{
  "best_metric": 0.4560202658176422,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.11421229209793704,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0005710614604896852,
      "grad_norm": 9.655982971191406,
      "learning_rate": 2.333333333333333e-06,
      "loss": 4.9642,
      "step": 1
    },
    {
      "epoch": 0.0005710614604896852,
      "eval_loss": 0.7650408744812012,
      "eval_runtime": 593.1149,
      "eval_samples_per_second": 4.974,
      "eval_steps_per_second": 1.244,
      "step": 1
    },
    {
      "epoch": 0.0011421229209793704,
      "grad_norm": 8.363570213317871,
      "learning_rate": 4.666666666666666e-06,
      "loss": 5.0461,
      "step": 2
    },
    {
      "epoch": 0.0017131843814690555,
      "grad_norm": 8.605849266052246,
      "learning_rate": 7e-06,
      "loss": 4.6528,
      "step": 3
    },
    {
      "epoch": 0.002284245841958741,
      "grad_norm": 10.32148551940918,
      "learning_rate": 9.333333333333333e-06,
      "loss": 5.4813,
      "step": 4
    },
    {
      "epoch": 0.002855307302448426,
      "grad_norm": 7.543620586395264,
      "learning_rate": 1.1666666666666665e-05,
      "loss": 4.938,
      "step": 5
    },
    {
      "epoch": 0.003426368762938111,
      "grad_norm": 7.9145426750183105,
      "learning_rate": 1.4e-05,
      "loss": 5.5469,
      "step": 6
    },
    {
      "epoch": 0.003997430223427797,
      "grad_norm": 7.476115703582764,
      "learning_rate": 1.633333333333333e-05,
      "loss": 5.4585,
      "step": 7
    },
    {
      "epoch": 0.004568491683917482,
      "grad_norm": 7.091987133026123,
      "learning_rate": 1.8666666666666665e-05,
      "loss": 5.2762,
      "step": 8
    },
    {
      "epoch": 0.005139553144407167,
      "grad_norm": 5.893141746520996,
      "learning_rate": 2.1e-05,
      "loss": 4.9478,
      "step": 9
    },
    {
      "epoch": 0.005710614604896852,
      "grad_norm": 5.861713886260986,
      "learning_rate": 2.333333333333333e-05,
      "loss": 5.3339,
      "step": 10
    },
    {
      "epoch": 0.006281676065386537,
      "grad_norm": 5.187582015991211,
      "learning_rate": 2.5666666666666663e-05,
      "loss": 4.6863,
      "step": 11
    },
    {
      "epoch": 0.006852737525876222,
      "grad_norm": 5.725977420806885,
      "learning_rate": 2.8e-05,
      "loss": 4.2163,
      "step": 12
    },
    {
      "epoch": 0.007423798986365908,
      "grad_norm": 6.390807628631592,
      "learning_rate": 3.0333333333333333e-05,
      "loss": 4.7941,
      "step": 13
    },
    {
      "epoch": 0.007994860446855594,
      "grad_norm": 6.105741500854492,
      "learning_rate": 3.266666666666666e-05,
      "loss": 5.1002,
      "step": 14
    },
    {
      "epoch": 0.008565921907345278,
      "grad_norm": 6.119107723236084,
      "learning_rate": 3.5e-05,
      "loss": 4.9678,
      "step": 15
    },
    {
      "epoch": 0.009136983367834963,
      "grad_norm": 5.37454891204834,
      "learning_rate": 3.733333333333333e-05,
      "loss": 4.6234,
      "step": 16
    },
    {
      "epoch": 0.00970804482832465,
      "grad_norm": 5.023365497589111,
      "learning_rate": 3.9666666666666664e-05,
      "loss": 4.5956,
      "step": 17
    },
    {
      "epoch": 0.010279106288814333,
      "grad_norm": 5.116684436798096,
      "learning_rate": 4.2e-05,
      "loss": 4.4032,
      "step": 18
    },
    {
      "epoch": 0.010850167749304019,
      "grad_norm": 4.848908424377441,
      "learning_rate": 4.4333333333333324e-05,
      "loss": 4.1467,
      "step": 19
    },
    {
      "epoch": 0.011421229209793705,
      "grad_norm": 5.300320625305176,
      "learning_rate": 4.666666666666666e-05,
      "loss": 4.711,
      "step": 20
    },
    {
      "epoch": 0.011992290670283389,
      "grad_norm": 4.577368259429932,
      "learning_rate": 4.899999999999999e-05,
      "loss": 4.1949,
      "step": 21
    },
    {
      "epoch": 0.012563352130773075,
      "grad_norm": 4.3569560050964355,
      "learning_rate": 5.1333333333333325e-05,
      "loss": 4.6698,
      "step": 22
    },
    {
      "epoch": 0.01313441359126276,
      "grad_norm": 4.732718467712402,
      "learning_rate": 5.3666666666666666e-05,
      "loss": 4.4054,
      "step": 23
    },
    {
      "epoch": 0.013705475051752444,
      "grad_norm": 4.412967681884766,
      "learning_rate": 5.6e-05,
      "loss": 4.3697,
      "step": 24
    },
    {
      "epoch": 0.01427653651224213,
      "grad_norm": 4.235969543457031,
      "learning_rate": 5.833333333333333e-05,
      "loss": 4.3441,
      "step": 25
    },
    {
      "epoch": 0.014847597972731816,
      "grad_norm": 4.274826526641846,
      "learning_rate": 6.0666666666666666e-05,
      "loss": 4.6215,
      "step": 26
    },
    {
      "epoch": 0.0154186594332215,
      "grad_norm": 4.126132488250732,
      "learning_rate": 6.3e-05,
      "loss": 4.3444,
      "step": 27
    },
    {
      "epoch": 0.015989720893711187,
      "grad_norm": 4.711647987365723,
      "learning_rate": 6.533333333333333e-05,
      "loss": 4.4501,
      "step": 28
    },
    {
      "epoch": 0.01656078235420087,
      "grad_norm": 4.457403659820557,
      "learning_rate": 6.766666666666667e-05,
      "loss": 4.5348,
      "step": 29
    },
    {
      "epoch": 0.017131843814690555,
      "grad_norm": 4.318757057189941,
      "learning_rate": 7e-05,
      "loss": 4.8379,
      "step": 30
    },
    {
      "epoch": 0.01770290527518024,
      "grad_norm": 4.192445755004883,
      "learning_rate": 6.999402376603183e-05,
      "loss": 4.8049,
      "step": 31
    },
    {
      "epoch": 0.018273966735669927,
      "grad_norm": 4.142071723937988,
      "learning_rate": 6.99760971050058e-05,
      "loss": 4.4266,
      "step": 32
    },
    {
      "epoch": 0.018845028196159613,
      "grad_norm": 4.6366167068481445,
      "learning_rate": 6.994622613886018e-05,
      "loss": 5.603,
      "step": 33
    },
    {
      "epoch": 0.0194160896566493,
      "grad_norm": 4.616098880767822,
      "learning_rate": 6.990442106850258e-05,
      "loss": 5.0995,
      "step": 34
    },
    {
      "epoch": 0.01998715111713898,
      "grad_norm": 4.354705333709717,
      "learning_rate": 6.98506961703262e-05,
      "loss": 4.8896,
      "step": 35
    },
    {
      "epoch": 0.020558212577628666,
      "grad_norm": 4.386688709259033,
      "learning_rate": 6.978506979133457e-05,
      "loss": 4.4294,
      "step": 36
    },
    {
      "epoch": 0.021129274038118352,
      "grad_norm": 4.3963093757629395,
      "learning_rate": 6.9707564342876e-05,
      "loss": 4.2744,
      "step": 37
    },
    {
      "epoch": 0.021700335498608038,
      "grad_norm": 4.365737438201904,
      "learning_rate": 6.96182062929901e-05,
      "loss": 3.9286,
      "step": 38
    },
    {
      "epoch": 0.022271396959097724,
      "grad_norm": 4.8030571937561035,
      "learning_rate": 6.951702615736908e-05,
      "loss": 4.4058,
      "step": 39
    },
    {
      "epoch": 0.02284245841958741,
      "grad_norm": 5.323755264282227,
      "learning_rate": 6.940405848893656e-05,
      "loss": 4.145,
      "step": 40
    },
    {
      "epoch": 0.023413519880077092,
      "grad_norm": 4.974182605743408,
      "learning_rate": 6.92793418660478e-05,
      "loss": 4.5518,
      "step": 41
    },
    {
      "epoch": 0.023984581340566778,
      "grad_norm": 5.099968910217285,
      "learning_rate": 6.914291887931528e-05,
      "loss": 4.0907,
      "step": 42
    },
    {
      "epoch": 0.024555642801056463,
      "grad_norm": 4.776078224182129,
      "learning_rate": 6.899483611706398e-05,
      "loss": 4.1122,
      "step": 43
    },
    {
      "epoch": 0.02512670426154615,
      "grad_norm": 4.876911640167236,
      "learning_rate": 6.883514414942155e-05,
      "loss": 4.1617,
      "step": 44
    },
    {
      "epoch": 0.025697765722035835,
      "grad_norm": 5.060526371002197,
      "learning_rate": 6.866389751104867e-05,
      "loss": 3.7848,
      "step": 45
    },
    {
      "epoch": 0.02626882718252552,
      "grad_norm": 5.239251136779785,
      "learning_rate": 6.848115468251542e-05,
      "loss": 4.0855,
      "step": 46
    },
    {
      "epoch": 0.026839888643015203,
      "grad_norm": 5.517544269561768,
      "learning_rate": 6.828697807033038e-05,
      "loss": 4.1965,
      "step": 47
    },
    {
      "epoch": 0.02741095010350489,
      "grad_norm": 5.758817672729492,
      "learning_rate": 6.808143398562868e-05,
      "loss": 4.0623,
      "step": 48
    },
    {
      "epoch": 0.027982011563994574,
      "grad_norm": 5.886153697967529,
      "learning_rate": 6.786459262152698e-05,
      "loss": 3.7384,
      "step": 49
    },
    {
      "epoch": 0.02855307302448426,
      "grad_norm": 8.272394180297852,
      "learning_rate": 6.763652802915244e-05,
      "loss": 3.7424,
      "step": 50
    },
    {
      "epoch": 0.02855307302448426,
      "eval_loss": 0.5147433876991272,
      "eval_runtime": 595.5275,
      "eval_samples_per_second": 4.954,
      "eval_steps_per_second": 1.239,
      "step": 50
    },
    {
      "epoch": 0.029124134484973946,
      "grad_norm": 4.8110432624816895,
      "learning_rate": 6.739731809235446e-05,
      "loss": 3.3424,
      "step": 51
    },
    {
      "epoch": 0.02969519594546363,
      "grad_norm": 4.688547611236572,
      "learning_rate": 6.71470445011073e-05,
      "loss": 3.6607,
      "step": 52
    },
    {
      "epoch": 0.030266257405953317,
      "grad_norm": 4.069744110107422,
      "learning_rate": 6.688579272361309e-05,
      "loss": 3.8212,
      "step": 53
    },
    {
      "epoch": 0.030837318866443,
      "grad_norm": 3.885037422180176,
      "learning_rate": 6.66136519771145e-05,
      "loss": 3.7556,
      "step": 54
    },
    {
      "epoch": 0.03140838032693269,
      "grad_norm": 3.9566140174865723,
      "learning_rate": 6.633071519742718e-05,
      "loss": 3.6849,
      "step": 55
    },
    {
      "epoch": 0.031979441787422375,
      "grad_norm": 4.207897186279297,
      "learning_rate": 6.603707900720217e-05,
      "loss": 3.3258,
      "step": 56
    },
    {
      "epoch": 0.03255050324791205,
      "grad_norm": 4.095637321472168,
      "learning_rate": 6.573284368292943e-05,
      "loss": 3.8785,
      "step": 57
    },
    {
      "epoch": 0.03312156470840174,
      "grad_norm": 3.729644536972046,
      "learning_rate": 6.541811312069348e-05,
      "loss": 3.2926,
      "step": 58
    },
    {
      "epoch": 0.033692626168891425,
      "grad_norm": 4.160499572753906,
      "learning_rate": 6.509299480069303e-05,
      "loss": 3.6083,
      "step": 59
    },
    {
      "epoch": 0.03426368762938111,
      "grad_norm": 3.895054578781128,
      "learning_rate": 6.47575997505365e-05,
      "loss": 3.4853,
      "step": 60
    },
    {
      "epoch": 0.034834749089870796,
      "grad_norm": 3.4832022190093994,
      "learning_rate": 6.441204250732624e-05,
      "loss": 3.1942,
      "step": 61
    },
    {
      "epoch": 0.03540581055036048,
      "grad_norm": 3.5753769874572754,
      "learning_rate": 6.405644107854427e-05,
      "loss": 3.6576,
      "step": 62
    },
    {
      "epoch": 0.03597687201085017,
      "grad_norm": 3.4942517280578613,
      "learning_rate": 6.369091690175273e-05,
      "loss": 3.5523,
      "step": 63
    },
    {
      "epoch": 0.036547933471339854,
      "grad_norm": 3.4585907459259033,
      "learning_rate": 6.331559480312315e-05,
      "loss": 3.4847,
      "step": 64
    },
    {
      "epoch": 0.03711899493182954,
      "grad_norm": 3.7623393535614014,
      "learning_rate": 6.293060295480838e-05,
      "loss": 3.9728,
      "step": 65
    },
    {
      "epoch": 0.037690056392319225,
      "grad_norm": 3.7405989170074463,
      "learning_rate": 6.25360728311719e-05,
      "loss": 3.6524,
      "step": 66
    },
    {
      "epoch": 0.03826111785280891,
      "grad_norm": 3.857524871826172,
      "learning_rate": 6.213213916388954e-05,
      "loss": 4.1654,
      "step": 67
    },
    {
      "epoch": 0.0388321793132986,
      "grad_norm": 3.8793821334838867,
      "learning_rate": 6.171893989593859e-05,
      "loss": 3.9822,
      "step": 68
    },
    {
      "epoch": 0.039403240773788276,
      "grad_norm": 3.9082939624786377,
      "learning_rate": 6.129661613449057e-05,
      "loss": 4.042,
      "step": 69
    },
    {
      "epoch": 0.03997430223427796,
      "grad_norm": 3.849015474319458,
      "learning_rate": 6.086531210272306e-05,
      "loss": 3.68,
      "step": 70
    },
    {
      "epoch": 0.04054536369476765,
      "grad_norm": 3.643329381942749,
      "learning_rate": 6.042517509056784e-05,
      "loss": 3.4066,
      "step": 71
    },
    {
      "epoch": 0.04111642515525733,
      "grad_norm": 3.8051083087921143,
      "learning_rate": 5.997635540441133e-05,
      "loss": 3.9113,
      "step": 72
    },
    {
      "epoch": 0.04168748661574702,
      "grad_norm": 3.7576425075531006,
      "learning_rate": 5.9519006315765176e-05,
      "loss": 3.9575,
      "step": 73
    },
    {
      "epoch": 0.042258548076236704,
      "grad_norm": 4.091893196105957,
      "learning_rate": 5.9053284008924185e-05,
      "loss": 4.044,
      "step": 74
    },
    {
      "epoch": 0.04282960953672639,
      "grad_norm": 3.665947437286377,
      "learning_rate": 5.85793475276295e-05,
      "loss": 3.7113,
      "step": 75
    },
    {
      "epoch": 0.043400670997216076,
      "grad_norm": 3.7637410163879395,
      "learning_rate": 5.809735872075529e-05,
      "loss": 3.9625,
      "step": 76
    },
    {
      "epoch": 0.04397173245770576,
      "grad_norm": 3.848707437515259,
      "learning_rate": 5.760748218703755e-05,
      "loss": 4.3046,
      "step": 77
    },
    {
      "epoch": 0.04454279391819545,
      "grad_norm": 3.8582873344421387,
      "learning_rate": 5.710988521886378e-05,
      "loss": 4.0915,
      "step": 78
    },
    {
      "epoch": 0.04511385537868513,
      "grad_norm": 3.9476304054260254,
      "learning_rate": 5.660473774514275e-05,
      "loss": 3.9352,
      "step": 79
    },
    {
      "epoch": 0.04568491683917482,
      "grad_norm": 3.7469491958618164,
      "learning_rate": 5.6092212273273975e-05,
      "loss": 3.8668,
      "step": 80
    },
    {
      "epoch": 0.046255978299664505,
      "grad_norm": 3.9498250484466553,
      "learning_rate": 5.557248383023655e-05,
      "loss": 4.3159,
      "step": 81
    },
    {
      "epoch": 0.046827039760154184,
      "grad_norm": 3.702901601791382,
      "learning_rate": 5.5045729902817676e-05,
      "loss": 4.057,
      "step": 82
    },
    {
      "epoch": 0.04739810122064387,
      "grad_norm": 4.221633434295654,
      "learning_rate": 5.4512130377000987e-05,
      "loss": 4.4756,
      "step": 83
    },
    {
      "epoch": 0.047969162681133555,
      "grad_norm": 3.9935688972473145,
      "learning_rate": 5.397186747653573e-05,
      "loss": 4.9131,
      "step": 84
    },
    {
      "epoch": 0.04854022414162324,
      "grad_norm": 3.9818077087402344,
      "learning_rate": 5.342512570070745e-05,
      "loss": 4.3006,
      "step": 85
    },
    {
      "epoch": 0.04911128560211293,
      "grad_norm": 3.9484941959381104,
      "learning_rate": 5.287209176133174e-05,
      "loss": 3.9452,
      "step": 86
    },
    {
      "epoch": 0.04968234706260261,
      "grad_norm": 3.9193077087402344,
      "learning_rate": 5.231295451899226e-05,
      "loss": 3.8865,
      "step": 87
    },
    {
      "epoch": 0.0502534085230923,
      "grad_norm": 4.1261305809021,
      "learning_rate": 5.174790491854502e-05,
      "loss": 4.026,
      "step": 88
    },
    {
      "epoch": 0.050824469983581984,
      "grad_norm": 4.0768351554870605,
      "learning_rate": 5.117713592391096e-05,
      "loss": 3.7968,
      "step": 89
    },
    {
      "epoch": 0.05139553144407167,
      "grad_norm": 4.356061935424805,
      "learning_rate": 5.060084245217884e-05,
      "loss": 4.048,
      "step": 90
    },
    {
      "epoch": 0.051966592904561355,
      "grad_norm": 4.245785236358643,
      "learning_rate": 5.0019221307041306e-05,
      "loss": 3.5643,
      "step": 91
    },
    {
      "epoch": 0.05253765436505104,
      "grad_norm": 4.084421634674072,
      "learning_rate": 4.943247111158662e-05,
      "loss": 3.5692,
      "step": 92
    },
    {
      "epoch": 0.05310871582554073,
      "grad_norm": 4.277196407318115,
      "learning_rate": 4.884079224046898e-05,
      "loss": 4.2165,
      "step": 93
    },
    {
      "epoch": 0.053679777286030406,
      "grad_norm": 4.16418981552124,
      "learning_rate": 4.824438675148086e-05,
      "loss": 3.3093,
      "step": 94
    },
    {
      "epoch": 0.05425083874652009,
      "grad_norm": 4.666293144226074,
      "learning_rate": 4.764345831655036e-05,
      "loss": 3.5172,
      "step": 95
    },
    {
      "epoch": 0.05482190020700978,
      "grad_norm": 4.601908206939697,
      "learning_rate": 4.703821215218748e-05,
      "loss": 3.3629,
      "step": 96
    },
    {
      "epoch": 0.05539296166749946,
      "grad_norm": 4.77185583114624,
      "learning_rate": 4.642885494940291e-05,
      "loss": 3.5901,
      "step": 97
    },
    {
      "epoch": 0.05596402312798915,
      "grad_norm": 5.006587028503418,
      "learning_rate": 4.581559480312316e-05,
      "loss": 3.2387,
      "step": 98
    },
    {
      "epoch": 0.056535084588478834,
      "grad_norm": 5.511896133422852,
      "learning_rate": 4.519864114112636e-05,
      "loss": 3.5257,
      "step": 99
    },
    {
      "epoch": 0.05710614604896852,
      "grad_norm": 7.265841007232666,
      "learning_rate": 4.45782046525229e-05,
      "loss": 3.6848,
      "step": 100
    },
    {
      "epoch": 0.05710614604896852,
      "eval_loss": 0.4799676537513733,
      "eval_runtime": 595.5075,
      "eval_samples_per_second": 4.954,
      "eval_steps_per_second": 1.239,
      "step": 100
    },
    {
      "epoch": 0.057677207509458206,
      "grad_norm": 4.521103382110596,
      "learning_rate": 4.3954497215805244e-05,
      "loss": 4.1905,
      "step": 101
    },
    {
      "epoch": 0.05824826896994789,
      "grad_norm": 3.9009766578674316,
      "learning_rate": 4.332773182649165e-05,
      "loss": 3.408,
      "step": 102
    },
    {
      "epoch": 0.05881933043043758,
      "grad_norm": 3.8264896869659424,
      "learning_rate": 4.2698122524388405e-05,
      "loss": 3.433,
      "step": 103
    },
    {
      "epoch": 0.05939039189092726,
      "grad_norm": 3.7116568088531494,
      "learning_rate": 4.206588432049535e-05,
      "loss": 3.5134,
      "step": 104
    },
    {
      "epoch": 0.05996145335141695,
      "grad_norm": 3.5707197189331055,
      "learning_rate": 4.143123312357996e-05,
      "loss": 3.0987,
      "step": 105
    },
    {
      "epoch": 0.060532514811906635,
      "grad_norm": 3.62729811668396,
      "learning_rate": 4.079438566644454e-05,
      "loss": 3.7314,
      "step": 106
    },
    {
      "epoch": 0.061103576272396314,
      "grad_norm": 3.6416237354278564,
      "learning_rate": 4.015555943191231e-05,
      "loss": 3.4506,
      "step": 107
    },
    {
      "epoch": 0.061674637732886,
      "grad_norm": 3.462261199951172,
      "learning_rate": 3.9514972578557114e-05,
      "loss": 3.3004,
      "step": 108
    },
    {
      "epoch": 0.062245699193375685,
      "grad_norm": 3.664212942123413,
      "learning_rate": 3.8872843866202525e-05,
      "loss": 3.3434,
      "step": 109
    },
    {
      "epoch": 0.06281676065386538,
      "grad_norm": 3.6109557151794434,
      "learning_rate": 3.8229392581215565e-05,
      "loss": 3.6783,
      "step": 110
    },
    {
      "epoch": 0.06338782211435506,
      "grad_norm": 3.6711537837982178,
      "learning_rate": 3.7584838461620587e-05,
      "loss": 3.5776,
      "step": 111
    },
    {
      "epoch": 0.06395888357484475,
      "grad_norm": 3.6779439449310303,
      "learning_rate": 3.693940162205895e-05,
      "loss": 3.6493,
      "step": 112
    },
    {
      "epoch": 0.06452994503533442,
      "grad_norm": 3.770522356033325,
      "learning_rate": 3.629330247862007e-05,
      "loss": 3.4767,
      "step": 113
    },
    {
      "epoch": 0.0651010064958241,
      "grad_norm": 3.7113234996795654,
      "learning_rate": 3.564676167356954e-05,
      "loss": 3.3844,
      "step": 114
    },
    {
      "epoch": 0.06567206795631379,
      "grad_norm": 3.5234830379486084,
      "learning_rate": 3.5e-05,
      "loss": 3.3633,
      "step": 115
    },
    {
      "epoch": 0.06624312941680348,
      "grad_norm": 3.8543918132781982,
      "learning_rate": 3.435323832643046e-05,
      "loss": 4.189,
      "step": 116
    },
    {
      "epoch": 0.06681419087729316,
      "grad_norm": 3.756317138671875,
      "learning_rate": 3.370669752137993e-05,
      "loss": 3.4077,
      "step": 117
    },
    {
      "epoch": 0.06738525233778285,
      "grad_norm": 3.5420141220092773,
      "learning_rate": 3.306059837794105e-05,
      "loss": 3.577,
      "step": 118
    },
    {
      "epoch": 0.06795631379827254,
      "grad_norm": 3.767037868499756,
      "learning_rate": 3.241516153837941e-05,
      "loss": 4.1839,
      "step": 119
    },
    {
      "epoch": 0.06852737525876222,
      "grad_norm": 3.63523268699646,
      "learning_rate": 3.177060741878443e-05,
      "loss": 3.8141,
      "step": 120
    },
    {
      "epoch": 0.06909843671925191,
      "grad_norm": 3.623882532119751,
      "learning_rate": 3.1127156133797475e-05,
      "loss": 3.5584,
      "step": 121
    },
    {
      "epoch": 0.06966949817974159,
      "grad_norm": 3.6835837364196777,
      "learning_rate": 3.048502742144289e-05,
      "loss": 3.6744,
      "step": 122
    },
    {
      "epoch": 0.07024055964023128,
      "grad_norm": 3.5777692794799805,
      "learning_rate": 2.984444056808768e-05,
      "loss": 3.8484,
      "step": 123
    },
    {
      "epoch": 0.07081162110072096,
      "grad_norm": 3.7121081352233887,
      "learning_rate": 2.9205614333555444e-05,
      "loss": 3.9062,
      "step": 124
    },
    {
      "epoch": 0.07138268256121065,
      "grad_norm": 3.627039909362793,
      "learning_rate": 2.856876687642003e-05,
      "loss": 3.8178,
      "step": 125
    },
    {
      "epoch": 0.07195374402170034,
      "grad_norm": 3.5459561347961426,
      "learning_rate": 2.7934115679504645e-05,
      "loss": 3.4396,
      "step": 126
    },
    {
      "epoch": 0.07252480548219002,
      "grad_norm": 3.5370867252349854,
      "learning_rate": 2.7301877475611606e-05,
      "loss": 3.5644,
      "step": 127
    },
    {
      "epoch": 0.07309586694267971,
      "grad_norm": 3.6710925102233887,
      "learning_rate": 2.667226817350835e-05,
      "loss": 3.8682,
      "step": 128
    },
    {
      "epoch": 0.0736669284031694,
      "grad_norm": 3.8747775554656982,
      "learning_rate": 2.604550278419475e-05,
      "loss": 4.4182,
      "step": 129
    },
    {
      "epoch": 0.07423798986365908,
      "grad_norm": 3.7661991119384766,
      "learning_rate": 2.54217953474771e-05,
      "loss": 4.1295,
      "step": 130
    },
    {
      "epoch": 0.07480905132414876,
      "grad_norm": 3.6781368255615234,
      "learning_rate": 2.4801358858873636e-05,
      "loss": 4.1132,
      "step": 131
    },
    {
      "epoch": 0.07538011278463845,
      "grad_norm": 3.6991829872131348,
      "learning_rate": 2.4184405196876842e-05,
      "loss": 4.3835,
      "step": 132
    },
    {
      "epoch": 0.07595117424512814,
      "grad_norm": 3.730729341506958,
      "learning_rate": 2.3571145050597088e-05,
      "loss": 4.2776,
      "step": 133
    },
    {
      "epoch": 0.07652223570561782,
      "grad_norm": 3.956230401992798,
      "learning_rate": 2.296178784781251e-05,
      "loss": 4.8423,
      "step": 134
    },
    {
      "epoch": 0.07709329716610751,
      "grad_norm": 3.769970417022705,
      "learning_rate": 2.2356541683449646e-05,
      "loss": 4.1285,
      "step": 135
    },
    {
      "epoch": 0.0776643586265972,
      "grad_norm": 4.056671619415283,
      "learning_rate": 2.175561324851914e-05,
      "loss": 4.1466,
      "step": 136
    },
    {
      "epoch": 0.07823542008708688,
      "grad_norm": 3.9425039291381836,
      "learning_rate": 2.1159207759531013e-05,
      "loss": 3.734,
      "step": 137
    },
    {
      "epoch": 0.07880648154757655,
      "grad_norm": 3.9520437717437744,
      "learning_rate": 2.0567528888413382e-05,
      "loss": 3.4524,
      "step": 138
    },
    {
      "epoch": 0.07937754300806624,
      "grad_norm": 3.9661831855773926,
      "learning_rate": 1.9980778692958684e-05,
      "loss": 3.6408,
      "step": 139
    },
    {
      "epoch": 0.07994860446855592,
      "grad_norm": 4.106236934661865,
      "learning_rate": 1.9399157547821162e-05,
      "loss": 3.5565,
      "step": 140
    },
    {
      "epoch": 0.08051966592904561,
      "grad_norm": 4.123175621032715,
      "learning_rate": 1.882286407608904e-05,
      "loss": 3.8822,
      "step": 141
    },
    {
      "epoch": 0.0810907273895353,
      "grad_norm": 4.1223859786987305,
      "learning_rate": 1.825209508145497e-05,
      "loss": 3.5751,
      "step": 142
    },
    {
      "epoch": 0.08166178885002498,
      "grad_norm": 4.237730979919434,
      "learning_rate": 1.7687045481007746e-05,
      "loss": 3.8547,
      "step": 143
    },
    {
      "epoch": 0.08223285031051467,
      "grad_norm": 4.241462707519531,
      "learning_rate": 1.712790823866826e-05,
      "loss": 3.5782,
      "step": 144
    },
    {
      "epoch": 0.08280391177100435,
      "grad_norm": 4.347367763519287,
      "learning_rate": 1.657487429929254e-05,
      "loss": 3.3161,
      "step": 145
    },
    {
      "epoch": 0.08337497323149404,
      "grad_norm": 4.680622577667236,
      "learning_rate": 1.602813252346427e-05,
      "loss": 3.7808,
      "step": 146
    },
    {
      "epoch": 0.08394603469198372,
      "grad_norm": 4.457460403442383,
      "learning_rate": 1.5487869622999004e-05,
      "loss": 3.2924,
      "step": 147
    },
    {
      "epoch": 0.08451709615247341,
      "grad_norm": 4.6479172706604,
      "learning_rate": 1.4954270097182317e-05,
      "loss": 3.3637,
      "step": 148
    },
    {
      "epoch": 0.0850881576129631,
      "grad_norm": 5.357884883880615,
      "learning_rate": 1.4427516169763444e-05,
      "loss": 3.3677,
      "step": 149
    },
    {
      "epoch": 0.08565921907345278,
      "grad_norm": 6.834778308868408,
      "learning_rate": 1.3907787726726029e-05,
      "loss": 3.6711,
      "step": 150
    },
    {
      "epoch": 0.08565921907345278,
      "eval_loss": 0.46051260828971863,
      "eval_runtime": 596.2501,
      "eval_samples_per_second": 4.948,
      "eval_steps_per_second": 1.238,
      "step": 150
    },
    {
      "epoch": 0.08623028053394247,
      "grad_norm": 3.791814088821411,
      "learning_rate": 1.339526225485725e-05,
      "loss": 3.4645,
      "step": 151
    },
    {
      "epoch": 0.08680134199443215,
      "grad_norm": 3.796891212463379,
      "learning_rate": 1.2890114781136224e-05,
      "loss": 3.4475,
      "step": 152
    },
    {
      "epoch": 0.08737240345492184,
      "grad_norm": 3.6403725147247314,
      "learning_rate": 1.239251781296245e-05,
      "loss": 3.1833,
      "step": 153
    },
    {
      "epoch": 0.08794346491541152,
      "grad_norm": 3.6974196434020996,
      "learning_rate": 1.1902641279244715e-05,
      "loss": 3.3696,
      "step": 154
    },
    {
      "epoch": 0.08851452637590121,
      "grad_norm": 3.578291893005371,
      "learning_rate": 1.1420652472370497e-05,
      "loss": 3.1136,
      "step": 155
    },
    {
      "epoch": 0.0890855878363909,
      "grad_norm": 3.4341652393341064,
      "learning_rate": 1.0946715991075805e-05,
      "loss": 2.9641,
      "step": 156
    },
    {
      "epoch": 0.08965664929688058,
      "grad_norm": 3.451934576034546,
      "learning_rate": 1.0480993684234815e-05,
      "loss": 3.1792,
      "step": 157
    },
    {
      "epoch": 0.09022771075737027,
      "grad_norm": 3.3800251483917236,
      "learning_rate": 1.0023644595588671e-05,
      "loss": 3.3144,
      "step": 158
    },
    {
      "epoch": 0.09079877221785995,
      "grad_norm": 3.479318618774414,
      "learning_rate": 9.57482490943216e-06,
      "loss": 3.1648,
      "step": 159
    },
    {
      "epoch": 0.09136983367834964,
      "grad_norm": 3.491649866104126,
      "learning_rate": 9.134687897276934e-06,
      "loss": 3.3233,
      "step": 160
    },
    {
      "epoch": 0.09194089513883932,
      "grad_norm": 3.484065055847168,
      "learning_rate": 8.703383865509432e-06,
      "loss": 3.5239,
      "step": 161
    },
    {
      "epoch": 0.09251195659932901,
      "grad_norm": 3.6290812492370605,
      "learning_rate": 8.281060104061394e-06,
      "loss": 3.2998,
      "step": 162
    },
    {
      "epoch": 0.09308301805981868,
      "grad_norm": 3.6152350902557373,
      "learning_rate": 7.867860836110453e-06,
      "loss": 3.7931,
      "step": 163
    },
    {
      "epoch": 0.09365407952030837,
      "grad_norm": 3.5166842937469482,
      "learning_rate": 7.463927168828087e-06,
      "loss": 3.5036,
      "step": 164
    },
    {
      "epoch": 0.09422514098079805,
      "grad_norm": 3.266004800796509,
      "learning_rate": 7.069397045191617e-06,
      "loss": 2.7839,
      "step": 165
    },
    {
      "epoch": 0.09479620244128774,
      "grad_norm": 3.5984013080596924,
      "learning_rate": 6.684405196876842e-06,
      "loss": 3.7792,
      "step": 166
    },
    {
      "epoch": 0.09536726390177742,
      "grad_norm": 3.4622771739959717,
      "learning_rate": 6.309083098247264e-06,
      "loss": 3.4876,
      "step": 167
    },
    {
      "epoch": 0.09593832536226711,
      "grad_norm": 3.397523880004883,
      "learning_rate": 5.943558921455733e-06,
      "loss": 3.202,
      "step": 168
    },
    {
      "epoch": 0.0965093868227568,
      "grad_norm": 3.561593770980835,
      "learning_rate": 5.587957492673759e-06,
      "loss": 3.5857,
      "step": 169
    },
    {
      "epoch": 0.09708044828324648,
      "grad_norm": 3.7165884971618652,
      "learning_rate": 5.2424002494635095e-06,
      "loss": 3.6163,
      "step": 170
    },
    {
      "epoch": 0.09765150974373617,
      "grad_norm": 3.710353374481201,
      "learning_rate": 4.9070051993069636e-06,
      "loss": 3.9078,
      "step": 171
    },
    {
      "epoch": 0.09822257120422585,
      "grad_norm": 3.53379225730896,
      "learning_rate": 4.581886879306507e-06,
      "loss": 3.5316,
      "step": 172
    },
    {
      "epoch": 0.09879363266471554,
      "grad_norm": 3.5342020988464355,
      "learning_rate": 4.2671563170705725e-06,
      "loss": 3.5556,
      "step": 173
    },
    {
      "epoch": 0.09936469412520522,
      "grad_norm": 3.5446951389312744,
      "learning_rate": 3.962920992797834e-06,
      "loss": 3.4027,
      "step": 174
    },
    {
      "epoch": 0.09993575558569491,
      "grad_norm": 3.7609660625457764,
      "learning_rate": 3.6692848025728216e-06,
      "loss": 4.0196,
      "step": 175
    },
    {
      "epoch": 0.1005068170461846,
      "grad_norm": 3.5949559211730957,
      "learning_rate": 3.38634802288549e-06,
      "loss": 3.7018,
      "step": 176
    },
    {
      "epoch": 0.10107787850667428,
      "grad_norm": 3.4899206161499023,
      "learning_rate": 3.1142072763869042e-06,
      "loss": 3.4402,
      "step": 177
    },
    {
      "epoch": 0.10164893996716397,
      "grad_norm": 3.4958362579345703,
      "learning_rate": 2.852955498892694e-06,
      "loss": 3.6609,
      "step": 178
    },
    {
      "epoch": 0.10222000142765365,
      "grad_norm": 3.7367665767669678,
      "learning_rate": 2.6026819076455325e-06,
      "loss": 3.9386,
      "step": 179
    },
    {
      "epoch": 0.10279106288814334,
      "grad_norm": 3.6101346015930176,
      "learning_rate": 2.36347197084755e-06,
      "loss": 3.589,
      "step": 180
    },
    {
      "epoch": 0.10336212434863302,
      "grad_norm": 3.714442491531372,
      "learning_rate": 2.1354073784730253e-06,
      "loss": 3.9742,
      "step": 181
    },
    {
      "epoch": 0.10393318580912271,
      "grad_norm": 3.8775899410247803,
      "learning_rate": 1.9185660143713184e-06,
      "loss": 4.4065,
      "step": 182
    },
    {
      "epoch": 0.1045042472696124,
      "grad_norm": 4.041444301605225,
      "learning_rate": 1.7130219296696263e-06,
      "loss": 4.5485,
      "step": 183
    },
    {
      "epoch": 0.10507530873010208,
      "grad_norm": 3.8321800231933594,
      "learning_rate": 1.5188453174845743e-06,
      "loss": 4.2913,
      "step": 184
    },
    {
      "epoch": 0.10564637019059177,
      "grad_norm": 3.849231481552124,
      "learning_rate": 1.3361024889513333e-06,
      "loss": 4.6328,
      "step": 185
    },
    {
      "epoch": 0.10621743165108145,
      "grad_norm": 3.960550546646118,
      "learning_rate": 1.16485585057844e-06,
      "loss": 4.6204,
      "step": 186
    },
    {
      "epoch": 0.10678849311157114,
      "grad_norm": 3.7694177627563477,
      "learning_rate": 1.0051638829360127e-06,
      "loss": 3.6095,
      "step": 187
    },
    {
      "epoch": 0.10735955457206081,
      "grad_norm": 3.936204671859741,
      "learning_rate": 8.570811206847189e-07,
      "loss": 4.067,
      "step": 188
    },
    {
      "epoch": 0.1079306160325505,
      "grad_norm": 4.248170852661133,
      "learning_rate": 7.206581339521939e-07,
      "loss": 4.3996,
      "step": 189
    },
    {
      "epoch": 0.10850167749304018,
      "grad_norm": 4.024265766143799,
      "learning_rate": 5.959415110634375e-07,
      "loss": 3.9334,
      "step": 190
    },
    {
      "epoch": 0.10907273895352987,
      "grad_norm": 4.193387508392334,
      "learning_rate": 4.829738426309099e-07,
      "loss": 3.9381,
      "step": 191
    },
    {
      "epoch": 0.10964380041401955,
      "grad_norm": 4.102433204650879,
      "learning_rate": 3.817937070098914e-07,
      "loss": 3.7026,
      "step": 192
    },
    {
      "epoch": 0.11021486187450924,
      "grad_norm": 4.2977142333984375,
      "learning_rate": 2.9243565712400384e-07,
      "loss": 3.4472,
      "step": 193
    },
    {
      "epoch": 0.11078592333499893,
      "grad_norm": 4.350393295288086,
      "learning_rate": 2.1493020866542365e-07,
      "loss": 3.5126,
      "step": 194
    },
    {
      "epoch": 0.11135698479548861,
      "grad_norm": 4.65367317199707,
      "learning_rate": 1.4930382967379363e-07,
      "loss": 3.6799,
      "step": 195
    },
    {
      "epoch": 0.1119280462559783,
      "grad_norm": 4.893271446228027,
      "learning_rate": 9.557893149741924e-08,
      "loss": 3.6397,
      "step": 196
    },
    {
      "epoch": 0.11249910771646798,
      "grad_norm": 4.931285381317139,
      "learning_rate": 5.377386113981197e-08,
      "loss": 3.6552,
      "step": 197
    },
    {
      "epoch": 0.11307016917695767,
      "grad_norm": 5.327000141143799,
      "learning_rate": 2.3902894994198286e-08,
      "loss": 3.3647,
      "step": 198
    },
    {
      "epoch": 0.11364123063744735,
      "grad_norm": 6.226569175720215,
      "learning_rate": 5.976233968155164e-09,
      "loss": 3.279,
      "step": 199
    },
    {
      "epoch": 0.11421229209793704,
      "grad_norm": 8.364439010620117,
      "learning_rate": 0.0,
      "loss": 3.8711,
      "step": 200
    },
    {
      "epoch": 0.11421229209793704,
      "eval_loss": 0.4560202658176422,
      "eval_runtime": 596.8378,
      "eval_samples_per_second": 4.943,
      "eval_steps_per_second": 1.237,
      "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": 4.849985165910344e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}