{
  "best_metric": 0.2916959226131439,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.36199095022624433,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0018099547511312218,
      "grad_norm": 0.18614302575588226,
      "learning_rate": 1e-05,
      "loss": 0.2229,
      "step": 1
    },
    {
      "epoch": 0.0018099547511312218,
      "eval_loss": 0.41589033603668213,
      "eval_runtime": 39.09,
      "eval_samples_per_second": 23.817,
      "eval_steps_per_second": 5.961,
      "step": 1
    },
    {
      "epoch": 0.0036199095022624436,
      "grad_norm": 0.20625001192092896,
      "learning_rate": 2e-05,
      "loss": 0.2624,
      "step": 2
    },
    {
      "epoch": 0.005429864253393665,
      "grad_norm": 0.2611289322376251,
      "learning_rate": 3e-05,
      "loss": 0.3303,
      "step": 3
    },
    {
      "epoch": 0.007239819004524887,
      "grad_norm": 0.2935875654220581,
      "learning_rate": 4e-05,
      "loss": 0.321,
      "step": 4
    },
    {
      "epoch": 0.00904977375565611,
      "grad_norm": 0.22957083582878113,
      "learning_rate": 5e-05,
      "loss": 0.3686,
      "step": 5
    },
    {
      "epoch": 0.01085972850678733,
      "grad_norm": 0.23430617153644562,
      "learning_rate": 6e-05,
      "loss": 0.3144,
      "step": 6
    },
    {
      "epoch": 0.012669683257918552,
      "grad_norm": 0.2339513748884201,
      "learning_rate": 7e-05,
      "loss": 0.2904,
      "step": 7
    },
    {
      "epoch": 0.014479638009049774,
      "grad_norm": 0.24033229053020477,
      "learning_rate": 8e-05,
      "loss": 0.317,
      "step": 8
    },
    {
      "epoch": 0.016289592760180997,
      "grad_norm": 0.2350195199251175,
      "learning_rate": 9e-05,
      "loss": 0.3623,
      "step": 9
    },
    {
      "epoch": 0.01809954751131222,
      "grad_norm": 0.2262275069952011,
      "learning_rate": 0.0001,
      "loss": 0.3626,
      "step": 10
    },
    {
      "epoch": 0.019909502262443438,
      "grad_norm": 0.20852632820606232,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.3052,
      "step": 11
    },
    {
      "epoch": 0.02171945701357466,
      "grad_norm": 0.2222484052181244,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.3014,
      "step": 12
    },
    {
      "epoch": 0.023529411764705882,
      "grad_norm": 0.20699352025985718,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.3292,
      "step": 13
    },
    {
      "epoch": 0.025339366515837104,
      "grad_norm": 0.21587537229061127,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.322,
      "step": 14
    },
    {
      "epoch": 0.027149321266968326,
      "grad_norm": 0.23950739204883575,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.3516,
      "step": 15
    },
    {
      "epoch": 0.02895927601809955,
      "grad_norm": 0.22088505327701569,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.3137,
      "step": 16
    },
    {
      "epoch": 0.03076923076923077,
      "grad_norm": 0.22170783579349518,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.3252,
      "step": 17
    },
    {
      "epoch": 0.03257918552036199,
      "grad_norm": 0.1999443769454956,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.2898,
      "step": 18
    },
    {
      "epoch": 0.034389140271493215,
      "grad_norm": 0.21578389406204224,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.3177,
      "step": 19
    },
    {
      "epoch": 0.03619909502262444,
      "grad_norm": 0.18774747848510742,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.3177,
      "step": 20
    },
    {
      "epoch": 0.03800904977375565,
      "grad_norm": 0.186977818608284,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.3026,
      "step": 21
    },
    {
      "epoch": 0.039819004524886875,
      "grad_norm": 0.198961079120636,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.3152,
      "step": 22
    },
    {
      "epoch": 0.0416289592760181,
      "grad_norm": 0.21934367716312408,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.3445,
      "step": 23
    },
    {
      "epoch": 0.04343891402714932,
      "grad_norm": 0.19348673522472382,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.3281,
      "step": 24
    },
    {
      "epoch": 0.04524886877828054,
      "grad_norm": 0.18486647307872772,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.2894,
      "step": 25
    },
    {
      "epoch": 0.047058823529411764,
      "grad_norm": 0.20630523562431335,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.3479,
      "step": 26
    },
    {
      "epoch": 0.048868778280542986,
      "grad_norm": 0.19784313440322876,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.3445,
      "step": 27
    },
    {
      "epoch": 0.05067873303167421,
      "grad_norm": 0.20148912072181702,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.3444,
      "step": 28
    },
    {
      "epoch": 0.05248868778280543,
      "grad_norm": 0.20969101786613464,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.3758,
      "step": 29
    },
    {
      "epoch": 0.05429864253393665,
      "grad_norm": 0.2041962891817093,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.335,
      "step": 30
    },
    {
      "epoch": 0.056108597285067875,
      "grad_norm": 0.21553747355937958,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.3569,
      "step": 31
    },
    {
      "epoch": 0.0579185520361991,
      "grad_norm": 0.2132917046546936,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.4062,
      "step": 32
    },
    {
      "epoch": 0.05972850678733032,
      "grad_norm": 0.2178078293800354,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.3536,
      "step": 33
    },
    {
      "epoch": 0.06153846153846154,
      "grad_norm": 0.21165508031845093,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.381,
      "step": 34
    },
    {
      "epoch": 0.06334841628959276,
      "grad_norm": 0.20810887217521667,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.3541,
      "step": 35
    },
    {
      "epoch": 0.06515837104072399,
      "grad_norm": 0.20564156770706177,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.3269,
      "step": 36
    },
    {
      "epoch": 0.0669683257918552,
      "grad_norm": 0.22891999781131744,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.3486,
      "step": 37
    },
    {
      "epoch": 0.06877828054298643,
      "grad_norm": 0.220924973487854,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.3653,
      "step": 38
    },
    {
      "epoch": 0.07058823529411765,
      "grad_norm": 0.2316446453332901,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.3627,
      "step": 39
    },
    {
      "epoch": 0.07239819004524888,
      "grad_norm": 0.2408294528722763,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.3627,
      "step": 40
    },
    {
      "epoch": 0.07420814479638009,
      "grad_norm": 0.2520692050457001,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.3423,
      "step": 41
    },
    {
      "epoch": 0.0760180995475113,
      "grad_norm": 0.24204710125923157,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.36,
      "step": 42
    },
    {
      "epoch": 0.07782805429864253,
      "grad_norm": 0.2555690407752991,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.3473,
      "step": 43
    },
    {
      "epoch": 0.07963800904977375,
      "grad_norm": 0.2731051743030548,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.4025,
      "step": 44
    },
    {
      "epoch": 0.08144796380090498,
      "grad_norm": 0.2501581907272339,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.2815,
      "step": 45
    },
    {
      "epoch": 0.0832579185520362,
      "grad_norm": 0.26206544041633606,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.3072,
      "step": 46
    },
    {
      "epoch": 0.08506787330316742,
      "grad_norm": 0.285060316324234,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.3257,
      "step": 47
    },
    {
      "epoch": 0.08687782805429864,
      "grad_norm": 0.26404353976249695,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.282,
      "step": 48
    },
    {
      "epoch": 0.08868778280542987,
      "grad_norm": 0.34065139293670654,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.3855,
      "step": 49
    },
    {
      "epoch": 0.09049773755656108,
      "grad_norm": 0.3714425563812256,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.321,
      "step": 50
    },
    {
      "epoch": 0.09049773755656108,
      "eval_loss": 0.31067702174186707,
      "eval_runtime": 39.3466,
      "eval_samples_per_second": 23.662,
      "eval_steps_per_second": 5.922,
      "step": 50
    },
    {
      "epoch": 0.09230769230769231,
      "grad_norm": 0.13869038224220276,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.2337,
      "step": 51
    },
    {
      "epoch": 0.09411764705882353,
      "grad_norm": 0.14860887825489044,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.262,
      "step": 52
    },
    {
      "epoch": 0.09592760180995476,
      "grad_norm": 0.14812035858631134,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.2606,
      "step": 53
    },
    {
      "epoch": 0.09773755656108597,
      "grad_norm": 0.12683004140853882,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.2267,
      "step": 54
    },
    {
      "epoch": 0.09954751131221719,
      "grad_norm": 0.14570578932762146,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.2887,
      "step": 55
    },
    {
      "epoch": 0.10135746606334842,
      "grad_norm": 0.1383831799030304,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.2361,
      "step": 56
    },
    {
      "epoch": 0.10316742081447963,
      "grad_norm": 0.13729773461818695,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.2325,
      "step": 57
    },
    {
      "epoch": 0.10497737556561086,
      "grad_norm": 0.15586471557617188,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.291,
      "step": 58
    },
    {
      "epoch": 0.10678733031674208,
      "grad_norm": 0.1725718230009079,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3337,
      "step": 59
    },
    {
      "epoch": 0.1085972850678733,
      "grad_norm": 0.16073912382125854,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.3078,
      "step": 60
    },
    {
      "epoch": 0.11040723981900452,
      "grad_norm": 0.16522738337516785,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.2984,
      "step": 61
    },
    {
      "epoch": 0.11221719457013575,
      "grad_norm": 0.17444974184036255,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.3126,
      "step": 62
    },
    {
      "epoch": 0.11402714932126697,
      "grad_norm": 0.17596791684627533,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.3073,
      "step": 63
    },
    {
      "epoch": 0.1158371040723982,
      "grad_norm": 0.17301933467388153,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.2792,
      "step": 64
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.1702631264925003,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.2701,
      "step": 65
    },
    {
      "epoch": 0.11945701357466064,
      "grad_norm": 0.19698326289653778,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.3594,
      "step": 66
    },
    {
      "epoch": 0.12126696832579185,
      "grad_norm": 0.17198459804058075,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.2696,
      "step": 67
    },
    {
      "epoch": 0.12307692307692308,
      "grad_norm": 0.1675131916999817,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3124,
      "step": 68
    },
    {
      "epoch": 0.1248868778280543,
      "grad_norm": 0.16875603795051575,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.2891,
      "step": 69
    },
    {
      "epoch": 0.12669683257918551,
      "grad_norm": 0.1794370412826538,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.3092,
      "step": 70
    },
    {
      "epoch": 0.12850678733031673,
      "grad_norm": 0.17048543691635132,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.3042,
      "step": 71
    },
    {
      "epoch": 0.13031674208144797,
      "grad_norm": 0.18122081458568573,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.3003,
      "step": 72
    },
    {
      "epoch": 0.1321266968325792,
      "grad_norm": 0.16847367584705353,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.301,
      "step": 73
    },
    {
      "epoch": 0.1339366515837104,
      "grad_norm": 0.17657122015953064,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.3595,
      "step": 74
    },
    {
      "epoch": 0.13574660633484162,
      "grad_norm": 0.17168860137462616,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.3214,
      "step": 75
    },
    {
      "epoch": 0.13755656108597286,
      "grad_norm": 0.17688295245170593,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.3285,
      "step": 76
    },
    {
      "epoch": 0.13936651583710408,
      "grad_norm": 0.1758047193288803,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.3468,
      "step": 77
    },
    {
      "epoch": 0.1411764705882353,
      "grad_norm": 0.17167751491069794,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.3268,
      "step": 78
    },
    {
      "epoch": 0.1429864253393665,
      "grad_norm": 0.19024083018302917,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.3589,
      "step": 79
    },
    {
      "epoch": 0.14479638009049775,
      "grad_norm": 0.19968771934509277,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.3881,
      "step": 80
    },
    {
      "epoch": 0.14660633484162897,
      "grad_norm": 0.19565021991729736,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.3708,
      "step": 81
    },
    {
      "epoch": 0.14841628959276018,
      "grad_norm": 0.18866311013698578,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.3329,
      "step": 82
    },
    {
      "epoch": 0.1502262443438914,
      "grad_norm": 0.19271686673164368,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.3591,
      "step": 83
    },
    {
      "epoch": 0.1520361990950226,
      "grad_norm": 0.19825728237628937,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.3173,
      "step": 84
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 0.2054145485162735,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.36,
      "step": 85
    },
    {
      "epoch": 0.15565610859728507,
      "grad_norm": 0.21366769075393677,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.3504,
      "step": 86
    },
    {
      "epoch": 0.15746606334841629,
      "grad_norm": 0.21033622324466705,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.316,
      "step": 87
    },
    {
      "epoch": 0.1592760180995475,
      "grad_norm": 0.20696823298931122,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3307,
      "step": 88
    },
    {
      "epoch": 0.16108597285067874,
      "grad_norm": 0.22416459023952484,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.3288,
      "step": 89
    },
    {
      "epoch": 0.16289592760180996,
      "grad_norm": 0.21869608759880066,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.3377,
      "step": 90
    },
    {
      "epoch": 0.16470588235294117,
      "grad_norm": 0.2249254584312439,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.3126,
      "step": 91
    },
    {
      "epoch": 0.1665158371040724,
      "grad_norm": 0.2334333211183548,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.3421,
      "step": 92
    },
    {
      "epoch": 0.16832579185520363,
      "grad_norm": 0.2551923990249634,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.3584,
      "step": 93
    },
    {
      "epoch": 0.17013574660633485,
      "grad_norm": 0.23733778297901154,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3061,
      "step": 94
    },
    {
      "epoch": 0.17194570135746606,
      "grad_norm": 0.2562201917171478,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.3073,
      "step": 95
    },
    {
      "epoch": 0.17375565610859728,
      "grad_norm": 0.26217159628868103,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.2965,
      "step": 96
    },
    {
      "epoch": 0.1755656108597285,
      "grad_norm": 0.27308839559555054,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.3463,
      "step": 97
    },
    {
      "epoch": 0.17737556561085974,
      "grad_norm": 0.30020132660865784,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3409,
      "step": 98
    },
    {
      "epoch": 0.17918552036199095,
      "grad_norm": 0.2935176491737366,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.2832,
      "step": 99
    },
    {
      "epoch": 0.18099547511312217,
      "grad_norm": 0.3104833662509918,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.2941,
      "step": 100
    },
    {
      "epoch": 0.18099547511312217,
      "eval_loss": 0.30154144763946533,
      "eval_runtime": 39.3437,
      "eval_samples_per_second": 23.663,
      "eval_steps_per_second": 5.922,
      "step": 100
    },
    {
      "epoch": 0.18280542986425338,
      "grad_norm": 0.13306333124637604,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.2146,
      "step": 101
    },
    {
      "epoch": 0.18461538461538463,
      "grad_norm": 0.14036741852760315,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.2436,
      "step": 102
    },
    {
      "epoch": 0.18642533936651584,
      "grad_norm": 0.12967729568481445,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.2253,
      "step": 103
    },
    {
      "epoch": 0.18823529411764706,
      "grad_norm": 0.14258146286010742,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.2665,
      "step": 104
    },
    {
      "epoch": 0.19004524886877827,
      "grad_norm": 0.14999476075172424,
      "learning_rate": 5e-05,
      "loss": 0.2834,
      "step": 105
    },
    {
      "epoch": 0.19185520361990951,
      "grad_norm": 0.13652700185775757,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.2496,
      "step": 106
    },
    {
      "epoch": 0.19366515837104073,
      "grad_norm": 0.14443832635879517,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.2412,
      "step": 107
    },
    {
      "epoch": 0.19547511312217195,
      "grad_norm": 0.1632201075553894,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.3317,
      "step": 108
    },
    {
      "epoch": 0.19728506787330316,
      "grad_norm": 0.14898152649402618,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.2651,
      "step": 109
    },
    {
      "epoch": 0.19909502262443438,
      "grad_norm": 0.16227401793003082,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3034,
      "step": 110
    },
    {
      "epoch": 0.20090497737556562,
      "grad_norm": 0.15319198369979858,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2623,
      "step": 111
    },
    {
      "epoch": 0.20271493212669683,
      "grad_norm": 0.1812855303287506,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.3117,
      "step": 112
    },
    {
      "epoch": 0.20452488687782805,
      "grad_norm": 0.1512429565191269,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.2569,
      "step": 113
    },
    {
      "epoch": 0.20633484162895926,
      "grad_norm": 0.17571544647216797,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3149,
      "step": 114
    },
    {
      "epoch": 0.2081447963800905,
      "grad_norm": 0.16053655743598938,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.2678,
      "step": 115
    },
    {
      "epoch": 0.20995475113122172,
      "grad_norm": 0.17989246547222137,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3223,
      "step": 116
    },
    {
      "epoch": 0.21176470588235294,
      "grad_norm": 0.16377779841423035,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.2826,
      "step": 117
    },
    {
      "epoch": 0.21357466063348415,
      "grad_norm": 0.16705048084259033,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.2527,
      "step": 118
    },
    {
      "epoch": 0.2153846153846154,
      "grad_norm": 0.16415028274059296,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.3103,
      "step": 119
    },
    {
      "epoch": 0.2171945701357466,
      "grad_norm": 0.16321642696857452,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.2818,
      "step": 120
    },
    {
      "epoch": 0.21900452488687783,
      "grad_norm": 0.1582905501127243,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.2723,
      "step": 121
    },
    {
      "epoch": 0.22081447963800904,
      "grad_norm": 0.17084002494812012,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.2823,
      "step": 122
    },
    {
      "epoch": 0.22262443438914026,
      "grad_norm": 0.15782323479652405,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.2679,
      "step": 123
    },
    {
      "epoch": 0.2244343891402715,
      "grad_norm": 0.17926311492919922,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.3344,
      "step": 124
    },
    {
      "epoch": 0.22624434389140272,
      "grad_norm": 0.17336390912532806,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.312,
      "step": 125
    },
    {
      "epoch": 0.22805429864253393,
      "grad_norm": 0.18395105004310608,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.3171,
      "step": 126
    },
    {
      "epoch": 0.22986425339366515,
      "grad_norm": 0.1807699352502823,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3279,
      "step": 127
    },
    {
      "epoch": 0.2316742081447964,
      "grad_norm": 0.1802460253238678,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.3436,
      "step": 128
    },
    {
      "epoch": 0.2334841628959276,
      "grad_norm": 0.1808416098356247,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.3456,
      "step": 129
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.18347255885601044,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.3289,
      "step": 130
    },
    {
      "epoch": 0.23710407239819004,
      "grad_norm": 0.1987457126379013,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.347,
      "step": 131
    },
    {
      "epoch": 0.23891402714932128,
      "grad_norm": 0.19039824604988098,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.3851,
      "step": 132
    },
    {
      "epoch": 0.2407239819004525,
      "grad_norm": 0.18834824860095978,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.3245,
      "step": 133
    },
    {
      "epoch": 0.2425339366515837,
      "grad_norm": 0.19769467413425446,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.347,
      "step": 134
    },
    {
      "epoch": 0.24434389140271492,
      "grad_norm": 0.20547893643379211,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3618,
      "step": 135
    },
    {
      "epoch": 0.24615384615384617,
      "grad_norm": 0.19622278213500977,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.3333,
      "step": 136
    },
    {
      "epoch": 0.24796380090497738,
      "grad_norm": 0.20859295129776,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.3261,
      "step": 137
    },
    {
      "epoch": 0.2497737556561086,
      "grad_norm": 0.2199597805738449,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.3564,
      "step": 138
    },
    {
      "epoch": 0.2515837104072398,
      "grad_norm": 0.2064392864704132,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3223,
      "step": 139
    },
    {
      "epoch": 0.25339366515837103,
      "grad_norm": 0.20289361476898193,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.2946,
      "step": 140
    },
    {
      "epoch": 0.25520361990950224,
      "grad_norm": 0.20882990956306458,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.2935,
      "step": 141
    },
    {
      "epoch": 0.25701357466063346,
      "grad_norm": 0.22584840655326843,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2945,
      "step": 142
    },
    {
      "epoch": 0.25882352941176473,
      "grad_norm": 0.2418498694896698,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.3246,
      "step": 143
    },
    {
      "epoch": 0.26063348416289595,
      "grad_norm": 0.24941790103912354,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.3562,
      "step": 144
    },
    {
      "epoch": 0.26244343891402716,
      "grad_norm": 0.2624957263469696,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.3686,
      "step": 145
    },
    {
      "epoch": 0.2642533936651584,
      "grad_norm": 0.2426348775625229,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.2794,
      "step": 146
    },
    {
      "epoch": 0.2660633484162896,
      "grad_norm": 0.2559041976928711,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.305,
      "step": 147
    },
    {
      "epoch": 0.2678733031674208,
      "grad_norm": 0.2829945981502533,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.3127,
      "step": 148
    },
    {
      "epoch": 0.269683257918552,
      "grad_norm": 0.282668799161911,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3238,
      "step": 149
    },
    {
      "epoch": 0.27149321266968324,
      "grad_norm": 0.31640326976776123,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.285,
      "step": 150
    },
    {
      "epoch": 0.27149321266968324,
      "eval_loss": 0.2943105399608612,
      "eval_runtime": 39.2713,
      "eval_samples_per_second": 23.707,
      "eval_steps_per_second": 5.933,
      "step": 150
    },
    {
      "epoch": 0.27330316742081445,
      "grad_norm": 0.10948589444160461,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.184,
      "step": 151
    },
    {
      "epoch": 0.2751131221719457,
      "grad_norm": 0.11907648295164108,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.2088,
      "step": 152
    },
    {
      "epoch": 0.27692307692307694,
      "grad_norm": 0.11452509462833405,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.2016,
      "step": 153
    },
    {
      "epoch": 0.27873303167420815,
      "grad_norm": 0.13300038874149323,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.2568,
      "step": 154
    },
    {
      "epoch": 0.28054298642533937,
      "grad_norm": 0.14113478362560272,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.2649,
      "step": 155
    },
    {
      "epoch": 0.2823529411764706,
      "grad_norm": 0.14260727167129517,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.2602,
      "step": 156
    },
    {
      "epoch": 0.2841628959276018,
      "grad_norm": 0.15744712948799133,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.2812,
      "step": 157
    },
    {
      "epoch": 0.285972850678733,
      "grad_norm": 0.1739368885755539,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.3274,
      "step": 158
    },
    {
      "epoch": 0.28778280542986423,
      "grad_norm": 0.16682900488376617,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.3011,
      "step": 159
    },
    {
      "epoch": 0.2895927601809955,
      "grad_norm": 0.16134510934352875,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.2991,
      "step": 160
    },
    {
      "epoch": 0.2914027149321267,
      "grad_norm": 0.1617824286222458,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.2845,
      "step": 161
    },
    {
      "epoch": 0.29321266968325793,
      "grad_norm": 0.16519419848918915,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.2796,
      "step": 162
    },
    {
      "epoch": 0.29502262443438915,
      "grad_norm": 0.1758502572774887,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.3036,
      "step": 163
    },
    {
      "epoch": 0.29683257918552036,
      "grad_norm": 0.19032685458660126,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.3232,
      "step": 164
    },
    {
      "epoch": 0.2986425339366516,
      "grad_norm": 0.15693630278110504,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.2423,
      "step": 165
    },
    {
      "epoch": 0.3004524886877828,
      "grad_norm": 0.16963694989681244,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.2737,
      "step": 166
    },
    {
      "epoch": 0.302262443438914,
      "grad_norm": 0.16014021635055542,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.2459,
      "step": 167
    },
    {
      "epoch": 0.3040723981900452,
      "grad_norm": 0.1551578789949417,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.2564,
      "step": 168
    },
    {
      "epoch": 0.3058823529411765,
      "grad_norm": 0.1587299406528473,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2976,
      "step": 169
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 0.1582556813955307,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.2714,
      "step": 170
    },
    {
      "epoch": 0.3095022624434389,
      "grad_norm": 0.16579100489616394,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.2841,
      "step": 171
    },
    {
      "epoch": 0.31131221719457014,
      "grad_norm": 0.16417637467384338,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.2978,
      "step": 172
    },
    {
      "epoch": 0.31312217194570136,
      "grad_norm": 0.16473717987537384,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.315,
      "step": 173
    },
    {
      "epoch": 0.31493212669683257,
      "grad_norm": 0.17040620744228363,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.3028,
      "step": 174
    },
    {
      "epoch": 0.3167420814479638,
      "grad_norm": 0.17260462045669556,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3251,
      "step": 175
    },
    {
      "epoch": 0.318552036199095,
      "grad_norm": 0.18109650909900665,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.3791,
      "step": 176
    },
    {
      "epoch": 0.32036199095022627,
      "grad_norm": 0.1642480343580246,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.3046,
      "step": 177
    },
    {
      "epoch": 0.3221719457013575,
      "grad_norm": 0.1727900356054306,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.3137,
      "step": 178
    },
    {
      "epoch": 0.3239819004524887,
      "grad_norm": 0.17912720143795013,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.3327,
      "step": 179
    },
    {
      "epoch": 0.3257918552036199,
      "grad_norm": 0.18002280592918396,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.3321,
      "step": 180
    },
    {
      "epoch": 0.32760180995475113,
      "grad_norm": 0.1779184192419052,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.3214,
      "step": 181
    },
    {
      "epoch": 0.32941176470588235,
      "grad_norm": 0.19277767837047577,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.3576,
      "step": 182
    },
    {
      "epoch": 0.33122171945701356,
      "grad_norm": 0.19387736916542053,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.3654,
      "step": 183
    },
    {
      "epoch": 0.3330316742081448,
      "grad_norm": 0.19607076048851013,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.3507,
      "step": 184
    },
    {
      "epoch": 0.334841628959276,
      "grad_norm": 0.1904783695936203,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.335,
      "step": 185
    },
    {
      "epoch": 0.33665158371040727,
      "grad_norm": 0.2086869776248932,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.3394,
      "step": 186
    },
    {
      "epoch": 0.3384615384615385,
      "grad_norm": 0.21611140668392181,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.3647,
      "step": 187
    },
    {
      "epoch": 0.3402714932126697,
      "grad_norm": 0.20826615393161774,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.3631,
      "step": 188
    },
    {
      "epoch": 0.3420814479638009,
      "grad_norm": 0.2229364812374115,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.3253,
      "step": 189
    },
    {
      "epoch": 0.3438914027149321,
      "grad_norm": 0.22350913286209106,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.3739,
      "step": 190
    },
    {
      "epoch": 0.34570135746606334,
      "grad_norm": 0.21008704602718353,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2811,
      "step": 191
    },
    {
      "epoch": 0.34751131221719456,
      "grad_norm": 0.22699537873268127,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.3241,
      "step": 192
    },
    {
      "epoch": 0.34932126696832577,
      "grad_norm": 0.23869360983371735,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.331,
      "step": 193
    },
    {
      "epoch": 0.351131221719457,
      "grad_norm": 0.24357366561889648,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.3504,
      "step": 194
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.24194666743278503,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.3329,
      "step": 195
    },
    {
      "epoch": 0.3547511312217195,
      "grad_norm": 0.25053972005844116,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.292,
      "step": 196
    },
    {
      "epoch": 0.3565610859728507,
      "grad_norm": 0.28680533170700073,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.3673,
      "step": 197
    },
    {
      "epoch": 0.3583710407239819,
      "grad_norm": 0.2748851478099823,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.3103,
      "step": 198
    },
    {
      "epoch": 0.3601809954751131,
      "grad_norm": 0.2849353551864624,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2762,
      "step": 199
    },
    {
      "epoch": 0.36199095022624433,
      "grad_norm": 0.35630470514297485,
      "learning_rate": 0.0,
      "loss": 0.3162,
      "step": 200
    },
    {
      "epoch": 0.36199095022624433,
      "eval_loss": 0.2916959226131439,
      "eval_runtime": 39.2323,
      "eval_samples_per_second": 23.73,
      "eval_steps_per_second": 5.939,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.864982716809216e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}