{
  "best_metric": 0.5849894285202026,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.7554296506137866,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003777148253068933,
      "grad_norm": 5.133665084838867,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.4055,
      "step": 1
    },
    {
      "epoch": 0.003777148253068933,
      "eval_loss": 2.09171986579895,
      "eval_runtime": 11.0131,
      "eval_samples_per_second": 40.497,
      "eval_steps_per_second": 20.249,
      "step": 1
    },
    {
      "epoch": 0.007554296506137866,
      "grad_norm": 5.148059368133545,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.4092,
      "step": 2
    },
    {
      "epoch": 0.0113314447592068,
      "grad_norm": 5.043951511383057,
      "learning_rate": 1e-05,
      "loss": 1.5684,
      "step": 3
    },
    {
      "epoch": 0.015108593012275733,
      "grad_norm": 6.672359466552734,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.4773,
      "step": 4
    },
    {
      "epoch": 0.018885741265344664,
      "grad_norm": 4.942115306854248,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.4355,
      "step": 5
    },
    {
      "epoch": 0.0226628895184136,
      "grad_norm": 3.8950068950653076,
      "learning_rate": 2e-05,
      "loss": 1.4504,
      "step": 6
    },
    {
      "epoch": 0.02644003777148253,
      "grad_norm": 3.630852222442627,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 1.1864,
      "step": 7
    },
    {
      "epoch": 0.030217186024551465,
      "grad_norm": 3.320261001586914,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 1.2669,
      "step": 8
    },
    {
      "epoch": 0.0339943342776204,
      "grad_norm": 3.4972527027130127,
      "learning_rate": 3e-05,
      "loss": 1.2724,
      "step": 9
    },
    {
      "epoch": 0.03777148253068933,
      "grad_norm": 3.515155792236328,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 1.2599,
      "step": 10
    },
    {
      "epoch": 0.04154863078375826,
      "grad_norm": 3.004218339920044,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 1.1397,
      "step": 11
    },
    {
      "epoch": 0.0453257790368272,
      "grad_norm": 3.0206212997436523,
      "learning_rate": 4e-05,
      "loss": 1.0672,
      "step": 12
    },
    {
      "epoch": 0.049102927289896126,
      "grad_norm": 3.522303819656372,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 1.0409,
      "step": 13
    },
    {
      "epoch": 0.05288007554296506,
      "grad_norm": 2.9816133975982666,
      "learning_rate": 4.666666666666667e-05,
      "loss": 1.0899,
      "step": 14
    },
    {
      "epoch": 0.056657223796033995,
      "grad_norm": 2.4707491397857666,
      "learning_rate": 5e-05,
      "loss": 1.074,
      "step": 15
    },
    {
      "epoch": 0.06043437204910293,
      "grad_norm": 2.226038694381714,
      "learning_rate": 5.333333333333333e-05,
      "loss": 1.0244,
      "step": 16
    },
    {
      "epoch": 0.06421152030217187,
      "grad_norm": 3.55509090423584,
      "learning_rate": 5.666666666666667e-05,
      "loss": 1.01,
      "step": 17
    },
    {
      "epoch": 0.0679886685552408,
      "grad_norm": 2.9011569023132324,
      "learning_rate": 6e-05,
      "loss": 0.98,
      "step": 18
    },
    {
      "epoch": 0.07176581680830972,
      "grad_norm": 2.353875160217285,
      "learning_rate": 6.333333333333333e-05,
      "loss": 1.0103,
      "step": 19
    },
    {
      "epoch": 0.07554296506137866,
      "grad_norm": 1.8924942016601562,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.975,
      "step": 20
    },
    {
      "epoch": 0.07932011331444759,
      "grad_norm": 2.2733869552612305,
      "learning_rate": 7e-05,
      "loss": 0.9898,
      "step": 21
    },
    {
      "epoch": 0.08309726156751653,
      "grad_norm": 2.248375654220581,
      "learning_rate": 7.333333333333333e-05,
      "loss": 0.9458,
      "step": 22
    },
    {
      "epoch": 0.08687440982058546,
      "grad_norm": 1.9557474851608276,
      "learning_rate": 7.666666666666667e-05,
      "loss": 0.9286,
      "step": 23
    },
    {
      "epoch": 0.0906515580736544,
      "grad_norm": 1.9946815967559814,
      "learning_rate": 8e-05,
      "loss": 1.0479,
      "step": 24
    },
    {
      "epoch": 0.09442870632672333,
      "grad_norm": 2.2438347339630127,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.9403,
      "step": 25
    },
    {
      "epoch": 0.09820585457979225,
      "grad_norm": 2.063390016555786,
      "learning_rate": 8.666666666666667e-05,
      "loss": 0.9661,
      "step": 26
    },
    {
      "epoch": 0.10198300283286119,
      "grad_norm": 1.9358409643173218,
      "learning_rate": 9e-05,
      "loss": 0.8136,
      "step": 27
    },
    {
      "epoch": 0.10576015108593012,
      "grad_norm": 2.092071771621704,
      "learning_rate": 9.333333333333334e-05,
      "loss": 0.9752,
      "step": 28
    },
    {
      "epoch": 0.10953729933899906,
      "grad_norm": 2.1448395252227783,
      "learning_rate": 9.666666666666667e-05,
      "loss": 0.8689,
      "step": 29
    },
    {
      "epoch": 0.11331444759206799,
      "grad_norm": 2.375199317932129,
      "learning_rate": 0.0001,
      "loss": 0.786,
      "step": 30
    },
    {
      "epoch": 0.11709159584513693,
      "grad_norm": 2.278336524963379,
      "learning_rate": 9.999146252290264e-05,
      "loss": 0.9043,
      "step": 31
    },
    {
      "epoch": 0.12086874409820586,
      "grad_norm": 2.0525617599487305,
      "learning_rate": 9.996585300715116e-05,
      "loss": 0.8588,
      "step": 32
    },
    {
      "epoch": 0.12464589235127478,
      "grad_norm": 2.440821647644043,
      "learning_rate": 9.99231801983717e-05,
      "loss": 0.8775,
      "step": 33
    },
    {
      "epoch": 0.12842304060434373,
      "grad_norm": 2.378626823425293,
      "learning_rate": 9.986345866928941e-05,
      "loss": 0.8597,
      "step": 34
    },
    {
      "epoch": 0.13220018885741266,
      "grad_norm": 2.09382700920105,
      "learning_rate": 9.978670881475172e-05,
      "loss": 0.8777,
      "step": 35
    },
    {
      "epoch": 0.1359773371104816,
      "grad_norm": 2.5226829051971436,
      "learning_rate": 9.96929568447637e-05,
      "loss": 0.8689,
      "step": 36
    },
    {
      "epoch": 0.1397544853635505,
      "grad_norm": 2.3331191539764404,
      "learning_rate": 9.958223477553714e-05,
      "loss": 0.9217,
      "step": 37
    },
    {
      "epoch": 0.14353163361661944,
      "grad_norm": 1.9742475748062134,
      "learning_rate": 9.94545804185573e-05,
      "loss": 0.7367,
      "step": 38
    },
    {
      "epoch": 0.14730878186968838,
      "grad_norm": 2.221008062362671,
      "learning_rate": 9.931003736767013e-05,
      "loss": 0.8081,
      "step": 39
    },
    {
      "epoch": 0.1510859301227573,
      "grad_norm": 2.59126877784729,
      "learning_rate": 9.91486549841951e-05,
      "loss": 0.9491,
      "step": 40
    },
    {
      "epoch": 0.15486307837582625,
      "grad_norm": 2.0537147521972656,
      "learning_rate": 9.89704883800683e-05,
      "loss": 0.7115,
      "step": 41
    },
    {
      "epoch": 0.15864022662889518,
      "grad_norm": 2.845020055770874,
      "learning_rate": 9.877559839902184e-05,
      "loss": 1.0225,
      "step": 42
    },
    {
      "epoch": 0.16241737488196412,
      "grad_norm": 2.8423526287078857,
      "learning_rate": 9.85640515958057e-05,
      "loss": 1.0292,
      "step": 43
    },
    {
      "epoch": 0.16619452313503305,
      "grad_norm": 2.9364428520202637,
      "learning_rate": 9.833592021345937e-05,
      "loss": 1.0661,
      "step": 44
    },
    {
      "epoch": 0.16997167138810199,
      "grad_norm": 3.533302068710327,
      "learning_rate": 9.809128215864097e-05,
      "loss": 1.3266,
      "step": 45
    },
    {
      "epoch": 0.17374881964117092,
      "grad_norm": 3.8001816272735596,
      "learning_rate": 9.783022097502204e-05,
      "loss": 1.2563,
      "step": 46
    },
    {
      "epoch": 0.17752596789423986,
      "grad_norm": 3.8458480834960938,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.2794,
      "step": 47
    },
    {
      "epoch": 0.1813031161473088,
      "grad_norm": 4.069799900054932,
      "learning_rate": 9.725919140804099e-05,
      "loss": 1.1926,
      "step": 48
    },
    {
      "epoch": 0.18508026440037773,
      "grad_norm": 3.908215045928955,
      "learning_rate": 9.694941803075283e-05,
      "loss": 1.2267,
      "step": 49
    },
    {
      "epoch": 0.18885741265344666,
      "grad_norm": 3.979827880859375,
      "learning_rate": 9.662361147021779e-05,
      "loss": 1.0749,
      "step": 50
    },
    {
      "epoch": 0.18885741265344666,
      "eval_loss": 1.1297073364257812,
      "eval_runtime": 10.9134,
      "eval_samples_per_second": 40.867,
      "eval_steps_per_second": 20.434,
      "step": 50
    },
    {
      "epoch": 0.19263456090651557,
      "grad_norm": 4.175277233123779,
      "learning_rate": 9.628188298907782e-05,
      "loss": 1.0744,
      "step": 51
    },
    {
      "epoch": 0.1964117091595845,
      "grad_norm": 4.684498310089111,
      "learning_rate": 9.592434928729616e-05,
      "loss": 1.1514,
      "step": 52
    },
    {
      "epoch": 0.20018885741265344,
      "grad_norm": 3.317908763885498,
      "learning_rate": 9.555113246230442e-05,
      "loss": 1.017,
      "step": 53
    },
    {
      "epoch": 0.20396600566572237,
      "grad_norm": 2.499777317047119,
      "learning_rate": 9.516235996730645e-05,
      "loss": 0.955,
      "step": 54
    },
    {
      "epoch": 0.2077431539187913,
      "grad_norm": 1.7710614204406738,
      "learning_rate": 9.475816456775313e-05,
      "loss": 0.8789,
      "step": 55
    },
    {
      "epoch": 0.21152030217186024,
      "grad_norm": 1.8278405666351318,
      "learning_rate": 9.43386842960031e-05,
      "loss": 0.8578,
      "step": 56
    },
    {
      "epoch": 0.21529745042492918,
      "grad_norm": 1.8348681926727295,
      "learning_rate": 9.39040624041849e-05,
      "loss": 0.8507,
      "step": 57
    },
    {
      "epoch": 0.2190745986779981,
      "grad_norm": 1.788751244544983,
      "learning_rate": 9.345444731527642e-05,
      "loss": 0.854,
      "step": 58
    },
    {
      "epoch": 0.22285174693106705,
      "grad_norm": 1.7203760147094727,
      "learning_rate": 9.298999257241863e-05,
      "loss": 0.7857,
      "step": 59
    },
    {
      "epoch": 0.22662889518413598,
      "grad_norm": 1.7681082487106323,
      "learning_rate": 9.251085678648072e-05,
      "loss": 0.7915,
      "step": 60
    },
    {
      "epoch": 0.23040604343720492,
      "grad_norm": 1.4763216972351074,
      "learning_rate": 9.201720358189464e-05,
      "loss": 0.7069,
      "step": 61
    },
    {
      "epoch": 0.23418319169027385,
      "grad_norm": 1.536072850227356,
      "learning_rate": 9.150920154077754e-05,
      "loss": 0.6653,
      "step": 62
    },
    {
      "epoch": 0.23796033994334279,
      "grad_norm": 1.6507514715194702,
      "learning_rate": 9.098702414536107e-05,
      "loss": 0.8034,
      "step": 63
    },
    {
      "epoch": 0.24173748819641172,
      "grad_norm": 1.5751417875289917,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.7442,
      "step": 64
    },
    {
      "epoch": 0.24551463644948066,
      "grad_norm": 1.8082925081253052,
      "learning_rate": 8.9900861364012e-05,
      "loss": 0.784,
      "step": 65
    },
    {
      "epoch": 0.24929178470254956,
      "grad_norm": 1.5581010580062866,
      "learning_rate": 8.933724690167417e-05,
      "loss": 0.7309,
      "step": 66
    },
    {
      "epoch": 0.2530689329556185,
      "grad_norm": 1.746187686920166,
      "learning_rate": 8.876019880555649e-05,
      "loss": 0.7594,
      "step": 67
    },
    {
      "epoch": 0.25684608120868746,
      "grad_norm": 1.4668420553207397,
      "learning_rate": 8.816991413705516e-05,
      "loss": 0.6373,
      "step": 68
    },
    {
      "epoch": 0.26062322946175637,
      "grad_norm": 1.521977186203003,
      "learning_rate": 8.756659447784368e-05,
      "loss": 0.5866,
      "step": 69
    },
    {
      "epoch": 0.26440037771482533,
      "grad_norm": 1.773505449295044,
      "learning_rate": 8.695044586103296e-05,
      "loss": 0.7318,
      "step": 70
    },
    {
      "epoch": 0.26817752596789424,
      "grad_norm": 1.6386646032333374,
      "learning_rate": 8.632167870081121e-05,
      "loss": 0.7247,
      "step": 71
    },
    {
      "epoch": 0.2719546742209632,
      "grad_norm": 1.5340347290039062,
      "learning_rate": 8.568050772058762e-05,
      "loss": 0.6559,
      "step": 72
    },
    {
      "epoch": 0.2757318224740321,
      "grad_norm": 1.6369562149047852,
      "learning_rate": 8.502715187966455e-05,
      "loss": 0.6877,
      "step": 73
    },
    {
      "epoch": 0.279508970727101,
      "grad_norm": 1.544015645980835,
      "learning_rate": 8.436183429846313e-05,
      "loss": 0.6584,
      "step": 74
    },
    {
      "epoch": 0.28328611898017,
      "grad_norm": 1.7706215381622314,
      "learning_rate": 8.368478218232787e-05,
      "loss": 0.7188,
      "step": 75
    },
    {
      "epoch": 0.2870632672332389,
      "grad_norm": 1.6866343021392822,
      "learning_rate": 8.299622674393614e-05,
      "loss": 0.6599,
      "step": 76
    },
    {
      "epoch": 0.29084041548630785,
      "grad_norm": 1.8714741468429565,
      "learning_rate": 8.229640312433937e-05,
      "loss": 0.6722,
      "step": 77
    },
    {
      "epoch": 0.29461756373937675,
      "grad_norm": 1.721524953842163,
      "learning_rate": 8.158555031266254e-05,
      "loss": 0.7256,
      "step": 78
    },
    {
      "epoch": 0.2983947119924457,
      "grad_norm": 1.5944390296936035,
      "learning_rate": 8.086391106448965e-05,
      "loss": 0.6347,
      "step": 79
    },
    {
      "epoch": 0.3021718602455146,
      "grad_norm": 1.5683523416519165,
      "learning_rate": 8.013173181896283e-05,
      "loss": 0.6439,
      "step": 80
    },
    {
      "epoch": 0.3059490084985836,
      "grad_norm": 1.6822518110275269,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.6437,
      "step": 81
    },
    {
      "epoch": 0.3097261567516525,
      "grad_norm": 1.9092936515808105,
      "learning_rate": 7.863675700402526e-05,
      "loss": 0.7138,
      "step": 82
    },
    {
      "epoch": 0.31350330500472146,
      "grad_norm": 1.9999340772628784,
      "learning_rate": 7.787447196714427e-05,
      "loss": 0.6435,
      "step": 83
    },
    {
      "epoch": 0.31728045325779036,
      "grad_norm": 1.9476672410964966,
      "learning_rate": 7.710266782362247e-05,
      "loss": 0.5772,
      "step": 84
    },
    {
      "epoch": 0.3210576015108593,
      "grad_norm": 1.825303316116333,
      "learning_rate": 7.63216081438678e-05,
      "loss": 0.649,
      "step": 85
    },
    {
      "epoch": 0.32483474976392823,
      "grad_norm": 1.7973113059997559,
      "learning_rate": 7.553155965904535e-05,
      "loss": 0.5625,
      "step": 86
    },
    {
      "epoch": 0.3286118980169972,
      "grad_norm": 1.678102731704712,
      "learning_rate": 7.473279216998895e-05,
      "loss": 0.5223,
      "step": 87
    },
    {
      "epoch": 0.3323890462700661,
      "grad_norm": 1.9203351736068726,
      "learning_rate": 7.392557845506432e-05,
      "loss": 0.6239,
      "step": 88
    },
    {
      "epoch": 0.336166194523135,
      "grad_norm": 1.9650973081588745,
      "learning_rate": 7.311019417701566e-05,
      "loss": 0.7045,
      "step": 89
    },
    {
      "epoch": 0.33994334277620397,
      "grad_norm": 2.0367164611816406,
      "learning_rate": 7.228691778882693e-05,
      "loss": 0.6919,
      "step": 90
    },
    {
      "epoch": 0.3437204910292729,
      "grad_norm": 1.9451732635498047,
      "learning_rate": 7.145603043863045e-05,
      "loss": 0.6438,
      "step": 91
    },
    {
      "epoch": 0.34749763928234184,
      "grad_norm": 2.4473257064819336,
      "learning_rate": 7.061781587369519e-05,
      "loss": 0.7729,
      "step": 92
    },
    {
      "epoch": 0.35127478753541075,
      "grad_norm": 2.3056561946868896,
      "learning_rate": 6.977256034352712e-05,
      "loss": 0.792,
      "step": 93
    },
    {
      "epoch": 0.3550519357884797,
      "grad_norm": 2.2371840476989746,
      "learning_rate": 6.892055250211552e-05,
      "loss": 0.7239,
      "step": 94
    },
    {
      "epoch": 0.3588290840415486,
      "grad_norm": 2.878354072570801,
      "learning_rate": 6.806208330935766e-05,
      "loss": 1.0358,
      "step": 95
    },
    {
      "epoch": 0.3626062322946176,
      "grad_norm": 3.273000717163086,
      "learning_rate": 6.719744593169641e-05,
      "loss": 0.9572,
      "step": 96
    },
    {
      "epoch": 0.3663833805476865,
      "grad_norm": 3.6380937099456787,
      "learning_rate": 6.632693564200416e-05,
      "loss": 0.9839,
      "step": 97
    },
    {
      "epoch": 0.37016052880075545,
      "grad_norm": 3.747563362121582,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.0543,
      "step": 98
    },
    {
      "epoch": 0.37393767705382436,
      "grad_norm": 3.951873302459717,
      "learning_rate": 6.456948734446624e-05,
      "loss": 1.0016,
      "step": 99
    },
    {
      "epoch": 0.3777148253068933,
      "grad_norm": 4.127007961273193,
      "learning_rate": 6.368314950360415e-05,
      "loss": 1.2259,
      "step": 100
    },
    {
      "epoch": 0.3777148253068933,
      "eval_loss": 0.7670125365257263,
      "eval_runtime": 11.0624,
      "eval_samples_per_second": 40.317,
      "eval_steps_per_second": 20.158,
      "step": 100
    },
    {
      "epoch": 0.3814919735599622,
      "grad_norm": 2.226933717727661,
      "learning_rate": 6.279213887972179e-05,
      "loss": 0.7909,
      "step": 101
    },
    {
      "epoch": 0.38526912181303113,
      "grad_norm": 2.067042350769043,
      "learning_rate": 6.189675975213094e-05,
      "loss": 0.768,
      "step": 102
    },
    {
      "epoch": 0.3890462700661001,
      "grad_norm": 2.1467435359954834,
      "learning_rate": 6.099731789198344e-05,
      "loss": 0.7788,
      "step": 103
    },
    {
      "epoch": 0.392823418319169,
      "grad_norm": 2.043851375579834,
      "learning_rate": 6.009412045785051e-05,
      "loss": 0.7792,
      "step": 104
    },
    {
      "epoch": 0.39660056657223797,
      "grad_norm": 1.6794301271438599,
      "learning_rate": 5.918747589082853e-05,
      "loss": 0.7373,
      "step": 105
    },
    {
      "epoch": 0.4003777148253069,
      "grad_norm": 1.5084600448608398,
      "learning_rate": 5.82776938092065e-05,
      "loss": 0.6173,
      "step": 106
    },
    {
      "epoch": 0.40415486307837584,
      "grad_norm": 1.2415573596954346,
      "learning_rate": 5.736508490273188e-05,
      "loss": 0.5869,
      "step": 107
    },
    {
      "epoch": 0.40793201133144474,
      "grad_norm": 1.3616746664047241,
      "learning_rate": 5.644996082651017e-05,
      "loss": 0.5805,
      "step": 108
    },
    {
      "epoch": 0.4117091595845137,
      "grad_norm": 1.5013140439987183,
      "learning_rate": 5.553263409457504e-05,
      "loss": 0.7008,
      "step": 109
    },
    {
      "epoch": 0.4154863078375826,
      "grad_norm": 1.5834474563598633,
      "learning_rate": 5.4613417973165106e-05,
      "loss": 0.6307,
      "step": 110
    },
    {
      "epoch": 0.4192634560906516,
      "grad_norm": 1.5312182903289795,
      "learning_rate": 5.3692626373743706e-05,
      "loss": 0.5383,
      "step": 111
    },
    {
      "epoch": 0.4230406043437205,
      "grad_norm": 1.578965663909912,
      "learning_rate": 5.27705737457985e-05,
      "loss": 0.5546,
      "step": 112
    },
    {
      "epoch": 0.42681775259678945,
      "grad_norm": 1.4108834266662598,
      "learning_rate": 5.184757496945726e-05,
      "loss": 0.5339,
      "step": 113
    },
    {
      "epoch": 0.43059490084985835,
      "grad_norm": 1.4769469499588013,
      "learning_rate": 5.092394524795649e-05,
      "loss": 0.5179,
      "step": 114
    },
    {
      "epoch": 0.4343720491029273,
      "grad_norm": 1.5313994884490967,
      "learning_rate": 5e-05,
      "loss": 0.6111,
      "step": 115
    },
    {
      "epoch": 0.4381491973559962,
      "grad_norm": 1.3974460363388062,
      "learning_rate": 4.907605475204352e-05,
      "loss": 0.5616,
      "step": 116
    },
    {
      "epoch": 0.44192634560906513,
      "grad_norm": 1.5315614938735962,
      "learning_rate": 4.8152425030542766e-05,
      "loss": 0.6007,
      "step": 117
    },
    {
      "epoch": 0.4457034938621341,
      "grad_norm": 1.595839500427246,
      "learning_rate": 4.72294262542015e-05,
      "loss": 0.6371,
      "step": 118
    },
    {
      "epoch": 0.449480642115203,
      "grad_norm": 1.528900384902954,
      "learning_rate": 4.6307373626256306e-05,
      "loss": 0.5995,
      "step": 119
    },
    {
      "epoch": 0.45325779036827196,
      "grad_norm": 1.5370620489120483,
      "learning_rate": 4.5386582026834906e-05,
      "loss": 0.5323,
      "step": 120
    },
    {
      "epoch": 0.45703493862134087,
      "grad_norm": 1.574083685874939,
      "learning_rate": 4.446736590542497e-05,
      "loss": 0.5331,
      "step": 121
    },
    {
      "epoch": 0.46081208687440983,
      "grad_norm": 1.5231963396072388,
      "learning_rate": 4.3550039173489845e-05,
      "loss": 0.5113,
      "step": 122
    },
    {
      "epoch": 0.46458923512747874,
      "grad_norm": 1.4855923652648926,
      "learning_rate": 4.2634915097268115e-05,
      "loss": 0.4917,
      "step": 123
    },
    {
      "epoch": 0.4683663833805477,
      "grad_norm": 1.496247410774231,
      "learning_rate": 4.1722306190793495e-05,
      "loss": 0.4791,
      "step": 124
    },
    {
      "epoch": 0.4721435316336166,
      "grad_norm": 1.812813401222229,
      "learning_rate": 4.0812524109171476e-05,
      "loss": 0.67,
      "step": 125
    },
    {
      "epoch": 0.47592067988668557,
      "grad_norm": 1.5825666189193726,
      "learning_rate": 3.99058795421495e-05,
      "loss": 0.5366,
      "step": 126
    },
    {
      "epoch": 0.4796978281397545,
      "grad_norm": 1.5532022714614868,
      "learning_rate": 3.9002682108016585e-05,
      "loss": 0.5429,
      "step": 127
    },
    {
      "epoch": 0.48347497639282344,
      "grad_norm": 1.653063416481018,
      "learning_rate": 3.8103240247869075e-05,
      "loss": 0.5422,
      "step": 128
    },
    {
      "epoch": 0.48725212464589235,
      "grad_norm": 1.6550402641296387,
      "learning_rate": 3.720786112027822e-05,
      "loss": 0.4998,
      "step": 129
    },
    {
      "epoch": 0.4910292728989613,
      "grad_norm": 1.5448733568191528,
      "learning_rate": 3.631685049639586e-05,
      "loss": 0.5118,
      "step": 130
    },
    {
      "epoch": 0.4948064211520302,
      "grad_norm": 1.6610890626907349,
      "learning_rate": 3.543051265553377e-05,
      "loss": 0.5189,
      "step": 131
    },
    {
      "epoch": 0.4985835694050991,
      "grad_norm": 1.691016674041748,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.5197,
      "step": 132
    },
    {
      "epoch": 0.502360717658168,
      "grad_norm": 1.66881263256073,
      "learning_rate": 3.367306435799584e-05,
      "loss": 0.5432,
      "step": 133
    },
    {
      "epoch": 0.506137865911237,
      "grad_norm": 2.1006433963775635,
      "learning_rate": 3.2802554068303596e-05,
      "loss": 0.6788,
      "step": 134
    },
    {
      "epoch": 0.509915014164306,
      "grad_norm": 1.8879708051681519,
      "learning_rate": 3.1937916690642356e-05,
      "loss": 0.5475,
      "step": 135
    },
    {
      "epoch": 0.5136921624173749,
      "grad_norm": 1.9183238744735718,
      "learning_rate": 3.107944749788449e-05,
      "loss": 0.594,
      "step": 136
    },
    {
      "epoch": 0.5174693106704438,
      "grad_norm": 1.836464524269104,
      "learning_rate": 3.0227439656472877e-05,
      "loss": 0.6261,
      "step": 137
    },
    {
      "epoch": 0.5212464589235127,
      "grad_norm": 2.068596124649048,
      "learning_rate": 2.9382184126304834e-05,
      "loss": 0.6147,
      "step": 138
    },
    {
      "epoch": 0.5250236071765817,
      "grad_norm": 1.8795504570007324,
      "learning_rate": 2.8543969561369556e-05,
      "loss": 0.625,
      "step": 139
    },
    {
      "epoch": 0.5288007554296507,
      "grad_norm": 2.140127420425415,
      "learning_rate": 2.771308221117309e-05,
      "loss": 0.7256,
      "step": 140
    },
    {
      "epoch": 0.5325779036827195,
      "grad_norm": 2.207440137863159,
      "learning_rate": 2.688980582298435e-05,
      "loss": 0.5535,
      "step": 141
    },
    {
      "epoch": 0.5363550519357885,
      "grad_norm": 2.187288761138916,
      "learning_rate": 2.607442154493568e-05,
      "loss": 0.664,
      "step": 142
    },
    {
      "epoch": 0.5401322001888574,
      "grad_norm": 2.49151873588562,
      "learning_rate": 2.5267207830011068e-05,
      "loss": 0.7329,
      "step": 143
    },
    {
      "epoch": 0.5439093484419264,
      "grad_norm": 3.0910847187042236,
      "learning_rate": 2.446844034095466e-05,
      "loss": 1.1256,
      "step": 144
    },
    {
      "epoch": 0.5476864966949953,
      "grad_norm": 3.17502760887146,
      "learning_rate": 2.3678391856132204e-05,
      "loss": 0.8965,
      "step": 145
    },
    {
      "epoch": 0.5514636449480642,
      "grad_norm": 3.321992874145508,
      "learning_rate": 2.2897332176377528e-05,
      "loss": 0.8985,
      "step": 146
    },
    {
      "epoch": 0.5552407932011332,
      "grad_norm": 3.996783971786499,
      "learning_rate": 2.2125528032855724e-05,
      "loss": 1.0264,
      "step": 147
    },
    {
      "epoch": 0.559017941454202,
      "grad_norm": 3.5252180099487305,
      "learning_rate": 2.136324299597474e-05,
      "loss": 0.9227,
      "step": 148
    },
    {
      "epoch": 0.562795089707271,
      "grad_norm": 3.9999685287475586,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.9694,
      "step": 149
    },
    {
      "epoch": 0.56657223796034,
      "grad_norm": 3.909895658493042,
      "learning_rate": 1.9868268181037185e-05,
      "loss": 0.9027,
      "step": 150
    },
    {
      "epoch": 0.56657223796034,
      "eval_loss": 0.6212329268455505,
      "eval_runtime": 10.931,
      "eval_samples_per_second": 40.801,
      "eval_steps_per_second": 20.401,
      "step": 150
    },
    {
      "epoch": 0.5703493862134089,
      "grad_norm": 1.6851472854614258,
      "learning_rate": 1.9136088935510362e-05,
      "loss": 0.6745,
      "step": 151
    },
    {
      "epoch": 0.5741265344664778,
      "grad_norm": 1.652190923690796,
      "learning_rate": 1.8414449687337464e-05,
      "loss": 0.6327,
      "step": 152
    },
    {
      "epoch": 0.5779036827195467,
      "grad_norm": 1.691318154335022,
      "learning_rate": 1.7703596875660645e-05,
      "loss": 0.7017,
      "step": 153
    },
    {
      "epoch": 0.5816808309726157,
      "grad_norm": 1.4943283796310425,
      "learning_rate": 1.700377325606388e-05,
      "loss": 0.562,
      "step": 154
    },
    {
      "epoch": 0.5854579792256847,
      "grad_norm": 1.4418638944625854,
      "learning_rate": 1.631521781767214e-05,
      "loss": 0.5831,
      "step": 155
    },
    {
      "epoch": 0.5892351274787535,
      "grad_norm": 1.6237192153930664,
      "learning_rate": 1.5638165701536868e-05,
      "loss": 0.6027,
      "step": 156
    },
    {
      "epoch": 0.5930122757318225,
      "grad_norm": 1.5733246803283691,
      "learning_rate": 1.4972848120335453e-05,
      "loss": 0.6051,
      "step": 157
    },
    {
      "epoch": 0.5967894239848914,
      "grad_norm": 1.422416090965271,
      "learning_rate": 1.4319492279412388e-05,
      "loss": 0.5212,
      "step": 158
    },
    {
      "epoch": 0.6005665722379604,
      "grad_norm": 1.4017753601074219,
      "learning_rate": 1.3678321299188801e-05,
      "loss": 0.5004,
      "step": 159
    },
    {
      "epoch": 0.6043437204910292,
      "grad_norm": 1.4192122220993042,
      "learning_rate": 1.3049554138967051e-05,
      "loss": 0.5081,
      "step": 160
    },
    {
      "epoch": 0.6081208687440982,
      "grad_norm": 1.505066156387329,
      "learning_rate": 1.2433405522156332e-05,
      "loss": 0.5414,
      "step": 161
    },
    {
      "epoch": 0.6118980169971672,
      "grad_norm": 1.3610177040100098,
      "learning_rate": 1.183008586294485e-05,
      "loss": 0.5256,
      "step": 162
    },
    {
      "epoch": 0.615675165250236,
      "grad_norm": 1.4742549657821655,
      "learning_rate": 1.1239801194443506e-05,
      "loss": 0.5877,
      "step": 163
    },
    {
      "epoch": 0.619452313503305,
      "grad_norm": 1.4625017642974854,
      "learning_rate": 1.066275309832584e-05,
      "loss": 0.4908,
      "step": 164
    },
    {
      "epoch": 0.623229461756374,
      "grad_norm": 1.3624998331069946,
      "learning_rate": 1.0099138635988026e-05,
      "loss": 0.4458,
      "step": 165
    },
    {
      "epoch": 0.6270066100094429,
      "grad_norm": 1.3607821464538574,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.4799,
      "step": 166
    },
    {
      "epoch": 0.6307837582625118,
      "grad_norm": 1.5529695749282837,
      "learning_rate": 9.012975854638949e-06,
      "loss": 0.5161,
      "step": 167
    },
    {
      "epoch": 0.6345609065155807,
      "grad_norm": 1.3452229499816895,
      "learning_rate": 8.490798459222476e-06,
      "loss": 0.4864,
      "step": 168
    },
    {
      "epoch": 0.6383380547686497,
      "grad_norm": 1.5585039854049683,
      "learning_rate": 7.982796418105371e-06,
      "loss": 0.5284,
      "step": 169
    },
    {
      "epoch": 0.6421152030217187,
      "grad_norm": 1.5944770574569702,
      "learning_rate": 7.489143213519301e-06,
      "loss": 0.5722,
      "step": 170
    },
    {
      "epoch": 0.6458923512747875,
      "grad_norm": 1.5475810766220093,
      "learning_rate": 7.010007427581378e-06,
      "loss": 0.5712,
      "step": 171
    },
    {
      "epoch": 0.6496694995278565,
      "grad_norm": 1.6774266958236694,
      "learning_rate": 6.5455526847235825e-06,
      "loss": 0.5642,
      "step": 172
    },
    {
      "epoch": 0.6534466477809254,
      "grad_norm": 1.5541495084762573,
      "learning_rate": 6.0959375958151045e-06,
      "loss": 0.4777,
      "step": 173
    },
    {
      "epoch": 0.6572237960339944,
      "grad_norm": 1.5276578664779663,
      "learning_rate": 5.6613157039969055e-06,
      "loss": 0.4975,
      "step": 174
    },
    {
      "epoch": 0.6610009442870632,
      "grad_norm": 1.5561702251434326,
      "learning_rate": 5.241835432246889e-06,
      "loss": 0.5383,
      "step": 175
    },
    {
      "epoch": 0.6647780925401322,
      "grad_norm": 1.5004448890686035,
      "learning_rate": 4.837640032693558e-06,
      "loss": 0.4556,
      "step": 176
    },
    {
      "epoch": 0.6685552407932012,
      "grad_norm": 1.4507856369018555,
      "learning_rate": 4.448867537695578e-06,
      "loss": 0.4388,
      "step": 177
    },
    {
      "epoch": 0.67233238904627,
      "grad_norm": 1.544404149055481,
      "learning_rate": 4.075650712703849e-06,
      "loss": 0.4649,
      "step": 178
    },
    {
      "epoch": 0.676109537299339,
      "grad_norm": 1.7769454717636108,
      "learning_rate": 3.71811701092219e-06,
      "loss": 0.6198,
      "step": 179
    },
    {
      "epoch": 0.6798866855524079,
      "grad_norm": 1.504475474357605,
      "learning_rate": 3.376388529782215e-06,
      "loss": 0.4049,
      "step": 180
    },
    {
      "epoch": 0.6836638338054769,
      "grad_norm": 1.4703073501586914,
      "learning_rate": 3.0505819692471792e-06,
      "loss": 0.4139,
      "step": 181
    },
    {
      "epoch": 0.6874409820585458,
      "grad_norm": 1.7930094003677368,
      "learning_rate": 2.7408085919590264e-06,
      "loss": 0.5649,
      "step": 182
    },
    {
      "epoch": 0.6912181303116147,
      "grad_norm": 1.8142603635787964,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.5327,
      "step": 183
    },
    {
      "epoch": 0.6949952785646837,
      "grad_norm": 1.7021881341934204,
      "learning_rate": 2.1697790249779636e-06,
      "loss": 0.5278,
      "step": 184
    },
    {
      "epoch": 0.6987724268177526,
      "grad_norm": 1.9536923170089722,
      "learning_rate": 1.908717841359048e-06,
      "loss": 0.519,
      "step": 185
    },
    {
      "epoch": 0.7025495750708215,
      "grad_norm": 1.837648630142212,
      "learning_rate": 1.6640797865406288e-06,
      "loss": 0.5552,
      "step": 186
    },
    {
      "epoch": 0.7063267233238905,
      "grad_norm": 1.8521840572357178,
      "learning_rate": 1.4359484041943038e-06,
      "loss": 0.5943,
      "step": 187
    },
    {
      "epoch": 0.7101038715769594,
      "grad_norm": 1.8110532760620117,
      "learning_rate": 1.2244016009781701e-06,
      "loss": 0.5854,
      "step": 188
    },
    {
      "epoch": 0.7138810198300283,
      "grad_norm": 1.8831785917282104,
      "learning_rate": 1.0295116199317057e-06,
      "loss": 0.5665,
      "step": 189
    },
    {
      "epoch": 0.7176581680830972,
      "grad_norm": 1.9565130472183228,
      "learning_rate": 8.513450158049108e-07,
      "loss": 0.538,
      "step": 190
    },
    {
      "epoch": 0.7214353163361662,
      "grad_norm": 1.9892494678497314,
      "learning_rate": 6.899626323298713e-07,
      "loss": 0.5818,
      "step": 191
    },
    {
      "epoch": 0.7252124645892352,
      "grad_norm": 2.109586238861084,
      "learning_rate": 5.454195814427021e-07,
      "loss": 0.6005,
      "step": 192
    },
    {
      "epoch": 0.728989612842304,
      "grad_norm": 2.2817015647888184,
      "learning_rate": 4.177652244628627e-07,
      "loss": 0.7264,
      "step": 193
    },
    {
      "epoch": 0.732766761095373,
      "grad_norm": 3.0464565753936768,
      "learning_rate": 3.0704315523631953e-07,
      "loss": 0.8336,
      "step": 194
    },
    {
      "epoch": 0.7365439093484419,
      "grad_norm": 3.6609790325164795,
      "learning_rate": 2.1329118524827662e-07,
      "loss": 0.9042,
      "step": 195
    },
    {
      "epoch": 0.7403210576015109,
      "grad_norm": 3.2964279651641846,
      "learning_rate": 1.3654133071059893e-07,
      "loss": 0.873,
      "step": 196
    },
    {
      "epoch": 0.7440982058545798,
      "grad_norm": 3.6157829761505127,
      "learning_rate": 7.681980162830282e-08,
      "loss": 0.9373,
      "step": 197
    },
    {
      "epoch": 0.7478753541076487,
      "grad_norm": 4.483926296234131,
      "learning_rate": 3.4146992848854695e-08,
      "loss": 1.0835,
      "step": 198
    },
    {
      "epoch": 0.7516525023607177,
      "grad_norm": 4.7277655601501465,
      "learning_rate": 8.537477097364522e-09,
      "loss": 1.15,
      "step": 199
    },
    {
      "epoch": 0.7554296506137866,
      "grad_norm": 5.080047130584717,
      "learning_rate": 0.0,
      "loss": 0.9157,
      "step": 200
    },
    {
      "epoch": 0.7554296506137866,
      "eval_loss": 0.5849894285202026,
      "eval_runtime": 10.9099,
      "eval_samples_per_second": 40.88,
      "eval_steps_per_second": 20.44,
      "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": 1.5476286069669888e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}