{
  "best_metric": 0.38390806317329407,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.061354398343431246,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003067719917171562,
      "grad_norm": 3.7495951652526855,
      "learning_rate": 7e-06,
      "loss": 1.3425,
      "step": 1
    },
    {
      "epoch": 0.0003067719917171562,
      "eval_loss": 0.5136561393737793,
      "eval_runtime": 430.5928,
      "eval_samples_per_second": 12.75,
      "eval_steps_per_second": 3.189,
      "step": 1
    },
    {
      "epoch": 0.0006135439834343124,
      "grad_norm": 4.0081987380981445,
      "learning_rate": 1.4e-05,
      "loss": 1.536,
      "step": 2
    },
    {
      "epoch": 0.0009203159751514687,
      "grad_norm": 4.015089988708496,
      "learning_rate": 2.1e-05,
      "loss": 1.6788,
      "step": 3
    },
    {
      "epoch": 0.0012270879668686248,
      "grad_norm": 3.7536773681640625,
      "learning_rate": 2.8e-05,
      "loss": 1.6255,
      "step": 4
    },
    {
      "epoch": 0.0015338599585857811,
      "grad_norm": 3.3928744792938232,
      "learning_rate": 3.5e-05,
      "loss": 1.8201,
      "step": 5
    },
    {
      "epoch": 0.0018406319503029374,
      "grad_norm": 3.9312448501586914,
      "learning_rate": 4.2e-05,
      "loss": 1.699,
      "step": 6
    },
    {
      "epoch": 0.0021474039420200934,
      "grad_norm": 4.036737442016602,
      "learning_rate": 4.899999999999999e-05,
      "loss": 1.893,
      "step": 7
    },
    {
      "epoch": 0.0024541759337372497,
      "grad_norm": 3.745652437210083,
      "learning_rate": 5.6e-05,
      "loss": 1.799,
      "step": 8
    },
    {
      "epoch": 0.002760947925454406,
      "grad_norm": 3.411588191986084,
      "learning_rate": 6.3e-05,
      "loss": 1.829,
      "step": 9
    },
    {
      "epoch": 0.0030677199171715622,
      "grad_norm": 3.5523102283477783,
      "learning_rate": 7e-05,
      "loss": 1.8915,
      "step": 10
    },
    {
      "epoch": 0.0033744919088887185,
      "grad_norm": 3.2218644618988037,
      "learning_rate": 6.999521567473641e-05,
      "loss": 1.936,
      "step": 11
    },
    {
      "epoch": 0.0036812639006058748,
      "grad_norm": 3.2266080379486084,
      "learning_rate": 6.998086400693241e-05,
      "loss": 1.7437,
      "step": 12
    },
    {
      "epoch": 0.003988035892323031,
      "grad_norm": 3.168818950653076,
      "learning_rate": 6.995694892019065e-05,
      "loss": 1.9077,
      "step": 13
    },
    {
      "epoch": 0.004294807884040187,
      "grad_norm": 3.1822073459625244,
      "learning_rate": 6.99234769526571e-05,
      "loss": 1.8901,
      "step": 14
    },
    {
      "epoch": 0.004601579875757343,
      "grad_norm": 3.098999500274658,
      "learning_rate": 6.988045725523343e-05,
      "loss": 1.8588,
      "step": 15
    },
    {
      "epoch": 0.004908351867474499,
      "grad_norm": 3.074695587158203,
      "learning_rate": 6.982790158907539e-05,
      "loss": 1.8403,
      "step": 16
    },
    {
      "epoch": 0.005215123859191656,
      "grad_norm": 3.289018154144287,
      "learning_rate": 6.976582432237733e-05,
      "loss": 2.0474,
      "step": 17
    },
    {
      "epoch": 0.005521895850908812,
      "grad_norm": 3.1470084190368652,
      "learning_rate": 6.969424242644413e-05,
      "loss": 1.774,
      "step": 18
    },
    {
      "epoch": 0.005828667842625968,
      "grad_norm": 3.417246103286743,
      "learning_rate": 6.961317547105138e-05,
      "loss": 1.9962,
      "step": 19
    },
    {
      "epoch": 0.0061354398343431244,
      "grad_norm": 3.2956278324127197,
      "learning_rate": 6.952264561909527e-05,
      "loss": 1.8747,
      "step": 20
    },
    {
      "epoch": 0.006442211826060281,
      "grad_norm": 3.1632611751556396,
      "learning_rate": 6.942267762053337e-05,
      "loss": 1.8501,
      "step": 21
    },
    {
      "epoch": 0.006748983817777437,
      "grad_norm": 3.133277416229248,
      "learning_rate": 6.931329880561832e-05,
      "loss": 1.9094,
      "step": 22
    },
    {
      "epoch": 0.007055755809494593,
      "grad_norm": 3.3376052379608154,
      "learning_rate": 6.919453907742597e-05,
      "loss": 2.1212,
      "step": 23
    },
    {
      "epoch": 0.0073625278012117495,
      "grad_norm": 3.125730276107788,
      "learning_rate": 6.90664309036802e-05,
      "loss": 1.8213,
      "step": 24
    },
    {
      "epoch": 0.007669299792928906,
      "grad_norm": 3.062613010406494,
      "learning_rate": 6.892900930787656e-05,
      "loss": 1.882,
      "step": 25
    },
    {
      "epoch": 0.007976071784646061,
      "grad_norm": 3.1579535007476807,
      "learning_rate": 6.87823118597072e-05,
      "loss": 1.8683,
      "step": 26
    },
    {
      "epoch": 0.008282843776363218,
      "grad_norm": 3.150268316268921,
      "learning_rate": 6.862637866478969e-05,
      "loss": 1.9691,
      "step": 27
    },
    {
      "epoch": 0.008589615768080374,
      "grad_norm": 3.1286489963531494,
      "learning_rate": 6.846125235370252e-05,
      "loss": 1.7716,
      "step": 28
    },
    {
      "epoch": 0.00889638775979753,
      "grad_norm": 3.434857130050659,
      "learning_rate": 6.828697807033038e-05,
      "loss": 1.8311,
      "step": 29
    },
    {
      "epoch": 0.009203159751514686,
      "grad_norm": 3.493105888366699,
      "learning_rate": 6.81036034595222e-05,
      "loss": 1.8715,
      "step": 30
    },
    {
      "epoch": 0.009509931743231843,
      "grad_norm": 3.204404354095459,
      "learning_rate": 6.791117865406564e-05,
      "loss": 1.635,
      "step": 31
    },
    {
      "epoch": 0.009816703734948999,
      "grad_norm": 3.2127530574798584,
      "learning_rate": 6.770975626098112e-05,
      "loss": 1.7038,
      "step": 32
    },
    {
      "epoch": 0.010123475726666156,
      "grad_norm": 3.0898277759552,
      "learning_rate": 6.749939134713974e-05,
      "loss": 1.5416,
      "step": 33
    },
    {
      "epoch": 0.010430247718383311,
      "grad_norm": 3.249939441680908,
      "learning_rate": 6.728014142420846e-05,
      "loss": 1.5933,
      "step": 34
    },
    {
      "epoch": 0.010737019710100468,
      "grad_norm": 3.3410327434539795,
      "learning_rate": 6.7052066432927e-05,
      "loss": 1.545,
      "step": 35
    },
    {
      "epoch": 0.011043791701817624,
      "grad_norm": 2.9800238609313965,
      "learning_rate": 6.681522872672069e-05,
      "loss": 1.4148,
      "step": 36
    },
    {
      "epoch": 0.011350563693534781,
      "grad_norm": 3.2996833324432373,
      "learning_rate": 6.656969305465356e-05,
      "loss": 1.4536,
      "step": 37
    },
    {
      "epoch": 0.011657335685251936,
      "grad_norm": 3.3368849754333496,
      "learning_rate": 6.631552654372672e-05,
      "loss": 1.5,
      "step": 38
    },
    {
      "epoch": 0.011964107676969093,
      "grad_norm": 3.478151321411133,
      "learning_rate": 6.60527986805264e-05,
      "loss": 1.5469,
      "step": 39
    },
    {
      "epoch": 0.012270879668686249,
      "grad_norm": 3.412599802017212,
      "learning_rate": 6.578158129222711e-05,
      "loss": 1.4183,
      "step": 40
    },
    {
      "epoch": 0.012577651660403406,
      "grad_norm": 3.7741832733154297,
      "learning_rate": 6.550194852695469e-05,
      "loss": 1.4992,
      "step": 41
    },
    {
      "epoch": 0.012884423652120561,
      "grad_norm": 3.9010705947875977,
      "learning_rate": 6.521397683351509e-05,
      "loss": 1.5144,
      "step": 42
    },
    {
      "epoch": 0.013191195643837717,
      "grad_norm": 4.069062232971191,
      "learning_rate": 6.491774494049386e-05,
      "loss": 1.4617,
      "step": 43
    },
    {
      "epoch": 0.013497967635554874,
      "grad_norm": 4.3247785568237305,
      "learning_rate": 6.461333383473272e-05,
      "loss": 1.6182,
      "step": 44
    },
    {
      "epoch": 0.01380473962727203,
      "grad_norm": 4.433003902435303,
      "learning_rate": 6.430082673918849e-05,
      "loss": 1.3577,
      "step": 45
    },
    {
      "epoch": 0.014111511618989186,
      "grad_norm": 4.900903224945068,
      "learning_rate": 6.398030909018069e-05,
      "loss": 1.6088,
      "step": 46
    },
    {
      "epoch": 0.014418283610706342,
      "grad_norm": 5.257229804992676,
      "learning_rate": 6.365186851403423e-05,
      "loss": 1.5353,
      "step": 47
    },
    {
      "epoch": 0.014725055602423499,
      "grad_norm": 5.429163932800293,
      "learning_rate": 6.331559480312315e-05,
      "loss": 1.322,
      "step": 48
    },
    {
      "epoch": 0.015031827594140654,
      "grad_norm": 7.3463053703308105,
      "learning_rate": 6.297157989132236e-05,
      "loss": 1.9904,
      "step": 49
    },
    {
      "epoch": 0.015338599585857812,
      "grad_norm": 22.010234832763672,
      "learning_rate": 6.261991782887377e-05,
      "loss": 1.6099,
      "step": 50
    },
    {
      "epoch": 0.015338599585857812,
      "eval_loss": 0.45144757628440857,
      "eval_runtime": 433.9082,
      "eval_samples_per_second": 12.652,
      "eval_steps_per_second": 3.164,
      "step": 50
    },
    {
      "epoch": 0.01564537157757497,
      "grad_norm": 2.9910173416137695,
      "learning_rate": 6.226070475667393e-05,
      "loss": 1.1536,
      "step": 51
    },
    {
      "epoch": 0.015952143569292122,
      "grad_norm": 2.9567184448242188,
      "learning_rate": 6.189403887999006e-05,
      "loss": 1.4466,
      "step": 52
    },
    {
      "epoch": 0.01625891556100928,
      "grad_norm": 2.5240225791931152,
      "learning_rate": 6.152002044161171e-05,
      "loss": 1.3853,
      "step": 53
    },
    {
      "epoch": 0.016565687552726437,
      "grad_norm": 2.495223045349121,
      "learning_rate": 6.113875169444539e-05,
      "loss": 1.5743,
      "step": 54
    },
    {
      "epoch": 0.016872459544443594,
      "grad_norm": 2.5516178607940674,
      "learning_rate": 6.0750336873559605e-05,
      "loss": 1.6522,
      "step": 55
    },
    {
      "epoch": 0.017179231536160747,
      "grad_norm": 2.406719446182251,
      "learning_rate": 6.035488216768811e-05,
      "loss": 1.7146,
      "step": 56
    },
    {
      "epoch": 0.017486003527877905,
      "grad_norm": 2.5311386585235596,
      "learning_rate": 5.9952495690198894e-05,
      "loss": 1.6618,
      "step": 57
    },
    {
      "epoch": 0.01779277551959506,
      "grad_norm": 2.566500186920166,
      "learning_rate": 5.954328744953709e-05,
      "loss": 1.659,
      "step": 58
    },
    {
      "epoch": 0.01809954751131222,
      "grad_norm": 2.463177442550659,
      "learning_rate": 5.91273693191498e-05,
      "loss": 1.6954,
      "step": 59
    },
    {
      "epoch": 0.018406319503029372,
      "grad_norm": 2.6135101318359375,
      "learning_rate": 5.870485500690094e-05,
      "loss": 1.932,
      "step": 60
    },
    {
      "epoch": 0.01871309149474653,
      "grad_norm": 2.574443817138672,
      "learning_rate": 5.827586002398468e-05,
      "loss": 1.7346,
      "step": 61
    },
    {
      "epoch": 0.019019863486463687,
      "grad_norm": 2.7927029132843018,
      "learning_rate": 5.784050165334589e-05,
      "loss": 1.8955,
      "step": 62
    },
    {
      "epoch": 0.01932663547818084,
      "grad_norm": 2.6441617012023926,
      "learning_rate": 5.739889891761608e-05,
      "loss": 1.824,
      "step": 63
    },
    {
      "epoch": 0.019633407469897998,
      "grad_norm": 2.5842678546905518,
      "learning_rate": 5.6951172546573794e-05,
      "loss": 1.9105,
      "step": 64
    },
    {
      "epoch": 0.019940179461615155,
      "grad_norm": 2.5029103755950928,
      "learning_rate": 5.6497444944138376e-05,
      "loss": 1.8022,
      "step": 65
    },
    {
      "epoch": 0.020246951453332312,
      "grad_norm": 2.5302553176879883,
      "learning_rate": 5.603784015490587e-05,
      "loss": 1.8158,
      "step": 66
    },
    {
      "epoch": 0.020553723445049465,
      "grad_norm": 2.478358507156372,
      "learning_rate": 5.557248383023655e-05,
      "loss": 1.7749,
      "step": 67
    },
    {
      "epoch": 0.020860495436766623,
      "grad_norm": 2.5742812156677246,
      "learning_rate": 5.510150319390302e-05,
      "loss": 1.8457,
      "step": 68
    },
    {
      "epoch": 0.02116726742848378,
      "grad_norm": 2.62701678276062,
      "learning_rate": 5.4625027007308546e-05,
      "loss": 1.8993,
      "step": 69
    },
    {
      "epoch": 0.021474039420200937,
      "grad_norm": 2.5461835861206055,
      "learning_rate": 5.414318553428494e-05,
      "loss": 1.779,
      "step": 70
    },
    {
      "epoch": 0.02178081141191809,
      "grad_norm": 2.6452932357788086,
      "learning_rate": 5.3656110505479776e-05,
      "loss": 1.8034,
      "step": 71
    },
    {
      "epoch": 0.022087583403635248,
      "grad_norm": 2.679868459701538,
      "learning_rate": 5.316393508234253e-05,
      "loss": 1.8786,
      "step": 72
    },
    {
      "epoch": 0.022394355395352405,
      "grad_norm": 2.657691717147827,
      "learning_rate": 5.266679382071953e-05,
      "loss": 1.7916,
      "step": 73
    },
    {
      "epoch": 0.022701127387069562,
      "grad_norm": 2.728360414505005,
      "learning_rate": 5.216482263406778e-05,
      "loss": 1.8655,
      "step": 74
    },
    {
      "epoch": 0.023007899378786716,
      "grad_norm": 2.6260228157043457,
      "learning_rate": 5.1658158756297576e-05,
      "loss": 1.7711,
      "step": 75
    },
    {
      "epoch": 0.023314671370503873,
      "grad_norm": 2.7247214317321777,
      "learning_rate": 5.114694070425407e-05,
      "loss": 1.8118,
      "step": 76
    },
    {
      "epoch": 0.02362144336222103,
      "grad_norm": 2.6050596237182617,
      "learning_rate": 5.063130823984823e-05,
      "loss": 1.6405,
      "step": 77
    },
    {
      "epoch": 0.023928215353938187,
      "grad_norm": 2.6807262897491455,
      "learning_rate": 5.011140233184724e-05,
      "loss": 1.7592,
      "step": 78
    },
    {
      "epoch": 0.02423498734565534,
      "grad_norm": 2.665363311767578,
      "learning_rate": 4.958736511733516e-05,
      "loss": 1.6636,
      "step": 79
    },
    {
      "epoch": 0.024541759337372498,
      "grad_norm": 2.8059120178222656,
      "learning_rate": 4.905933986285393e-05,
      "loss": 1.7264,
      "step": 80
    },
    {
      "epoch": 0.024848531329089655,
      "grad_norm": 2.7553577423095703,
      "learning_rate": 4.8527470925235824e-05,
      "loss": 1.6919,
      "step": 81
    },
    {
      "epoch": 0.025155303320806812,
      "grad_norm": 2.834160804748535,
      "learning_rate": 4.799190371213772e-05,
      "loss": 1.7664,
      "step": 82
    },
    {
      "epoch": 0.025462075312523966,
      "grad_norm": 2.881916046142578,
      "learning_rate": 4.745278464228808e-05,
      "loss": 1.7277,
      "step": 83
    },
    {
      "epoch": 0.025768847304241123,
      "grad_norm": 2.841578960418701,
      "learning_rate": 4.69102611054575e-05,
      "loss": 1.4881,
      "step": 84
    },
    {
      "epoch": 0.02607561929595828,
      "grad_norm": 2.90185809135437,
      "learning_rate": 4.6364481422163926e-05,
      "loss": 1.6276,
      "step": 85
    },
    {
      "epoch": 0.026382391287675434,
      "grad_norm": 3.0205485820770264,
      "learning_rate": 4.581559480312316e-05,
      "loss": 1.6431,
      "step": 86
    },
    {
      "epoch": 0.02668916327939259,
      "grad_norm": 2.986945629119873,
      "learning_rate": 4.526375130845627e-05,
      "loss": 1.4983,
      "step": 87
    },
    {
      "epoch": 0.026995935271109748,
      "grad_norm": 3.107217788696289,
      "learning_rate": 4.4709101806664554e-05,
      "loss": 1.5031,
      "step": 88
    },
    {
      "epoch": 0.027302707262826905,
      "grad_norm": 3.0583534240722656,
      "learning_rate": 4.4151797933383685e-05,
      "loss": 1.3852,
      "step": 89
    },
    {
      "epoch": 0.02760947925454406,
      "grad_norm": 3.250265598297119,
      "learning_rate": 4.359199204992797e-05,
      "loss": 1.3967,
      "step": 90
    },
    {
      "epoch": 0.027916251246261216,
      "grad_norm": 3.308723211288452,
      "learning_rate": 4.30298372016363e-05,
      "loss": 1.3185,
      "step": 91
    },
    {
      "epoch": 0.028223023237978373,
      "grad_norm": 3.400557041168213,
      "learning_rate": 4.246548707603114e-05,
      "loss": 1.3707,
      "step": 92
    },
    {
      "epoch": 0.02852979522969553,
      "grad_norm": 3.396674633026123,
      "learning_rate": 4.1899095960801805e-05,
      "loss": 1.2966,
      "step": 93
    },
    {
      "epoch": 0.028836567221412684,
      "grad_norm": 3.7319447994232178,
      "learning_rate": 4.133081870162385e-05,
      "loss": 1.4938,
      "step": 94
    },
    {
      "epoch": 0.02914333921312984,
      "grad_norm": 3.7182352542877197,
      "learning_rate": 4.076081065982569e-05,
      "loss": 1.3977,
      "step": 95
    },
    {
      "epoch": 0.029450111204846998,
      "grad_norm": 3.861218214035034,
      "learning_rate": 4.018922766991447e-05,
      "loss": 1.3694,
      "step": 96
    },
    {
      "epoch": 0.029756883196564155,
      "grad_norm": 4.477148056030273,
      "learning_rate": 3.961622599697241e-05,
      "loss": 1.6513,
      "step": 97
    },
    {
      "epoch": 0.03006365518828131,
      "grad_norm": 4.932404518127441,
      "learning_rate": 3.9041962293935516e-05,
      "loss": 1.3965,
      "step": 98
    },
    {
      "epoch": 0.030370427179998466,
      "grad_norm": 5.398231506347656,
      "learning_rate": 3.84665935587662e-05,
      "loss": 1.3526,
      "step": 99
    },
    {
      "epoch": 0.030677199171715623,
      "grad_norm": 8.890914916992188,
      "learning_rate": 3.7890277091531636e-05,
      "loss": 1.3629,
      "step": 100
    },
    {
      "epoch": 0.030677199171715623,
      "eval_loss": 0.41832274198532104,
      "eval_runtime": 433.466,
      "eval_samples_per_second": 12.665,
      "eval_steps_per_second": 3.167,
      "step": 100
    },
    {
      "epoch": 0.03098397116343278,
      "grad_norm": 3.0474488735198975,
      "learning_rate": 3.7313170451399475e-05,
      "loss": 1.2796,
      "step": 101
    },
    {
      "epoch": 0.03129074315514994,
      "grad_norm": 2.5932962894439697,
      "learning_rate": 3.673543141356278e-05,
      "loss": 1.3824,
      "step": 102
    },
    {
      "epoch": 0.03159751514686709,
      "grad_norm": 2.409973621368408,
      "learning_rate": 3.6157217926105783e-05,
      "loss": 1.4556,
      "step": 103
    },
    {
      "epoch": 0.031904287138584245,
      "grad_norm": 2.359910726547241,
      "learning_rate": 3.557868806682255e-05,
      "loss": 1.6339,
      "step": 104
    },
    {
      "epoch": 0.0322110591303014,
      "grad_norm": 2.1797034740448,
      "learning_rate": 3.5e-05,
      "loss": 1.5914,
      "step": 105
    },
    {
      "epoch": 0.03251783112201856,
      "grad_norm": 2.2220871448516846,
      "learning_rate": 3.442131193317745e-05,
      "loss": 1.6824,
      "step": 106
    },
    {
      "epoch": 0.032824603113735716,
      "grad_norm": 2.2529358863830566,
      "learning_rate": 3.384278207389421e-05,
      "loss": 1.8415,
      "step": 107
    },
    {
      "epoch": 0.03313137510545287,
      "grad_norm": 2.340041399002075,
      "learning_rate": 3.3264568586437216e-05,
      "loss": 1.7342,
      "step": 108
    },
    {
      "epoch": 0.03343814709717003,
      "grad_norm": 2.256408452987671,
      "learning_rate": 3.268682954860052e-05,
      "loss": 1.6666,
      "step": 109
    },
    {
      "epoch": 0.03374491908888719,
      "grad_norm": 2.5215020179748535,
      "learning_rate": 3.210972290846837e-05,
      "loss": 1.8225,
      "step": 110
    },
    {
      "epoch": 0.03405169108060434,
      "grad_norm": 2.403073310852051,
      "learning_rate": 3.15334064412338e-05,
      "loss": 1.7805,
      "step": 111
    },
    {
      "epoch": 0.034358463072321495,
      "grad_norm": 2.3578317165374756,
      "learning_rate": 3.0958037706064485e-05,
      "loss": 1.7553,
      "step": 112
    },
    {
      "epoch": 0.03466523506403865,
      "grad_norm": 2.45162034034729,
      "learning_rate": 3.038377400302758e-05,
      "loss": 1.8525,
      "step": 113
    },
    {
      "epoch": 0.03497200705575581,
      "grad_norm": 2.365379571914673,
      "learning_rate": 2.9810772330085524e-05,
      "loss": 1.752,
      "step": 114
    },
    {
      "epoch": 0.035278779047472966,
      "grad_norm": 2.4482245445251465,
      "learning_rate": 2.9239189340174306e-05,
      "loss": 1.8622,
      "step": 115
    },
    {
      "epoch": 0.03558555103919012,
      "grad_norm": 2.4054617881774902,
      "learning_rate": 2.8669181298376163e-05,
      "loss": 1.7601,
      "step": 116
    },
    {
      "epoch": 0.03589232303090728,
      "grad_norm": 2.410846710205078,
      "learning_rate": 2.8100904039198193e-05,
      "loss": 1.78,
      "step": 117
    },
    {
      "epoch": 0.03619909502262444,
      "grad_norm": 3.7372984886169434,
      "learning_rate": 2.7534512923968863e-05,
      "loss": 1.8017,
      "step": 118
    },
    {
      "epoch": 0.03650586701434159,
      "grad_norm": 2.4058003425598145,
      "learning_rate": 2.6970162798363695e-05,
      "loss": 1.7363,
      "step": 119
    },
    {
      "epoch": 0.036812639006058745,
      "grad_norm": 2.4123528003692627,
      "learning_rate": 2.640800795007203e-05,
      "loss": 1.813,
      "step": 120
    },
    {
      "epoch": 0.0371194109977759,
      "grad_norm": 2.4493489265441895,
      "learning_rate": 2.5848202066616305e-05,
      "loss": 1.855,
      "step": 121
    },
    {
      "epoch": 0.03742618298949306,
      "grad_norm": 2.418009042739868,
      "learning_rate": 2.5290898193335446e-05,
      "loss": 1.758,
      "step": 122
    },
    {
      "epoch": 0.037732954981210216,
      "grad_norm": 2.4469807147979736,
      "learning_rate": 2.4736248691543736e-05,
      "loss": 1.703,
      "step": 123
    },
    {
      "epoch": 0.038039726972927373,
      "grad_norm": 2.555224895477295,
      "learning_rate": 2.4184405196876842e-05,
      "loss": 1.8747,
      "step": 124
    },
    {
      "epoch": 0.03834649896464453,
      "grad_norm": 2.63464093208313,
      "learning_rate": 2.363551857783608e-05,
      "loss": 1.7816,
      "step": 125
    },
    {
      "epoch": 0.03865327095636168,
      "grad_norm": 2.608996629714966,
      "learning_rate": 2.308973889454249e-05,
      "loss": 1.8312,
      "step": 126
    },
    {
      "epoch": 0.03896004294807884,
      "grad_norm": 2.573441982269287,
      "learning_rate": 2.2547215357711918e-05,
      "loss": 1.6914,
      "step": 127
    },
    {
      "epoch": 0.039266814939795995,
      "grad_norm": 2.606969118118286,
      "learning_rate": 2.2008096287862266e-05,
      "loss": 1.6899,
      "step": 128
    },
    {
      "epoch": 0.03957358693151315,
      "grad_norm": 2.78878116607666,
      "learning_rate": 2.1472529074764177e-05,
      "loss": 1.7626,
      "step": 129
    },
    {
      "epoch": 0.03988035892323031,
      "grad_norm": 2.5659425258636475,
      "learning_rate": 2.0940660137146074e-05,
      "loss": 1.522,
      "step": 130
    },
    {
      "epoch": 0.040187130914947466,
      "grad_norm": 2.8045814037323,
      "learning_rate": 2.041263488266484e-05,
      "loss": 1.7199,
      "step": 131
    },
    {
      "epoch": 0.040493902906664624,
      "grad_norm": 2.6969640254974365,
      "learning_rate": 1.988859766815275e-05,
      "loss": 1.5729,
      "step": 132
    },
    {
      "epoch": 0.04080067489838178,
      "grad_norm": 2.780867576599121,
      "learning_rate": 1.9368691760151773e-05,
      "loss": 1.5697,
      "step": 133
    },
    {
      "epoch": 0.04110744689009893,
      "grad_norm": 2.7369863986968994,
      "learning_rate": 1.885305929574593e-05,
      "loss": 1.4807,
      "step": 134
    },
    {
      "epoch": 0.04141421888181609,
      "grad_norm": 2.8618247509002686,
      "learning_rate": 1.8341841243702424e-05,
      "loss": 1.4941,
      "step": 135
    },
    {
      "epoch": 0.041720990873533245,
      "grad_norm": 2.736384868621826,
      "learning_rate": 1.7835177365932225e-05,
      "loss": 1.3604,
      "step": 136
    },
    {
      "epoch": 0.0420277628652504,
      "grad_norm": 2.8534905910491943,
      "learning_rate": 1.7333206179280478e-05,
      "loss": 1.3728,
      "step": 137
    },
    {
      "epoch": 0.04233453485696756,
      "grad_norm": 2.8800296783447266,
      "learning_rate": 1.6836064917657478e-05,
      "loss": 1.3325,
      "step": 138
    },
    {
      "epoch": 0.04264130684868472,
      "grad_norm": 3.1749067306518555,
      "learning_rate": 1.6343889494520224e-05,
      "loss": 1.5899,
      "step": 139
    },
    {
      "epoch": 0.042948078840401874,
      "grad_norm": 2.708810567855835,
      "learning_rate": 1.5856814465715064e-05,
      "loss": 1.0585,
      "step": 140
    },
    {
      "epoch": 0.04325485083211903,
      "grad_norm": 2.938464403152466,
      "learning_rate": 1.5374972992691458e-05,
      "loss": 1.2942,
      "step": 141
    },
    {
      "epoch": 0.04356162282383618,
      "grad_norm": 3.3561573028564453,
      "learning_rate": 1.4898496806096974e-05,
      "loss": 1.508,
      "step": 142
    },
    {
      "epoch": 0.04386839481555334,
      "grad_norm": 3.163780927658081,
      "learning_rate": 1.4427516169763444e-05,
      "loss": 1.3587,
      "step": 143
    },
    {
      "epoch": 0.044175166807270495,
      "grad_norm": 3.5286014080047607,
      "learning_rate": 1.396215984509412e-05,
      "loss": 1.4554,
      "step": 144
    },
    {
      "epoch": 0.04448193879898765,
      "grad_norm": 3.4253334999084473,
      "learning_rate": 1.3502555055861625e-05,
      "loss": 1.1663,
      "step": 145
    },
    {
      "epoch": 0.04478871079070481,
      "grad_norm": 3.902454376220703,
      "learning_rate": 1.3048827453426203e-05,
      "loss": 1.3661,
      "step": 146
    },
    {
      "epoch": 0.04509548278242197,
      "grad_norm": 4.289845943450928,
      "learning_rate": 1.2601101082383917e-05,
      "loss": 1.353,
      "step": 147
    },
    {
      "epoch": 0.045402254774139124,
      "grad_norm": 4.6054768562316895,
      "learning_rate": 1.2159498346654094e-05,
      "loss": 1.2282,
      "step": 148
    },
    {
      "epoch": 0.045709026765856274,
      "grad_norm": 5.709343433380127,
      "learning_rate": 1.1724139976015306e-05,
      "loss": 1.4336,
      "step": 149
    },
    {
      "epoch": 0.04601579875757343,
      "grad_norm": 9.284481048583984,
      "learning_rate": 1.1295144993099068e-05,
      "loss": 1.5566,
      "step": 150
    },
    {
      "epoch": 0.04601579875757343,
      "eval_loss": 0.387969970703125,
      "eval_runtime": 433.356,
      "eval_samples_per_second": 12.669,
      "eval_steps_per_second": 3.168,
      "step": 150
    },
    {
      "epoch": 0.04632257074929059,
      "grad_norm": 1.7831456661224365,
      "learning_rate": 1.0872630680850196e-05,
      "loss": 1.1049,
      "step": 151
    },
    {
      "epoch": 0.046629342741007745,
      "grad_norm": 1.7289410829544067,
      "learning_rate": 1.0456712550462898e-05,
      "loss": 0.9961,
      "step": 152
    },
    {
      "epoch": 0.0469361147327249,
      "grad_norm": 1.969765543937683,
      "learning_rate": 1.0047504309801104e-05,
      "loss": 1.3164,
      "step": 153
    },
    {
      "epoch": 0.04724288672444206,
      "grad_norm": 2.031634569168091,
      "learning_rate": 9.645117832311886e-06,
      "loss": 1.3276,
      "step": 154
    },
    {
      "epoch": 0.04754965871615922,
      "grad_norm": 2.1028342247009277,
      "learning_rate": 9.249663126440394e-06,
      "loss": 1.3828,
      "step": 155
    },
    {
      "epoch": 0.047856430707876374,
      "grad_norm": 2.131300687789917,
      "learning_rate": 8.861248305554624e-06,
      "loss": 1.5613,
      "step": 156
    },
    {
      "epoch": 0.048163202699593524,
      "grad_norm": 2.277365207672119,
      "learning_rate": 8.47997955838829e-06,
      "loss": 1.6239,
      "step": 157
    },
    {
      "epoch": 0.04846997469131068,
      "grad_norm": 2.1458370685577393,
      "learning_rate": 8.10596112000994e-06,
      "loss": 1.5341,
      "step": 158
    },
    {
      "epoch": 0.04877674668302784,
      "grad_norm": 2.270092487335205,
      "learning_rate": 7.739295243326067e-06,
      "loss": 1.6929,
      "step": 159
    },
    {
      "epoch": 0.049083518674744996,
      "grad_norm": 2.3331100940704346,
      "learning_rate": 7.380082171126228e-06,
      "loss": 1.7919,
      "step": 160
    },
    {
      "epoch": 0.04939029066646215,
      "grad_norm": 2.266509532928467,
      "learning_rate": 7.028420108677635e-06,
      "loss": 1.7085,
      "step": 161
    },
    {
      "epoch": 0.04969706265817931,
      "grad_norm": 2.32977294921875,
      "learning_rate": 6.684405196876842e-06,
      "loss": 1.7049,
      "step": 162
    },
    {
      "epoch": 0.05000383464989647,
      "grad_norm": 2.3114678859710693,
      "learning_rate": 6.3481314859657675e-06,
      "loss": 1.8119,
      "step": 163
    },
    {
      "epoch": 0.050310606641613624,
      "grad_norm": 2.366704225540161,
      "learning_rate": 6.019690909819298e-06,
      "loss": 1.6592,
      "step": 164
    },
    {
      "epoch": 0.050617378633330774,
      "grad_norm": 2.2613394260406494,
      "learning_rate": 5.6991732608115e-06,
      "loss": 1.595,
      "step": 165
    },
    {
      "epoch": 0.05092415062504793,
      "grad_norm": 2.284902811050415,
      "learning_rate": 5.386666165267256e-06,
      "loss": 1.7036,
      "step": 166
    },
    {
      "epoch": 0.05123092261676509,
      "grad_norm": 2.395000457763672,
      "learning_rate": 5.08225505950613e-06,
      "loss": 1.8109,
      "step": 167
    },
    {
      "epoch": 0.051537694608482246,
      "grad_norm": 2.420766592025757,
      "learning_rate": 4.786023166484913e-06,
      "loss": 1.7471,
      "step": 168
    },
    {
      "epoch": 0.0518444666001994,
      "grad_norm": 2.4190926551818848,
      "learning_rate": 4.498051473045291e-06,
      "loss": 1.8043,
      "step": 169
    },
    {
      "epoch": 0.05215123859191656,
      "grad_norm": 2.334238290786743,
      "learning_rate": 4.218418707772886e-06,
      "loss": 1.6306,
      "step": 170
    },
    {
      "epoch": 0.05245801058363372,
      "grad_norm": 2.4227912425994873,
      "learning_rate": 3.947201319473587e-06,
      "loss": 1.7711,
      "step": 171
    },
    {
      "epoch": 0.05276478257535087,
      "grad_norm": 2.4554636478424072,
      "learning_rate": 3.684473456273278e-06,
      "loss": 1.7153,
      "step": 172
    },
    {
      "epoch": 0.053071554567068024,
      "grad_norm": 2.431055784225464,
      "learning_rate": 3.4303069453464383e-06,
      "loss": 1.7223,
      "step": 173
    },
    {
      "epoch": 0.05337832655878518,
      "grad_norm": 2.550487756729126,
      "learning_rate": 3.184771273279312e-06,
      "loss": 1.725,
      "step": 174
    },
    {
      "epoch": 0.05368509855050234,
      "grad_norm": 2.4287781715393066,
      "learning_rate": 2.947933567072987e-06,
      "loss": 1.5635,
      "step": 175
    },
    {
      "epoch": 0.053991870542219496,
      "grad_norm": 2.4226584434509277,
      "learning_rate": 2.719858575791534e-06,
      "loss": 1.4968,
      "step": 176
    },
    {
      "epoch": 0.05429864253393665,
      "grad_norm": 2.7177352905273438,
      "learning_rate": 2.500608652860256e-06,
      "loss": 1.7893,
      "step": 177
    },
    {
      "epoch": 0.05460541452565381,
      "grad_norm": 2.628464698791504,
      "learning_rate": 2.2902437390188737e-06,
      "loss": 1.6047,
      "step": 178
    },
    {
      "epoch": 0.05491218651737097,
      "grad_norm": 2.6525192260742188,
      "learning_rate": 2.0888213459343587e-06,
      "loss": 1.6634,
      "step": 179
    },
    {
      "epoch": 0.05521895850908812,
      "grad_norm": 2.5231614112854004,
      "learning_rate": 1.8963965404777875e-06,
      "loss": 1.4988,
      "step": 180
    },
    {
      "epoch": 0.055525730500805275,
      "grad_norm": 2.6935818195343018,
      "learning_rate": 1.7130219296696263e-06,
      "loss": 1.5274,
      "step": 181
    },
    {
      "epoch": 0.05583250249252243,
      "grad_norm": 2.7685813903808594,
      "learning_rate": 1.5387476462974824e-06,
      "loss": 1.5963,
      "step": 182
    },
    {
      "epoch": 0.05613927448423959,
      "grad_norm": 2.7521631717681885,
      "learning_rate": 1.3736213352103147e-06,
      "loss": 1.47,
      "step": 183
    },
    {
      "epoch": 0.056446046475956746,
      "grad_norm": 2.575249195098877,
      "learning_rate": 1.2176881402928002e-06,
      "loss": 1.3513,
      "step": 184
    },
    {
      "epoch": 0.0567528184676739,
      "grad_norm": 2.744507312774658,
      "learning_rate": 1.0709906921234367e-06,
      "loss": 1.4346,
      "step": 185
    },
    {
      "epoch": 0.05705959045939106,
      "grad_norm": 2.918213129043579,
      "learning_rate": 9.33569096319799e-07,
      "loss": 1.5346,
      "step": 186
    },
    {
      "epoch": 0.05736636245110821,
      "grad_norm": 2.8689889907836914,
      "learning_rate": 8.054609225740255e-07,
      "loss": 1.4716,
      "step": 187
    },
    {
      "epoch": 0.05767313444282537,
      "grad_norm": 2.719186305999756,
      "learning_rate": 6.867011943816724e-07,
      "loss": 1.2268,
      "step": 188
    },
    {
      "epoch": 0.057979906434542525,
      "grad_norm": 3.0783579349517822,
      "learning_rate": 5.77322379466617e-07,
      "loss": 1.4754,
      "step": 189
    },
    {
      "epoch": 0.05828667842625968,
      "grad_norm": 3.0656843185424805,
      "learning_rate": 4.773543809047186e-07,
      "loss": 1.3997,
      "step": 190
    },
    {
      "epoch": 0.05859345041797684,
      "grad_norm": 3.3066554069519043,
      "learning_rate": 3.868245289486027e-07,
      "loss": 1.3656,
      "step": 191
    },
    {
      "epoch": 0.058900222409693996,
      "grad_norm": 3.2034060955047607,
      "learning_rate": 3.0575757355586817e-07,
      "loss": 1.3425,
      "step": 192
    },
    {
      "epoch": 0.05920699440141115,
      "grad_norm": 3.28908634185791,
      "learning_rate": 2.3417567762266497e-07,
      "loss": 1.3068,
      "step": 193
    },
    {
      "epoch": 0.05951376639312831,
      "grad_norm": 3.2599215507507324,
      "learning_rate": 1.7209841092460043e-07,
      "loss": 1.1922,
      "step": 194
    },
    {
      "epoch": 0.05982053838484546,
      "grad_norm": 3.786778450012207,
      "learning_rate": 1.1954274476655534e-07,
      "loss": 1.4809,
      "step": 195
    },
    {
      "epoch": 0.06012731037656262,
      "grad_norm": 3.6888482570648193,
      "learning_rate": 7.652304734289127e-08,
      "loss": 1.3954,
      "step": 196
    },
    {
      "epoch": 0.060434082368279775,
      "grad_norm": 4.1001505851745605,
      "learning_rate": 4.30510798093342e-08,
      "loss": 1.3962,
      "step": 197
    },
    {
      "epoch": 0.06074085435999693,
      "grad_norm": 4.536400318145752,
      "learning_rate": 1.9135993067588284e-08,
      "loss": 1.4452,
      "step": 198
    },
    {
      "epoch": 0.06104762635171409,
      "grad_norm": 5.1919403076171875,
      "learning_rate": 4.784325263584854e-09,
      "loss": 1.278,
      "step": 199
    },
    {
      "epoch": 0.061354398343431246,
      "grad_norm": 7.395133018493652,
      "learning_rate": 0.0,
      "loss": 1.3814,
      "step": 200
    },
    {
      "epoch": 0.061354398343431246,
      "eval_loss": 0.38390806317329407,
      "eval_runtime": 433.8363,
      "eval_samples_per_second": 12.655,
      "eval_steps_per_second": 3.165,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 4,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.955014214474793e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}