
    },
    {
      "epoch": 0.7663551401869159,
      "grad_norm": 0.2838684618473053,
      "learning_rate": 8.99561086464603e-05,
      "loss": 6.889,
      "step": 41
    },
    {
      "epoch": 0.7850467289719626,
      "grad_norm": 0.29884856939315796,
      "learning_rate": 8.93221332154777e-05,
      "loss": 6.8815,
      "step": 42
    },
    {
      "epoch": 0.8037383177570093,
      "grad_norm": 0.3151189386844635,
      "learning_rate": 8.86711374827494e-05,
      "loss": 6.8698,
      "step": 43
    },
    {
      "epoch": 0.822429906542056,
      "grad_norm": 0.31718558073043823,
      "learning_rate": 8.800340322708291e-05,
      "loss": 6.8722,
      "step": 44
    },
    {
      "epoch": 0.8411214953271028,
      "grad_norm": 0.3026714026927948,
      "learning_rate": 8.731921947243469e-05,
      "loss": 6.8764,
      "step": 45
    },
    {
      "epoch": 0.8598130841121495,
      "grad_norm": 0.2692744731903076,
      "learning_rate": 8.661888236280813e-05,
      "loss": 6.8804,
      "step": 46
    },
    {
      "epoch": 0.8785046728971962,
      "grad_norm": 0.23935897648334503,
      "learning_rate": 8.590269503406985e-05,
      "loss": 6.8864,
      "step": 47
    },
    {
      "epoch": 0.897196261682243,
      "grad_norm": 0.2529131770133972,
      "learning_rate": 8.517096748273951e-05,
      "loss": 6.9087,
      "step": 48
    },
    {
      "epoch": 0.9158878504672897,
      "grad_norm": 0.23005519807338715,
      "learning_rate": 8.442401643181e-05,
      "loss": 6.8997,
      "step": 49
    },
    {
      "epoch": 0.9345794392523364,
      "grad_norm": 0.2434474676847458,
      "learning_rate": 8.366216519365621e-05,
      "loss": 6.9071,
      "step": 50
    },
    {
      "epoch": 0.9345794392523364,
      "eval_loss": 6.879741191864014,
      "eval_runtime": 0.1279,
      "eval_samples_per_second": 703.723,
      "eval_steps_per_second": 179.84,
      "step": 50
    },
    {
      "epoch": 0.9532710280373832,
      "grad_norm": 0.2575473487377167,
      "learning_rate": 8.288574353009164e-05,
      "loss": 6.8969,
      "step": 51
    },
    {
      "epoch": 0.9719626168224299,
      "grad_norm": 0.26143884658813477,
      "learning_rate": 8.209508750963328e-05,
      "loss": 6.8978,
      "step": 52
    },
    {
      "epoch": 0.9906542056074766,
      "grad_norm": 0.2457752525806427,
      "learning_rate": 8.129053936203687e-05,
      "loss": 6.8713,
      "step": 53
    },
    {
      "epoch": 1.0093457943925233,
      "grad_norm": 0.3516462743282318,
      "learning_rate": 8.047244733016522e-05,
      "loss": 10.8882,
      "step": 54
    },
    {
      "epoch": 1.02803738317757,
      "grad_norm": 0.2064395397901535,
      "learning_rate": 7.964116551925365e-05,
      "loss": 6.5968,
      "step": 55
    },
    {
      "epoch": 1.0467289719626167,
      "grad_norm": 0.2576603293418884,
      "learning_rate": 7.879705374363831e-05,
      "loss": 6.4406,
      "step": 56
    },
    {
      "epoch": 1.0654205607476634,
      "grad_norm": 0.29941892623901367,
      "learning_rate": 7.794047737101297e-05,
      "loss": 7.1305,
      "step": 57
    },
    {
      "epoch": 1.0841121495327102,
      "grad_norm": 0.2890705168247223,
      "learning_rate": 7.707180716428237e-05,
      "loss": 6.8306,
      "step": 58
    },
    {
      "epoch": 1.102803738317757,
      "grad_norm": 0.26566511392593384,
      "learning_rate": 7.619141912108008e-05,
      "loss": 6.8318,
      "step": 59
    },
    {
      "epoch": 1.1214953271028036,
      "grad_norm": 0.26670339703559875,
      "learning_rate": 7.529969431102064e-05,
      "loss": 6.5757,
      "step": 60
    },
    {
      "epoch": 1.1401869158878504,
      "grad_norm": 0.21569110453128815,
      "learning_rate": 7.439701871075641e-05,
      "loss": 6.9524,
      "step": 61
    },
    {
      "epoch": 1.158878504672897,
      "grad_norm": 0.20257577300071716,
      "learning_rate": 7.34837830369103e-05,
      "loss": 6.9015,
      "step": 62
    },
    {
      "epoch": 1.1775700934579438,
      "grad_norm": 0.20113326609134674,
      "learning_rate": 7.256038257695687e-05,
      "loss": 7.0185,
      "step": 63
    },
    {
      "epoch": 1.1962616822429906,
      "grad_norm": 0.20851078629493713,
      "learning_rate": 7.162721701812505e-05,
      "loss": 6.7535,
      "step": 64
    },
    {
      "epoch": 1.2149532710280373,
      "grad_norm": 0.23403489589691162,
      "learning_rate": 7.068469027439642e-05,
      "loss": 7.2187,
      "step": 65
    },
    {
      "epoch": 1.233644859813084,
      "grad_norm": 0.2020825296640396,
      "learning_rate": 6.973321031167383e-05,
      "loss": 7.1485,
      "step": 66
    },
    {
      "epoch": 1.2523364485981308,
      "grad_norm": 0.21163055300712585,
      "learning_rate": 6.87731889711965e-05,
      "loss": 7.1758,
      "step": 67
    },
    {
      "epoch": 1.2710280373831775,
      "grad_norm": 0.2162235528230667,
      "learning_rate": 6.780504179127734e-05,
      "loss": 6.5901,
      "step": 68
    },
    {
      "epoch": 1.2897196261682242,
      "grad_norm": 0.22019129991531372,
      "learning_rate": 6.682918782744032e-05,
      "loss": 6.6025,
      "step": 69
    },
    {
      "epoch": 1.308411214953271,
      "grad_norm": 0.2604824900627136,
      "learning_rate": 6.584604947103514e-05,
      "loss": 6.7204,
      "step": 70
    },
    {
      "epoch": 1.3271028037383177,
      "grad_norm": 0.2431761920452118,
      "learning_rate": 6.485605226640837e-05,
      "loss": 6.8815,
      "step": 71
    },
    {
      "epoch": 1.3457943925233644,
      "grad_norm": 0.2991611361503601,
      "learning_rate": 6.385962472670953e-05,
      "loss": 6.8737,
      "step": 72
    },
    {
      "epoch": 1.3644859813084111,
      "grad_norm": 0.3000471889972687,
      "learning_rate": 6.28571981484123e-05,
      "loss": 6.4497,
      "step": 73
    },
    {
      "epoch": 1.3831775700934579,
      "grad_norm": 0.3030647337436676,
      "learning_rate": 6.184920642463094e-05,
      "loss": 6.3958,
      "step": 74
    },
    {
      "epoch": 1.4018691588785046,
      "grad_norm": 0.2308041751384735,
      "learning_rate": 6.083608585731283e-05,
      "loss": 7.7412,
      "step": 75
    },
    {
      "epoch": 1.4205607476635513,
      "grad_norm": 0.2113001048564911,
      "learning_rate": 5.981827496838822e-05,
      "loss": 6.8193,
      "step": 76
    },
    {
      "epoch": 1.439252336448598,
      "grad_norm": 0.19752728939056396,
      "learning_rate": 5.8796214309959276e-05,
      "loss": 7.0249,
      "step": 77
    },
    {
      "epoch": 1.4579439252336448,
      "grad_norm": 0.17403879761695862,
      "learning_rate": 5.7770346273610254e-05,
      "loss": 6.8756,
      "step": 78
    },
    {
      "epoch": 1.4766355140186915,
      "grad_norm": 0.2119225412607193,
      "learning_rate": 5.674111489892144e-05,
      "loss": 7.2423,
      "step": 79
    },
    {
      "epoch": 1.4953271028037383,
      "grad_norm": 0.21908855438232422,
      "learning_rate": 5.570896568126993e-05,
      "loss": 6.9398,
      "step": 80
    },
    {
      "epoch": 1.514018691588785,
      "grad_norm": 0.29880014061927795,
      "learning_rate": 5.4674345379e-05,
      "loss": 6.3031,
      "step": 81
    },
    {
      "epoch": 1.5327102803738317,
      "grad_norm": 0.2394288182258606,
      "learning_rate": 5.36377018200472e-05,
      "loss": 6.7421,
      "step": 82
    },
    {
      "epoch": 1.5514018691588785,
      "grad_norm": 0.29015251994132996,
      "learning_rate": 5.2599483708099016e-05,
      "loss": 6.901,
      "step": 83
    },
    {
      "epoch": 1.5700934579439252,
      "grad_norm": 0.23098227381706238,
      "learning_rate": 5.1560140428376956e-05,
      "loss": 6.8375,
      "step": 84
    },
    {
      "epoch": 1.588785046728972,
      "grad_norm": 0.3023429214954376,
      "learning_rate": 5.052012185312322e-05,
      "loss": 7.3421,
      "step": 85
    },
    {
      "epoch": 1.6074766355140186,
      "grad_norm": 0.2821560800075531,
      "learning_rate": 4.9479878146876794e-05,
      "loss": 6.1029,
      "step": 86
    },
    {
      "epoch": 1.6261682242990654,
      "grad_norm": 0.24317847192287445,
      "learning_rate": 4.8439859571623035e-05,
      "loss": 6.855,
      "step": 87
    },
    {
      "epoch": 1.644859813084112,
      "grad_norm": 0.211278036236763,
      "learning_rate": 4.740051629190099e-05,
      "loss": 7.4823,
      "step": 88
    },
    {
      "epoch": 1.6635514018691588,
      "grad_norm": 0.16635450720787048,
      "learning_rate": 4.636229817995281e-05,
      "loss": 6.8219,
      "step": 89
    },
    {
      "epoch": 1.6822429906542056,
      "grad_norm": 0.18342377245426178,
      "learning_rate": 4.532565462099999e-05,
      "loss": 6.675,
      "step": 90
    },
    {
      "epoch": 1.7009345794392523,
      "grad_norm": 0.18881668150424957,
      "learning_rate": 4.429103431873009e-05,
      "loss": 7.0165,
      "step": 91
    },
    {
      "epoch": 1.719626168224299,
      "grad_norm": 0.21899811923503876,
      "learning_rate": 4.325888510107856e-05,
      "loss": 7.1536,
      "step": 92
    },
    {
      "epoch": 1.7383177570093458,
      "grad_norm": 0.1800474375486374,
      "learning_rate": 4.2229653726389765e-05,
      "loss": 7.3534,
      "step": 93
    },
    {
      "epoch": 1.7570093457943925,
      "grad_norm": 0.2928697466850281,
      "learning_rate": 4.120378569004074e-05,
      "loss": 5.8943,
      "step": 94
    },
    {
      "epoch": 1.7757009345794392,
      "grad_norm": 0.28914374113082886,
      "learning_rate": 4.0181725031611795e-05,
      "loss": 6.6657,
      "step": 95
    },
    {
      "epoch": 1.794392523364486,
      "grad_norm": 0.28518757224082947,
      "learning_rate": 3.9163914142687184e-05,
      "loss": 7.0581,
      "step": 96
    },
    {
      "epoch": 1.8130841121495327,
      "grad_norm": 0.33743077516555786,
      "learning_rate": 3.8150793575369066e-05,
      "loss": 7.1519,
      "step": 97
    },
    {
      "epoch": 1.8317757009345794,
      "grad_norm": 0.32854652404785156,
      "learning_rate": 3.714280185158771e-05,
      "loss": 6.8511,
      "step": 98
    },
    {
      "epoch": 1.8504672897196262,
      "grad_norm": 0.2562345564365387,
      "learning_rate": 3.614037527329048e-05,
      "loss": 6.0772,
      "step": 99
    },
    {
      "epoch": 1.8691588785046729,
      "grad_norm": 0.18754076957702637,
      "learning_rate": 3.514394773359163e-05,
      "loss": 7.3537,
      "step": 100
    },
    {
      "epoch": 1.8691588785046729,
      "eval_loss": 6.858725547790527,
      "eval_runtime": 0.1246,
      "eval_samples_per_second": 722.397,
      "eval_steps_per_second": 184.613,
      "step": 100
    },
    {
      "epoch": 1.8878504672897196,
      "grad_norm": 0.18902823328971863,
      "learning_rate": 3.415395052896487e-05,
      "loss": 7.089,
      "step": 101
    },
    {
      "epoch": 1.9065420560747663,
      "grad_norm": 0.18071438372135162,
      "learning_rate": 3.31708121725597e-05,
      "loss": 6.6421,
      "step": 102
    },
    {
      "epoch": 1.925233644859813,
      "grad_norm": 0.22222863137722015,
      "learning_rate": 3.219495820872265e-05,
      "loss": 7.017,
      "step": 103
    },
    {
      "epoch": 1.9439252336448598,
      "grad_norm": 0.18907681107521057,
      "learning_rate": 3.122681102880352e-05,
      "loss": 6.9228,
      "step": 104
    },
    {
      "epoch": 1.9626168224299065,
      "grad_norm": 0.2056841105222702,
      "learning_rate": 3.0266789688326186e-05,
      "loss": 7.1821,
      "step": 105
    },
    {
      "epoch": 1.9813084112149533,
      "grad_norm": 0.2206508368253708,
      "learning_rate": 2.9315309725603596e-05,
      "loss": 7.0746,
      "step": 106
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.28758642077445984,
      "learning_rate": 2.8372782981874963e-05,
      "loss": 9.763,
      "step": 107
    },
    {
      "epoch": 2.0186915887850465,
      "grad_norm": 0.23370413482189178,
      "learning_rate": 2.7439617423043145e-05,
      "loss": 6.8612,
      "step": 108
    },
    {
      "epoch": 2.0373831775700935,
      "grad_norm": 0.27056005597114563,
      "learning_rate": 2.6516216963089698e-05,
      "loss": 6.8443,
      "step": 109
    },
    {
      "epoch": 2.05607476635514,
      "grad_norm": 0.25762417912483215,
      "learning_rate": 2.5602981289243578e-05,
      "loss": 6.8467,
      "step": 110
    },
    {
      "epoch": 2.074766355140187,
      "grad_norm": 0.27386221289634705,
      "learning_rate": 2.470030568897938e-05,
      "loss": 6.8381,
      "step": 111
    },
    {
      "epoch": 2.0934579439252334,
      "grad_norm": 0.2707161009311676,
      "learning_rate": 2.3808580878919946e-05,
      "loss": 6.8459,
      "step": 112
    },
    {
      "epoch": 2.1121495327102804,
      "grad_norm": 0.27950215339660645,
      "learning_rate": 2.2928192835717644e-05,
      "loss": 6.8356,
      "step": 113
    },
    {
      "epoch": 2.130841121495327,
      "grad_norm": 0.3068324327468872,
      "learning_rate": 2.205952262898704e-05,
      "loss": 6.8478,
      "step": 114
    },
    {
      "epoch": 2.149532710280374,
      "grad_norm": 0.18191057443618774,
      "learning_rate": 2.120294625636171e-05,
      "loss": 6.8803,
      "step": 115
    },
    {
      "epoch": 2.1682242990654204,
      "grad_norm": 0.18903179466724396,
      "learning_rate": 2.0358834480746365e-05,
      "loss": 6.8878,
      "step": 116
    },
    {
      "epoch": 2.1869158878504673,
      "grad_norm": 0.1874402016401291,
      "learning_rate": 1.9527552669834798e-05,
      "loss": 6.8868,
      "step": 117
    },
    {
      "epoch": 2.205607476635514,
      "grad_norm": 0.18762698769569397,
      "learning_rate": 1.8709460637963123e-05,
      "loss": 6.8844,
      "step": 118
    },
    {
      "epoch": 2.2242990654205608,
      "grad_norm": 0.16527535021305084,
      "learning_rate": 1.790491249036672e-05,
      "loss": 6.8865,
      "step": 119
    },
    {
      "epoch": 2.2429906542056073,
      "grad_norm": 0.20380131900310516,
      "learning_rate": 1.711425646990838e-05,
      "loss": 6.8812,
      "step": 120
    },
    {
      "epoch": 2.2616822429906542,
      "grad_norm": 0.34213146567344666,
      "learning_rate": 1.6337834806343783e-05,
      "loss": 6.8717,
      "step": 121
    },
    {
      "epoch": 2.2803738317757007,
      "grad_norm": 0.28527557849884033,
      "learning_rate": 1.557598356819e-05,
      "loss": 6.8607,
      "step": 122
    },
    {
      "epoch": 2.2990654205607477,
      "grad_norm": 0.24322444200515747,
      "learning_rate": 1.4829032517260489e-05,
      "loss": 6.85,
      "step": 123
    },
    {
      "epoch": 2.317757009345794,
      "grad_norm": 0.3315684199333191,
      "learning_rate": 1.4097304965930158e-05,
      "loss": 6.8333,
      "step": 124
    },
    {
      "epoch": 2.336448598130841,
      "grad_norm": 0.29074937105178833,
      "learning_rate": 1.3381117637191886e-05,
      "loss": 6.8366,
      "step": 125
    },
    {
      "epoch": 2.3551401869158877,
      "grad_norm": 0.2741944193840027,
      "learning_rate": 1.268078052756531e-05,
      "loss": 6.8362,
      "step": 126
    },
    {
      "epoch": 2.3738317757009346,
      "grad_norm": 0.31636983156204224,
      "learning_rate": 1.199659677291709e-05,
      "loss": 6.8514,
      "step": 127
    },
    {
      "epoch": 2.392523364485981,
      "grad_norm": 0.1865837574005127,
      "learning_rate": 1.132886251725061e-05,
      "loss": 6.8833,
      "step": 128
    },
    {
      "epoch": 2.411214953271028,
      "grad_norm": 0.16030381619930267,
      "learning_rate": 1.0677866784522317e-05,
      "loss": 6.8839,
      "step": 129
    },
    {
      "epoch": 2.4299065420560746,
      "grad_norm": 0.1735234260559082,
      "learning_rate": 1.004389135353972e-05,
      "loss": 6.881,
      "step": 130
    },
    {
      "epoch": 2.4485981308411215,
      "grad_norm": 0.19315017759799957,
      "learning_rate": 9.427210635995482e-06,
      "loss": 6.8902,
      "step": 131
    },
    {
      "epoch": 2.467289719626168,
      "grad_norm": 0.20848211646080017,
      "learning_rate": 8.828091557690289e-06,
      "loss": 6.8857,
      "step": 132
    },
    {
      "epoch": 2.485981308411215,
      "grad_norm": 0.18371738493442535,
      "learning_rate": 8.246793442995954e-06,
      "loss": 6.882,
      "step": 133
    },
    {
      "epoch": 2.5046728971962615,
      "grad_norm": 0.28223395347595215,
      "learning_rate": 7.683567902608729e-06,
      "loss": 6.858,
      "step": 134
    },
    {
      "epoch": 2.5233644859813085,
      "grad_norm": 0.2955135703086853,
      "learning_rate": 7.138658724641417e-06,
      "loss": 6.8461,
      "step": 135
    },
    {
      "epoch": 2.542056074766355,
      "grad_norm": 0.26921555399894714,
      "learning_rate": 6.6123017691014645e-06,
      "loss": 6.8326,
      "step": 136
    },
    {
      "epoch": 2.560747663551402,
      "grad_norm": 0.29341185092926025,
      "learning_rate": 6.104724865800665e-06,
      "loss": 6.8405,
      "step": 137
    },
    {
      "epoch": 2.5794392523364484,
      "grad_norm": 0.2668432593345642,
      "learning_rate": 5.616147715740611e-06,
      "loss": 6.8386,
      "step": 138
    },
    {
      "epoch": 2.5981308411214954,
      "grad_norm": 0.29058679938316345,
      "learning_rate": 5.1467817960167975e-06,
      "loss": 6.8378,
      "step": 139
    },
    {
      "epoch": 2.616822429906542,
      "grad_norm": 0.24260707199573517,
      "learning_rate": 4.696830268282204e-06,
      "loss": 6.8482,
      "step": 140
    },
    {
      "epoch": 2.635514018691589,
      "grad_norm": 0.17562568187713623,
      "learning_rate": 4.266487890810256e-06,
      "loss": 6.865,
      "step": 141
    },
    {
      "epoch": 2.6542056074766354,
      "grad_norm": 0.21645016968250275,
      "learning_rate": 3.855940934195146e-06,
      "loss": 6.8823,
      "step": 142
    },
    {
      "epoch": 2.6728971962616823,
      "grad_norm": 0.17004626989364624,
      "learning_rate": 3.465367100725908e-06,
      "loss": 6.8824,
      "step": 143
    },
    {
      "epoch": 2.691588785046729,
      "grad_norm": 0.1897551566362381,
      "learning_rate": 3.0949354474692937e-06,
      "loss": 6.8868,
      "step": 144
    },
    {
      "epoch": 2.710280373831776,
      "grad_norm": 0.1884983330965042,
      "learning_rate": 2.7448063130946224e-06,
      "loss": 6.885,
      "step": 145
    },
    {
      "epoch": 2.7289719626168223,
      "grad_norm": 0.20524746179580688,
      "learning_rate": 2.4151312484723465e-06,
      "loss": 6.8843,
      "step": 146
    },
    {
      "epoch": 2.7476635514018692,
      "grad_norm": 0.22424666583538055,
      "learning_rate": 2.106052951076365e-06,
      "loss": 6.8523,
      "step": 147
    },
    {
      "epoch": 2.7663551401869158,
      "grad_norm": 0.23577845096588135,
      "learning_rate": 1.8177052032184283e-06,
      "loss": 6.8535,
      "step": 148
    },
    {
      "epoch": 2.7850467289719627,
      "grad_norm": 0.2570821940898895,
      "learning_rate": 1.5502128141414495e-06,
      "loss": 6.8354,
      "step": 149
    },
    {
      "epoch": 2.803738317757009,
      "grad_norm": 0.2860844135284424,
      "learning_rate": 1.303691565996712e-06,
      "loss": 6.8349,
      "step": 150
    },
    {
      "epoch": 2.803738317757009,
      "eval_loss": 6.854687690734863,
      "eval_runtime": 0.1264,
      "eval_samples_per_second": 711.858,
      "eval_steps_per_second": 181.919,
      "step": 150
    },
    {
      "epoch": 2.822429906542056,
      "grad_norm": 0.27185508608818054,
      "learning_rate": 1.0782481637284013e-06,
      "loss": 6.842,
      "step": 151
    },
    {
      "epoch": 2.8411214953271027,
      "grad_norm": 0.34425657987594604,
      "learning_rate": 8.739801888871469e-07,
      "loss": 6.8271,
      "step": 152
    },
    {
      "epoch": 2.8598130841121496,
      "grad_norm": 0.30129221081733704,
      "learning_rate": 6.909760573925561e-07,
      "loss": 6.8476,
      "step": 153
    },
    {
      "epoch": 2.878504672897196,
      "grad_norm": 0.19908937811851501,
      "learning_rate": 5.29314981262985e-07,
      "loss": 6.8637,
      "step": 154
    },
    {
      "epoch": 2.897196261682243,
      "grad_norm": 0.183157280087471,
      "learning_rate": 3.8906693432924634e-07,
      "loss": 6.8853,
      "step": 155
    },
    {
      "epoch": 2.9158878504672896,
      "grad_norm": 0.17316141724586487,
      "learning_rate": 2.702926219468882e-07,
      "loss": 6.8954,
      "step": 156
    },
    {
      "epoch": 2.9345794392523366,
      "grad_norm": 0.18594498932361603,
      "learning_rate": 1.7304345472035631e-07,
      "loss": 6.8887,
      "step": 157
    },
    {
      "epoch": 2.953271028037383,
      "grad_norm": 0.16382110118865967,
      "learning_rate": 9.73615262502503e-08,
      "loss": 6.8816,
      "step": 158
    },
    {
      "epoch": 2.97196261682243,
      "grad_norm": 0.2256663292646408,
      "learning_rate": 4.3279594913447905e-08,
      "loss": 6.8817,
      "step": 159
    },
    {
      "epoch": 2.9906542056074765,
      "grad_norm": 0.2136058360338211,
      "learning_rate": 1.082106968385288e-08,
      "loss": 6.8499,
      "step": 160
    },
    {
      "epoch": 3.0093457943925235,
      "grad_norm": 0.29850977659225464,
      "learning_rate": 0.0,
      "loss": 10.9924,
      "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": 3091228053504.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}