{
  "best_metric": 1.6908611059188843,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.2581477896095515,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0012907389480477573,
      "grad_norm": 2.2166810035705566,
      "learning_rate": 5e-06,
      "loss": 2.0499,
      "step": 1
    },
    {
      "epoch": 0.0012907389480477573,
      "eval_loss": 3.3415772914886475,
      "eval_runtime": 104.3878,
      "eval_samples_per_second": 12.501,
      "eval_steps_per_second": 6.256,
      "step": 1
    },
    {
      "epoch": 0.0025814778960955146,
      "grad_norm": 2.6439626216888428,
      "learning_rate": 1e-05,
      "loss": 2.3745,
      "step": 2
    },
    {
      "epoch": 0.003872216844143272,
      "grad_norm": 2.6274795532226562,
      "learning_rate": 1.5e-05,
      "loss": 2.3509,
      "step": 3
    },
    {
      "epoch": 0.005162955792191029,
      "grad_norm": 2.625005006790161,
      "learning_rate": 2e-05,
      "loss": 2.3948,
      "step": 4
    },
    {
      "epoch": 0.006453694740238786,
      "grad_norm": 2.6761314868927,
      "learning_rate": 2.5e-05,
      "loss": 2.3423,
      "step": 5
    },
    {
      "epoch": 0.007744433688286544,
      "grad_norm": 2.6907224655151367,
      "learning_rate": 3e-05,
      "loss": 2.6107,
      "step": 6
    },
    {
      "epoch": 0.0090351726363343,
      "grad_norm": 3.0346169471740723,
      "learning_rate": 3.5e-05,
      "loss": 2.2783,
      "step": 7
    },
    {
      "epoch": 0.010325911584382058,
      "grad_norm": 3.174612522125244,
      "learning_rate": 4e-05,
      "loss": 2.3886,
      "step": 8
    },
    {
      "epoch": 0.011616650532429816,
      "grad_norm": 2.7802975177764893,
      "learning_rate": 4.5e-05,
      "loss": 2.2565,
      "step": 9
    },
    {
      "epoch": 0.012907389480477573,
      "grad_norm": 2.869988203048706,
      "learning_rate": 5e-05,
      "loss": 1.9725,
      "step": 10
    },
    {
      "epoch": 0.01419812842852533,
      "grad_norm": 3.1507372856140137,
      "learning_rate": 5.500000000000001e-05,
      "loss": 2.072,
      "step": 11
    },
    {
      "epoch": 0.015488867376573089,
      "grad_norm": 3.5297839641571045,
      "learning_rate": 6e-05,
      "loss": 2.1756,
      "step": 12
    },
    {
      "epoch": 0.016779606324620847,
      "grad_norm": 3.8862404823303223,
      "learning_rate": 6.500000000000001e-05,
      "loss": 2.1328,
      "step": 13
    },
    {
      "epoch": 0.0180703452726686,
      "grad_norm": 3.4827890396118164,
      "learning_rate": 7e-05,
      "loss": 1.972,
      "step": 14
    },
    {
      "epoch": 0.01936108422071636,
      "grad_norm": 3.575754165649414,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.7775,
      "step": 15
    },
    {
      "epoch": 0.020651823168764117,
      "grad_norm": 2.96850323677063,
      "learning_rate": 8e-05,
      "loss": 1.8565,
      "step": 16
    },
    {
      "epoch": 0.021942562116811875,
      "grad_norm": 3.2041714191436768,
      "learning_rate": 8.5e-05,
      "loss": 1.7362,
      "step": 17
    },
    {
      "epoch": 0.023233301064859633,
      "grad_norm": 3.659710168838501,
      "learning_rate": 9e-05,
      "loss": 1.6243,
      "step": 18
    },
    {
      "epoch": 0.02452404001290739,
      "grad_norm": 3.0214896202087402,
      "learning_rate": 9.5e-05,
      "loss": 1.798,
      "step": 19
    },
    {
      "epoch": 0.025814778960955145,
      "grad_norm": 2.942141056060791,
      "learning_rate": 0.0001,
      "loss": 1.6809,
      "step": 20
    },
    {
      "epoch": 0.027105517909002903,
      "grad_norm": 2.947040319442749,
      "learning_rate": 9.999238475781957e-05,
      "loss": 1.5544,
      "step": 21
    },
    {
      "epoch": 0.02839625685705066,
      "grad_norm": 2.7564430236816406,
      "learning_rate": 9.99695413509548e-05,
      "loss": 1.9095,
      "step": 22
    },
    {
      "epoch": 0.02968699580509842,
      "grad_norm": 2.6305737495422363,
      "learning_rate": 9.99314767377287e-05,
      "loss": 1.7639,
      "step": 23
    },
    {
      "epoch": 0.030977734753146177,
      "grad_norm": 2.743706464767456,
      "learning_rate": 9.987820251299122e-05,
      "loss": 1.7814,
      "step": 24
    },
    {
      "epoch": 0.032268473701193935,
      "grad_norm": 2.832125663757324,
      "learning_rate": 9.980973490458728e-05,
      "loss": 1.5542,
      "step": 25
    },
    {
      "epoch": 0.03355921264924169,
      "grad_norm": 2.9831314086914062,
      "learning_rate": 9.972609476841367e-05,
      "loss": 1.8095,
      "step": 26
    },
    {
      "epoch": 0.03484995159728945,
      "grad_norm": 3.3297057151794434,
      "learning_rate": 9.962730758206611e-05,
      "loss": 1.6017,
      "step": 27
    },
    {
      "epoch": 0.0361406905453372,
      "grad_norm": 3.300830602645874,
      "learning_rate": 9.951340343707852e-05,
      "loss": 1.7396,
      "step": 28
    },
    {
      "epoch": 0.03743142949338496,
      "grad_norm": 2.9726502895355225,
      "learning_rate": 9.938441702975689e-05,
      "loss": 1.7024,
      "step": 29
    },
    {
      "epoch": 0.03872216844143272,
      "grad_norm": 2.793461799621582,
      "learning_rate": 9.924038765061042e-05,
      "loss": 1.6474,
      "step": 30
    },
    {
      "epoch": 0.040012907389480476,
      "grad_norm": 3.061504602432251,
      "learning_rate": 9.908135917238321e-05,
      "loss": 1.7849,
      "step": 31
    },
    {
      "epoch": 0.041303646337528234,
      "grad_norm": 2.6581826210021973,
      "learning_rate": 9.890738003669029e-05,
      "loss": 1.6826,
      "step": 32
    },
    {
      "epoch": 0.04259438528557599,
      "grad_norm": 2.923095703125,
      "learning_rate": 9.871850323926177e-05,
      "loss": 1.8471,
      "step": 33
    },
    {
      "epoch": 0.04388512423362375,
      "grad_norm": 3.102031946182251,
      "learning_rate": 9.851478631379982e-05,
      "loss": 1.5859,
      "step": 34
    },
    {
      "epoch": 0.04517586318167151,
      "grad_norm": 2.6008760929107666,
      "learning_rate": 9.829629131445342e-05,
      "loss": 1.7703,
      "step": 35
    },
    {
      "epoch": 0.046466602129719266,
      "grad_norm": 2.5209734439849854,
      "learning_rate": 9.806308479691595e-05,
      "loss": 1.9813,
      "step": 36
    },
    {
      "epoch": 0.047757341077767024,
      "grad_norm": 3.011399507522583,
      "learning_rate": 9.781523779815179e-05,
      "loss": 1.5864,
      "step": 37
    },
    {
      "epoch": 0.04904808002581478,
      "grad_norm": 3.1950185298919678,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.7848,
      "step": 38
    },
    {
      "epoch": 0.05033881897386254,
      "grad_norm": 3.280881643295288,
      "learning_rate": 9.727592877996585e-05,
      "loss": 2.119,
      "step": 39
    },
    {
      "epoch": 0.05162955792191029,
      "grad_norm": 3.732508420944214,
      "learning_rate": 9.698463103929542e-05,
      "loss": 2.2015,
      "step": 40
    },
    {
      "epoch": 0.05292029686995805,
      "grad_norm": 4.052702903747559,
      "learning_rate": 9.667902132486009e-05,
      "loss": 1.9312,
      "step": 41
    },
    {
      "epoch": 0.05421103581800581,
      "grad_norm": 3.4176535606384277,
      "learning_rate": 9.635919272833938e-05,
      "loss": 1.952,
      "step": 42
    },
    {
      "epoch": 0.055501774766053565,
      "grad_norm": 3.460026502609253,
      "learning_rate": 9.602524267262203e-05,
      "loss": 1.8142,
      "step": 43
    },
    {
      "epoch": 0.05679251371410132,
      "grad_norm": 3.964930295944214,
      "learning_rate": 9.567727288213005e-05,
      "loss": 1.8545,
      "step": 44
    },
    {
      "epoch": 0.05808325266214908,
      "grad_norm": 3.852332830429077,
      "learning_rate": 9.53153893518325e-05,
      "loss": 1.9221,
      "step": 45
    },
    {
      "epoch": 0.05937399161019684,
      "grad_norm": 3.6937999725341797,
      "learning_rate": 9.493970231495835e-05,
      "loss": 1.9574,
      "step": 46
    },
    {
      "epoch": 0.060664730558244596,
      "grad_norm": 4.124794006347656,
      "learning_rate": 9.45503262094184e-05,
      "loss": 2.039,
      "step": 47
    },
    {
      "epoch": 0.061955469506292354,
      "grad_norm": 3.9496965408325195,
      "learning_rate": 9.414737964294636e-05,
      "loss": 2.2507,
      "step": 48
    },
    {
      "epoch": 0.0632462084543401,
      "grad_norm": 5.036445140838623,
      "learning_rate": 9.373098535696979e-05,
      "loss": 2.7065,
      "step": 49
    },
    {
      "epoch": 0.06453694740238787,
      "grad_norm": 7.1888203620910645,
      "learning_rate": 9.330127018922194e-05,
      "loss": 2.9681,
      "step": 50
    },
    {
      "epoch": 0.06453694740238787,
      "eval_loss": 1.908481478691101,
      "eval_runtime": 106.1031,
      "eval_samples_per_second": 12.299,
      "eval_steps_per_second": 6.154,
      "step": 50
    },
    {
      "epoch": 0.06582768635043562,
      "grad_norm": 3.540147304534912,
      "learning_rate": 9.285836503510562e-05,
      "loss": 1.7768,
      "step": 51
    },
    {
      "epoch": 0.06711842529848339,
      "grad_norm": 3.4239327907562256,
      "learning_rate": 9.24024048078213e-05,
      "loss": 1.8248,
      "step": 52
    },
    {
      "epoch": 0.06840916424653114,
      "grad_norm": 2.5307228565216064,
      "learning_rate": 9.193352839727121e-05,
      "loss": 1.7018,
      "step": 53
    },
    {
      "epoch": 0.0696999031945789,
      "grad_norm": 2.3973517417907715,
      "learning_rate": 9.145187862775209e-05,
      "loss": 1.6536,
      "step": 54
    },
    {
      "epoch": 0.07099064214262665,
      "grad_norm": 2.189422369003296,
      "learning_rate": 9.09576022144496e-05,
      "loss": 1.6938,
      "step": 55
    },
    {
      "epoch": 0.0722813810906744,
      "grad_norm": 2.136927366256714,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5783,
      "step": 56
    },
    {
      "epoch": 0.07357212003872217,
      "grad_norm": 2.03843355178833,
      "learning_rate": 8.993177550236464e-05,
      "loss": 1.2809,
      "step": 57
    },
    {
      "epoch": 0.07486285898676992,
      "grad_norm": 2.4228107929229736,
      "learning_rate": 8.940053768033609e-05,
      "loss": 1.5598,
      "step": 58
    },
    {
      "epoch": 0.07615359793481769,
      "grad_norm": 2.2377545833587646,
      "learning_rate": 8.885729807284856e-05,
      "loss": 1.5813,
      "step": 59
    },
    {
      "epoch": 0.07744433688286544,
      "grad_norm": 2.3037893772125244,
      "learning_rate": 8.83022221559489e-05,
      "loss": 1.7104,
      "step": 60
    },
    {
      "epoch": 0.0787350758309132,
      "grad_norm": 2.3972346782684326,
      "learning_rate": 8.773547901113862e-05,
      "loss": 1.5551,
      "step": 61
    },
    {
      "epoch": 0.08002581477896095,
      "grad_norm": 2.1127376556396484,
      "learning_rate": 8.715724127386972e-05,
      "loss": 1.8186,
      "step": 62
    },
    {
      "epoch": 0.08131655372700872,
      "grad_norm": 2.3053057193756104,
      "learning_rate": 8.656768508095853e-05,
      "loss": 1.3372,
      "step": 63
    },
    {
      "epoch": 0.08260729267505647,
      "grad_norm": 2.5323338508605957,
      "learning_rate": 8.596699001693255e-05,
      "loss": 1.6374,
      "step": 64
    },
    {
      "epoch": 0.08389803162310423,
      "grad_norm": 2.2257132530212402,
      "learning_rate": 8.535533905932738e-05,
      "loss": 1.4145,
      "step": 65
    },
    {
      "epoch": 0.08518877057115198,
      "grad_norm": 2.6799635887145996,
      "learning_rate": 8.473291852294987e-05,
      "loss": 1.6006,
      "step": 66
    },
    {
      "epoch": 0.08647950951919975,
      "grad_norm": 2.298165798187256,
      "learning_rate": 8.409991800312493e-05,
      "loss": 1.586,
      "step": 67
    },
    {
      "epoch": 0.0877702484672475,
      "grad_norm": 2.6040775775909424,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.6148,
      "step": 68
    },
    {
      "epoch": 0.08906098741529525,
      "grad_norm": 2.55844783782959,
      "learning_rate": 8.280295144952536e-05,
      "loss": 1.3866,
      "step": 69
    },
    {
      "epoch": 0.09035172636334302,
      "grad_norm": 2.483966112136841,
      "learning_rate": 8.213938048432697e-05,
      "loss": 1.4356,
      "step": 70
    },
    {
      "epoch": 0.09164246531139077,
      "grad_norm": 2.592611074447632,
      "learning_rate": 8.146601955249188e-05,
      "loss": 1.5875,
      "step": 71
    },
    {
      "epoch": 0.09293320425943853,
      "grad_norm": 2.5707600116729736,
      "learning_rate": 8.07830737662829e-05,
      "loss": 1.6118,
      "step": 72
    },
    {
      "epoch": 0.09422394320748628,
      "grad_norm": 2.953272819519043,
      "learning_rate": 8.009075115760243e-05,
      "loss": 1.6945,
      "step": 73
    },
    {
      "epoch": 0.09551468215553405,
      "grad_norm": 2.6150481700897217,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.6591,
      "step": 74
    },
    {
      "epoch": 0.0968054211035818,
      "grad_norm": 2.6085891723632812,
      "learning_rate": 7.86788218175523e-05,
      "loss": 1.6226,
      "step": 75
    },
    {
      "epoch": 0.09809616005162956,
      "grad_norm": 2.596640110015869,
      "learning_rate": 7.795964517353735e-05,
      "loss": 1.5355,
      "step": 76
    },
    {
      "epoch": 0.09938689899967731,
      "grad_norm": 2.2675673961639404,
      "learning_rate": 7.723195175075136e-05,
      "loss": 1.8316,
      "step": 77
    },
    {
      "epoch": 0.10067763794772508,
      "grad_norm": 2.3651790618896484,
      "learning_rate": 7.649596321166024e-05,
      "loss": 1.7175,
      "step": 78
    },
    {
      "epoch": 0.10196837689577283,
      "grad_norm": 2.2982771396636963,
      "learning_rate": 7.575190374550272e-05,
      "loss": 1.8389,
      "step": 79
    },
    {
      "epoch": 0.10325911584382058,
      "grad_norm": 2.134028196334839,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.7727,
      "step": 80
    },
    {
      "epoch": 0.10454985479186835,
      "grad_norm": 2.5334267616271973,
      "learning_rate": 7.424048101231686e-05,
      "loss": 1.6886,
      "step": 81
    },
    {
      "epoch": 0.1058405937399161,
      "grad_norm": 2.494502067565918,
      "learning_rate": 7.347357813929454e-05,
      "loss": 1.7609,
      "step": 82
    },
    {
      "epoch": 0.10713133268796386,
      "grad_norm": 2.443568468093872,
      "learning_rate": 7.269952498697734e-05,
      "loss": 1.6298,
      "step": 83
    },
    {
      "epoch": 0.10842207163601161,
      "grad_norm": 2.6569809913635254,
      "learning_rate": 7.191855733945387e-05,
      "loss": 1.7955,
      "step": 84
    },
    {
      "epoch": 0.10971281058405938,
      "grad_norm": 2.6799306869506836,
      "learning_rate": 7.113091308703498e-05,
      "loss": 1.9591,
      "step": 85
    },
    {
      "epoch": 0.11100354953210713,
      "grad_norm": 2.3563599586486816,
      "learning_rate": 7.033683215379002e-05,
      "loss": 1.8815,
      "step": 86
    },
    {
      "epoch": 0.1122942884801549,
      "grad_norm": 2.684368133544922,
      "learning_rate": 6.953655642446368e-05,
      "loss": 1.7106,
      "step": 87
    },
    {
      "epoch": 0.11358502742820265,
      "grad_norm": 2.620560646057129,
      "learning_rate": 6.873032967079561e-05,
      "loss": 1.8121,
      "step": 88
    },
    {
      "epoch": 0.11487576637625041,
      "grad_norm": 2.813100814819336,
      "learning_rate": 6.7918397477265e-05,
      "loss": 1.6104,
      "step": 89
    },
    {
      "epoch": 0.11616650532429816,
      "grad_norm": 2.898761034011841,
      "learning_rate": 6.710100716628344e-05,
      "loss": 1.6963,
      "step": 90
    },
    {
      "epoch": 0.11745724427234591,
      "grad_norm": 2.6569104194641113,
      "learning_rate": 6.627840772285784e-05,
      "loss": 2.0084,
      "step": 91
    },
    {
      "epoch": 0.11874798322039368,
      "grad_norm": 2.7398810386657715,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.9913,
      "step": 92
    },
    {
      "epoch": 0.12003872216844143,
      "grad_norm": 2.800980567932129,
      "learning_rate": 6.461858523613684e-05,
      "loss": 1.9152,
      "step": 93
    },
    {
      "epoch": 0.12132946111648919,
      "grad_norm": 2.9425337314605713,
      "learning_rate": 6.378186779084995e-05,
      "loss": 1.7623,
      "step": 94
    },
    {
      "epoch": 0.12262020006453694,
      "grad_norm": 2.9889039993286133,
      "learning_rate": 6.294095225512603e-05,
      "loss": 2.115,
      "step": 95
    },
    {
      "epoch": 0.12391093901258471,
      "grad_norm": 2.6212713718414307,
      "learning_rate": 6.209609477998338e-05,
      "loss": 2.0041,
      "step": 96
    },
    {
      "epoch": 0.12520167796063247,
      "grad_norm": 3.1693074703216553,
      "learning_rate": 6.124755271719325e-05,
      "loss": 1.9586,
      "step": 97
    },
    {
      "epoch": 0.1264924169086802,
      "grad_norm": 3.050912618637085,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 2.0861,
      "step": 98
    },
    {
      "epoch": 0.12778315585672798,
      "grad_norm": 3.8726887702941895,
      "learning_rate": 5.9540449768827246e-05,
      "loss": 2.4497,
      "step": 99
    },
    {
      "epoch": 0.12907389480477574,
      "grad_norm": 5.545790195465088,
      "learning_rate": 5.868240888334653e-05,
      "loss": 2.8716,
      "step": 100
    },
    {
      "epoch": 0.12907389480477574,
      "eval_loss": 1.827061653137207,
      "eval_runtime": 106.0648,
      "eval_samples_per_second": 12.304,
      "eval_steps_per_second": 6.157,
      "step": 100
    },
    {
      "epoch": 0.13036463375282348,
      "grad_norm": 2.503622531890869,
      "learning_rate": 5.782172325201155e-05,
      "loss": 1.5123,
      "step": 101
    },
    {
      "epoch": 0.13165537270087124,
      "grad_norm": 3.0901057720184326,
      "learning_rate": 5.695865504800327e-05,
      "loss": 1.7133,
      "step": 102
    },
    {
      "epoch": 0.132946111648919,
      "grad_norm": 2.7562739849090576,
      "learning_rate": 5.6093467170257374e-05,
      "loss": 1.6712,
      "step": 103
    },
    {
      "epoch": 0.13423685059696677,
      "grad_norm": 2.6895506381988525,
      "learning_rate": 5.522642316338268e-05,
      "loss": 1.5337,
      "step": 104
    },
    {
      "epoch": 0.1355275895450145,
      "grad_norm": 2.453566074371338,
      "learning_rate": 5.435778713738292e-05,
      "loss": 1.6975,
      "step": 105
    },
    {
      "epoch": 0.13681832849306227,
      "grad_norm": 2.23386287689209,
      "learning_rate": 5.348782368720626e-05,
      "loss": 1.8449,
      "step": 106
    },
    {
      "epoch": 0.13810906744111004,
      "grad_norm": 2.1461331844329834,
      "learning_rate": 5.26167978121472e-05,
      "loss": 1.3803,
      "step": 107
    },
    {
      "epoch": 0.1393998063891578,
      "grad_norm": 2.254761219024658,
      "learning_rate": 5.174497483512506e-05,
      "loss": 1.6804,
      "step": 108
    },
    {
      "epoch": 0.14069054533720554,
      "grad_norm": 2.3938405513763428,
      "learning_rate": 5.0872620321864185e-05,
      "loss": 1.7504,
      "step": 109
    },
    {
      "epoch": 0.1419812842852533,
      "grad_norm": 2.0230281352996826,
      "learning_rate": 5e-05,
      "loss": 1.4922,
      "step": 110
    },
    {
      "epoch": 0.14327202323330107,
      "grad_norm": 2.0155575275421143,
      "learning_rate": 4.912737967813583e-05,
      "loss": 1.5318,
      "step": 111
    },
    {
      "epoch": 0.1445627621813488,
      "grad_norm": 2.3189070224761963,
      "learning_rate": 4.825502516487497e-05,
      "loss": 1.2708,
      "step": 112
    },
    {
      "epoch": 0.14585350112939657,
      "grad_norm": 2.32184100151062,
      "learning_rate": 4.738320218785281e-05,
      "loss": 1.6515,
      "step": 113
    },
    {
      "epoch": 0.14714424007744434,
      "grad_norm": 2.3235177993774414,
      "learning_rate": 4.6512176312793736e-05,
      "loss": 1.53,
      "step": 114
    },
    {
      "epoch": 0.1484349790254921,
      "grad_norm": 2.426671028137207,
      "learning_rate": 4.564221286261709e-05,
      "loss": 1.5062,
      "step": 115
    },
    {
      "epoch": 0.14972571797353984,
      "grad_norm": 2.169921398162842,
      "learning_rate": 4.477357683661734e-05,
      "loss": 1.7131,
      "step": 116
    },
    {
      "epoch": 0.1510164569215876,
      "grad_norm": 2.2873425483703613,
      "learning_rate": 4.390653282974264e-05,
      "loss": 1.8223,
      "step": 117
    },
    {
      "epoch": 0.15230719586963537,
      "grad_norm": 2.208829402923584,
      "learning_rate": 4.3041344951996746e-05,
      "loss": 1.4369,
      "step": 118
    },
    {
      "epoch": 0.15359793481768313,
      "grad_norm": 2.6913652420043945,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 1.5725,
      "step": 119
    },
    {
      "epoch": 0.15488867376573087,
      "grad_norm": 2.1297287940979004,
      "learning_rate": 4.131759111665349e-05,
      "loss": 1.7007,
      "step": 120
    },
    {
      "epoch": 0.15617941271377864,
      "grad_norm": 2.1201767921447754,
      "learning_rate": 4.045955023117276e-05,
      "loss": 1.4326,
      "step": 121
    },
    {
      "epoch": 0.1574701516618264,
      "grad_norm": 2.3217813968658447,
      "learning_rate": 3.960441545911204e-05,
      "loss": 1.7533,
      "step": 122
    },
    {
      "epoch": 0.15876089060987417,
      "grad_norm": 2.3463990688323975,
      "learning_rate": 3.875244728280676e-05,
      "loss": 1.6176,
      "step": 123
    },
    {
      "epoch": 0.1600516295579219,
      "grad_norm": 2.1427035331726074,
      "learning_rate": 3.790390522001662e-05,
      "loss": 1.6453,
      "step": 124
    },
    {
      "epoch": 0.16134236850596967,
      "grad_norm": 2.5846805572509766,
      "learning_rate": 3.705904774487396e-05,
      "loss": 1.6022,
      "step": 125
    },
    {
      "epoch": 0.16263310745401743,
      "grad_norm": 2.215284824371338,
      "learning_rate": 3.6218132209150045e-05,
      "loss": 1.5465,
      "step": 126
    },
    {
      "epoch": 0.16392384640206517,
      "grad_norm": 2.5978410243988037,
      "learning_rate": 3.5381414763863166e-05,
      "loss": 1.802,
      "step": 127
    },
    {
      "epoch": 0.16521458535011294,
      "grad_norm": 2.1787612438201904,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.6055,
      "step": 128
    },
    {
      "epoch": 0.1665053242981607,
      "grad_norm": 2.518327236175537,
      "learning_rate": 3.372159227714218e-05,
      "loss": 1.6895,
      "step": 129
    },
    {
      "epoch": 0.16779606324620847,
      "grad_norm": 2.384491443634033,
      "learning_rate": 3.289899283371657e-05,
      "loss": 1.3833,
      "step": 130
    },
    {
      "epoch": 0.1690868021942562,
      "grad_norm": 2.4606215953826904,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 1.5171,
      "step": 131
    },
    {
      "epoch": 0.17037754114230397,
      "grad_norm": 2.3431456089019775,
      "learning_rate": 3.12696703292044e-05,
      "loss": 1.5017,
      "step": 132
    },
    {
      "epoch": 0.17166828009035173,
      "grad_norm": 2.696519374847412,
      "learning_rate": 3.046344357553632e-05,
      "loss": 1.825,
      "step": 133
    },
    {
      "epoch": 0.1729590190383995,
      "grad_norm": 2.515021324157715,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 1.6135,
      "step": 134
    },
    {
      "epoch": 0.17424975798644723,
      "grad_norm": 2.7084922790527344,
      "learning_rate": 2.886908691296504e-05,
      "loss": 1.7145,
      "step": 135
    },
    {
      "epoch": 0.175540496934495,
      "grad_norm": 2.8446033000946045,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 1.8448,
      "step": 136
    },
    {
      "epoch": 0.17683123588254276,
      "grad_norm": 2.9814133644104004,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 1.72,
      "step": 137
    },
    {
      "epoch": 0.1781219748305905,
      "grad_norm": 2.513272285461426,
      "learning_rate": 2.6526421860705473e-05,
      "loss": 1.9783,
      "step": 138
    },
    {
      "epoch": 0.17941271377863827,
      "grad_norm": 2.867854595184326,
      "learning_rate": 2.575951898768315e-05,
      "loss": 1.8063,
      "step": 139
    },
    {
      "epoch": 0.18070345272668603,
      "grad_norm": 2.507193088531494,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.556,
      "step": 140
    },
    {
      "epoch": 0.1819941916747338,
      "grad_norm": 2.899620771408081,
      "learning_rate": 2.4248096254497288e-05,
      "loss": 1.704,
      "step": 141
    },
    {
      "epoch": 0.18328493062278153,
      "grad_norm": 2.8599815368652344,
      "learning_rate": 2.350403678833976e-05,
      "loss": 1.9382,
      "step": 142
    },
    {
      "epoch": 0.1845756695708293,
      "grad_norm": 2.761918067932129,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 1.779,
      "step": 143
    },
    {
      "epoch": 0.18586640851887706,
      "grad_norm": 3.3469927310943604,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 1.7328,
      "step": 144
    },
    {
      "epoch": 0.18715714746692483,
      "grad_norm": 2.946146249771118,
      "learning_rate": 2.132117818244771e-05,
      "loss": 1.9915,
      "step": 145
    },
    {
      "epoch": 0.18844788641497257,
      "grad_norm": 3.1667943000793457,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.8595,
      "step": 146
    },
    {
      "epoch": 0.18973862536302033,
      "grad_norm": 4.6375555992126465,
      "learning_rate": 1.9909248842397584e-05,
      "loss": 1.5824,
      "step": 147
    },
    {
      "epoch": 0.1910293643110681,
      "grad_norm": 3.8069305419921875,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 2.0351,
      "step": 148
    },
    {
      "epoch": 0.19232010325911583,
      "grad_norm": 4.175755023956299,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 2.3699,
      "step": 149
    },
    {
      "epoch": 0.1936108422071636,
      "grad_norm": 5.496400833129883,
      "learning_rate": 1.7860619515673033e-05,
      "loss": 2.7766,
      "step": 150
    },
    {
      "epoch": 0.1936108422071636,
      "eval_loss": 1.7084441184997559,
      "eval_runtime": 106.226,
      "eval_samples_per_second": 12.285,
      "eval_steps_per_second": 6.147,
      "step": 150
    },
    {
      "epoch": 0.19490158115521136,
      "grad_norm": 1.8199388980865479,
      "learning_rate": 1.7197048550474643e-05,
      "loss": 1.5286,
      "step": 151
    },
    {
      "epoch": 0.19619232010325913,
      "grad_norm": 1.9473209381103516,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 1.5828,
      "step": 152
    },
    {
      "epoch": 0.19748305905130686,
      "grad_norm": 2.1355364322662354,
      "learning_rate": 1.5900081996875083e-05,
      "loss": 1.507,
      "step": 153
    },
    {
      "epoch": 0.19877379799935463,
      "grad_norm": 2.227328300476074,
      "learning_rate": 1.526708147705013e-05,
      "loss": 1.5442,
      "step": 154
    },
    {
      "epoch": 0.2000645369474024,
      "grad_norm": 1.9992481470108032,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 1.3718,
      "step": 155
    },
    {
      "epoch": 0.20135527589545016,
      "grad_norm": 1.9334080219268799,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 1.6259,
      "step": 156
    },
    {
      "epoch": 0.2026460148434979,
      "grad_norm": 2.0036137104034424,
      "learning_rate": 1.3432314919041478e-05,
      "loss": 1.716,
      "step": 157
    },
    {
      "epoch": 0.20393675379154566,
      "grad_norm": 1.9384994506835938,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 1.371,
      "step": 158
    },
    {
      "epoch": 0.20522749273959343,
      "grad_norm": 2.747986316680908,
      "learning_rate": 1.22645209888614e-05,
      "loss": 1.5368,
      "step": 159
    },
    {
      "epoch": 0.20651823168764116,
      "grad_norm": 2.3735859394073486,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 1.319,
      "step": 160
    },
    {
      "epoch": 0.20780897063568893,
      "grad_norm": 2.167597770690918,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 1.5141,
      "step": 161
    },
    {
      "epoch": 0.2090997095837367,
      "grad_norm": 2.1857247352600098,
      "learning_rate": 1.0599462319663905e-05,
      "loss": 1.5138,
      "step": 162
    },
    {
      "epoch": 0.21039044853178446,
      "grad_norm": 2.2942934036254883,
      "learning_rate": 1.006822449763537e-05,
      "loss": 1.5684,
      "step": 163
    },
    {
      "epoch": 0.2116811874798322,
      "grad_norm": 2.193618059158325,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.5603,
      "step": 164
    },
    {
      "epoch": 0.21297192642787996,
      "grad_norm": 1.9363298416137695,
      "learning_rate": 9.042397785550405e-06,
      "loss": 1.7234,
      "step": 165
    },
    {
      "epoch": 0.21426266537592772,
      "grad_norm": 2.138582944869995,
      "learning_rate": 8.548121372247918e-06,
      "loss": 1.5614,
      "step": 166
    },
    {
      "epoch": 0.2155534043239755,
      "grad_norm": 2.4650990962982178,
      "learning_rate": 8.066471602728803e-06,
      "loss": 1.6118,
      "step": 167
    },
    {
      "epoch": 0.21684414327202323,
      "grad_norm": 2.180103302001953,
      "learning_rate": 7.597595192178702e-06,
      "loss": 1.4313,
      "step": 168
    },
    {
      "epoch": 0.218134882220071,
      "grad_norm": 2.1173620223999023,
      "learning_rate": 7.1416349648943894e-06,
      "loss": 1.456,
      "step": 169
    },
    {
      "epoch": 0.21942562116811876,
      "grad_norm": 2.1869735717773438,
      "learning_rate": 6.698729810778065e-06,
      "loss": 1.4179,
      "step": 170
    },
    {
      "epoch": 0.2207163601161665,
      "grad_norm": 2.301262617111206,
      "learning_rate": 6.269014643030213e-06,
      "loss": 1.4807,
      "step": 171
    },
    {
      "epoch": 0.22200709906421426,
      "grad_norm": 2.313802480697632,
      "learning_rate": 5.852620357053651e-06,
      "loss": 1.5051,
      "step": 172
    },
    {
      "epoch": 0.22329783801226202,
      "grad_norm": 2.1395950317382812,
      "learning_rate": 5.449673790581611e-06,
      "loss": 1.5932,
      "step": 173
    },
    {
      "epoch": 0.2245885769603098,
      "grad_norm": 2.3838601112365723,
      "learning_rate": 5.060297685041659e-06,
      "loss": 1.7347,
      "step": 174
    },
    {
      "epoch": 0.22587931590835753,
      "grad_norm": 2.4559905529022217,
      "learning_rate": 4.684610648167503e-06,
      "loss": 1.547,
      "step": 175
    },
    {
      "epoch": 0.2271700548564053,
      "grad_norm": 2.238723039627075,
      "learning_rate": 4.322727117869951e-06,
      "loss": 1.3255,
      "step": 176
    },
    {
      "epoch": 0.22846079380445306,
      "grad_norm": 2.7607717514038086,
      "learning_rate": 3.974757327377981e-06,
      "loss": 1.5829,
      "step": 177
    },
    {
      "epoch": 0.22975153275250082,
      "grad_norm": 2.4810197353363037,
      "learning_rate": 3.6408072716606346e-06,
      "loss": 1.8075,
      "step": 178
    },
    {
      "epoch": 0.23104227170054856,
      "grad_norm": 2.4974825382232666,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 1.5672,
      "step": 179
    },
    {
      "epoch": 0.23233301064859632,
      "grad_norm": 2.491542339324951,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 1.5251,
      "step": 180
    },
    {
      "epoch": 0.2336237495966441,
      "grad_norm": 2.520451545715332,
      "learning_rate": 2.724071220034158e-06,
      "loss": 1.693,
      "step": 181
    },
    {
      "epoch": 0.23491448854469182,
      "grad_norm": 2.5069587230682373,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.2409,
      "step": 182
    },
    {
      "epoch": 0.2362052274927396,
      "grad_norm": 2.4972262382507324,
      "learning_rate": 2.1847622018482283e-06,
      "loss": 1.8058,
      "step": 183
    },
    {
      "epoch": 0.23749596644078735,
      "grad_norm": 2.5082268714904785,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 1.5417,
      "step": 184
    },
    {
      "epoch": 0.23878670538883512,
      "grad_norm": 2.5240094661712646,
      "learning_rate": 1.70370868554659e-06,
      "loss": 1.8962,
      "step": 185
    },
    {
      "epoch": 0.24007744433688286,
      "grad_norm": 2.768125534057617,
      "learning_rate": 1.4852136862001764e-06,
      "loss": 1.7997,
      "step": 186
    },
    {
      "epoch": 0.24136818328493062,
      "grad_norm": 2.2773830890655518,
      "learning_rate": 1.2814967607382432e-06,
      "loss": 1.725,
      "step": 187
    },
    {
      "epoch": 0.24265892223297839,
      "grad_norm": 2.690159797668457,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 1.7546,
      "step": 188
    },
    {
      "epoch": 0.24394966118102615,
      "grad_norm": 2.6133649349212646,
      "learning_rate": 9.186408276168013e-07,
      "loss": 1.5532,
      "step": 189
    },
    {
      "epoch": 0.2452404001290739,
      "grad_norm": 2.8169918060302734,
      "learning_rate": 7.596123493895991e-07,
      "loss": 2.0924,
      "step": 190
    },
    {
      "epoch": 0.24653113907712165,
      "grad_norm": 2.8741047382354736,
      "learning_rate": 6.15582970243117e-07,
      "loss": 2.1608,
      "step": 191
    },
    {
      "epoch": 0.24782187802516942,
      "grad_norm": 3.2508883476257324,
      "learning_rate": 4.865965629214819e-07,
      "loss": 1.6941,
      "step": 192
    },
    {
      "epoch": 0.24911261697321715,
      "grad_norm": 2.766706705093384,
      "learning_rate": 3.7269241793390085e-07,
      "loss": 1.7249,
      "step": 193
    },
    {
      "epoch": 0.25040335592126495,
      "grad_norm": 3.09800386428833,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 2.0962,
      "step": 194
    },
    {
      "epoch": 0.25169409486931266,
      "grad_norm": 3.3121016025543213,
      "learning_rate": 1.9026509541272275e-07,
      "loss": 1.9174,
      "step": 195
    },
    {
      "epoch": 0.2529848338173604,
      "grad_norm": 2.8629212379455566,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 1.7519,
      "step": 196
    },
    {
      "epoch": 0.2542755727654082,
      "grad_norm": 3.389427900314331,
      "learning_rate": 6.852326227130834e-08,
      "loss": 2.0535,
      "step": 197
    },
    {
      "epoch": 0.25556631171345595,
      "grad_norm": 3.361931324005127,
      "learning_rate": 3.04586490452119e-08,
      "loss": 2.2871,
      "step": 198
    },
    {
      "epoch": 0.2568570506615037,
      "grad_norm": 3.408776044845581,
      "learning_rate": 7.615242180436522e-09,
      "loss": 2.2055,
      "step": 199
    },
    {
      "epoch": 0.2581477896095515,
      "grad_norm": 5.525600433349609,
      "learning_rate": 0.0,
      "loss": 2.6682,
      "step": 200
    },
    {
      "epoch": 0.2581477896095515,
      "eval_loss": 1.6908611059188843,
      "eval_runtime": 106.2598,
      "eval_samples_per_second": 12.281,
      "eval_steps_per_second": 6.145,
      "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": 3.017027157491712e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}