{
  "best_metric": 1.9855155944824219,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.011819981679028398,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.909990839514199e-05,
      "grad_norm": 14.701292037963867,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 4.2367,
      "step": 1
    },
    {
      "epoch": 5.909990839514199e-05,
      "eval_loss": 5.2180633544921875,
      "eval_runtime": 685.1037,
      "eval_samples_per_second": 41.597,
      "eval_steps_per_second": 20.798,
      "step": 1
    },
    {
      "epoch": 0.00011819981679028398,
      "grad_norm": 17.779050827026367,
      "learning_rate": 6.666666666666667e-06,
      "loss": 4.8007,
      "step": 2
    },
    {
      "epoch": 0.00017729972518542596,
      "grad_norm": 15.671966552734375,
      "learning_rate": 1e-05,
      "loss": 4.2807,
      "step": 3
    },
    {
      "epoch": 0.00023639963358056796,
      "grad_norm": 15.129173278808594,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 4.8199,
      "step": 4
    },
    {
      "epoch": 0.00029549954197570993,
      "grad_norm": 13.176082611083984,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 4.1705,
      "step": 5
    },
    {
      "epoch": 0.0003545994503708519,
      "grad_norm": 15.571830749511719,
      "learning_rate": 2e-05,
      "loss": 4.6829,
      "step": 6
    },
    {
      "epoch": 0.0004136993587659939,
      "grad_norm": 14.50023078918457,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 4.5119,
      "step": 7
    },
    {
      "epoch": 0.0004727992671611359,
      "grad_norm": 14.558034896850586,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 4.1539,
      "step": 8
    },
    {
      "epoch": 0.0005318991755562779,
      "grad_norm": 10.521626472473145,
      "learning_rate": 3e-05,
      "loss": 3.6708,
      "step": 9
    },
    {
      "epoch": 0.0005909990839514199,
      "grad_norm": 10.474567413330078,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 3.6137,
      "step": 10
    },
    {
      "epoch": 0.0006500989923465619,
      "grad_norm": 8.635924339294434,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 3.2574,
      "step": 11
    },
    {
      "epoch": 0.0007091989007417038,
      "grad_norm": 13.226350784301758,
      "learning_rate": 4e-05,
      "loss": 3.3157,
      "step": 12
    },
    {
      "epoch": 0.0007682988091368458,
      "grad_norm": 8.407894134521484,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 3.3735,
      "step": 13
    },
    {
      "epoch": 0.0008273987175319878,
      "grad_norm": 7.1230974197387695,
      "learning_rate": 4.666666666666667e-05,
      "loss": 3.0306,
      "step": 14
    },
    {
      "epoch": 0.0008864986259271298,
      "grad_norm": 7.134531021118164,
      "learning_rate": 5e-05,
      "loss": 2.9162,
      "step": 15
    },
    {
      "epoch": 0.0009455985343222718,
      "grad_norm": 6.953376770019531,
      "learning_rate": 5.333333333333333e-05,
      "loss": 2.6598,
      "step": 16
    },
    {
      "epoch": 0.0010046984427174138,
      "grad_norm": 6.143302917480469,
      "learning_rate": 5.666666666666667e-05,
      "loss": 2.7871,
      "step": 17
    },
    {
      "epoch": 0.0010637983511125558,
      "grad_norm": 5.899278163909912,
      "learning_rate": 6e-05,
      "loss": 2.8841,
      "step": 18
    },
    {
      "epoch": 0.0011228982595076977,
      "grad_norm": 5.721312999725342,
      "learning_rate": 6.333333333333333e-05,
      "loss": 2.5476,
      "step": 19
    },
    {
      "epoch": 0.0011819981679028397,
      "grad_norm": 5.896458625793457,
      "learning_rate": 6.666666666666667e-05,
      "loss": 2.5949,
      "step": 20
    },
    {
      "epoch": 0.0012410980762979818,
      "grad_norm": 5.44040584564209,
      "learning_rate": 7e-05,
      "loss": 2.3684,
      "step": 21
    },
    {
      "epoch": 0.0013001979846931238,
      "grad_norm": 5.645882606506348,
      "learning_rate": 7.333333333333333e-05,
      "loss": 2.5693,
      "step": 22
    },
    {
      "epoch": 0.0013592978930882656,
      "grad_norm": 5.0001606941223145,
      "learning_rate": 7.666666666666667e-05,
      "loss": 2.1486,
      "step": 23
    },
    {
      "epoch": 0.0014183978014834077,
      "grad_norm": 5.0876240730285645,
      "learning_rate": 8e-05,
      "loss": 2.29,
      "step": 24
    },
    {
      "epoch": 0.0014774977098785497,
      "grad_norm": 5.2396368980407715,
      "learning_rate": 8.333333333333334e-05,
      "loss": 2.1141,
      "step": 25
    },
    {
      "epoch": 0.0015365976182736916,
      "grad_norm": 5.895060062408447,
      "learning_rate": 8.666666666666667e-05,
      "loss": 2.3359,
      "step": 26
    },
    {
      "epoch": 0.0015956975266688336,
      "grad_norm": 5.065176486968994,
      "learning_rate": 9e-05,
      "loss": 2.0715,
      "step": 27
    },
    {
      "epoch": 0.0016547974350639757,
      "grad_norm": 4.62436056137085,
      "learning_rate": 9.333333333333334e-05,
      "loss": 2.1005,
      "step": 28
    },
    {
      "epoch": 0.0017138973434591177,
      "grad_norm": 5.012561798095703,
      "learning_rate": 9.666666666666667e-05,
      "loss": 2.2374,
      "step": 29
    },
    {
      "epoch": 0.0017729972518542596,
      "grad_norm": 3.960059642791748,
      "learning_rate": 0.0001,
      "loss": 1.9641,
      "step": 30
    },
    {
      "epoch": 0.0018320971602494016,
      "grad_norm": 4.35403299331665,
      "learning_rate": 9.999146252290264e-05,
      "loss": 1.9797,
      "step": 31
    },
    {
      "epoch": 0.0018911970686445437,
      "grad_norm": 4.6017985343933105,
      "learning_rate": 9.996585300715116e-05,
      "loss": 2.3441,
      "step": 32
    },
    {
      "epoch": 0.0019502969770396855,
      "grad_norm": 4.478118419647217,
      "learning_rate": 9.99231801983717e-05,
      "loss": 1.8137,
      "step": 33
    },
    {
      "epoch": 0.0020093968854348276,
      "grad_norm": 4.417632102966309,
      "learning_rate": 9.986345866928941e-05,
      "loss": 2.0547,
      "step": 34
    },
    {
      "epoch": 0.0020684967938299696,
      "grad_norm": 5.099286079406738,
      "learning_rate": 9.978670881475172e-05,
      "loss": 2.1801,
      "step": 35
    },
    {
      "epoch": 0.0021275967022251117,
      "grad_norm": 5.606115341186523,
      "learning_rate": 9.96929568447637e-05,
      "loss": 2.2527,
      "step": 36
    },
    {
      "epoch": 0.0021866966106202537,
      "grad_norm": 4.479214668273926,
      "learning_rate": 9.958223477553714e-05,
      "loss": 2.1026,
      "step": 37
    },
    {
      "epoch": 0.0022457965190153953,
      "grad_norm": 4.82413911819458,
      "learning_rate": 9.94545804185573e-05,
      "loss": 1.9819,
      "step": 38
    },
    {
      "epoch": 0.0023048964274105374,
      "grad_norm": 4.534383773803711,
      "learning_rate": 9.931003736767013e-05,
      "loss": 1.8833,
      "step": 39
    },
    {
      "epoch": 0.0023639963358056794,
      "grad_norm": 4.473276615142822,
      "learning_rate": 9.91486549841951e-05,
      "loss": 2.3477,
      "step": 40
    },
    {
      "epoch": 0.0024230962442008215,
      "grad_norm": 4.184545993804932,
      "learning_rate": 9.89704883800683e-05,
      "loss": 2.4562,
      "step": 41
    },
    {
      "epoch": 0.0024821961525959635,
      "grad_norm": 4.254863739013672,
      "learning_rate": 9.877559839902184e-05,
      "loss": 2.0597,
      "step": 42
    },
    {
      "epoch": 0.0025412960609911056,
      "grad_norm": 3.5156869888305664,
      "learning_rate": 9.85640515958057e-05,
      "loss": 2.0459,
      "step": 43
    },
    {
      "epoch": 0.0026003959693862476,
      "grad_norm": 3.2430365085601807,
      "learning_rate": 9.833592021345937e-05,
      "loss": 1.9904,
      "step": 44
    },
    {
      "epoch": 0.0026594958777813892,
      "grad_norm": 4.73977518081665,
      "learning_rate": 9.809128215864097e-05,
      "loss": 2.048,
      "step": 45
    },
    {
      "epoch": 0.0027185957861765313,
      "grad_norm": 3.6739819049835205,
      "learning_rate": 9.783022097502204e-05,
      "loss": 2.0881,
      "step": 46
    },
    {
      "epoch": 0.0027776956945716733,
      "grad_norm": 3.2717761993408203,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.0992,
      "step": 47
    },
    {
      "epoch": 0.0028367956029668154,
      "grad_norm": 3.5809576511383057,
      "learning_rate": 9.725919140804099e-05,
      "loss": 2.0799,
      "step": 48
    },
    {
      "epoch": 0.0028958955113619574,
      "grad_norm": 4.07869815826416,
      "learning_rate": 9.694941803075283e-05,
      "loss": 1.9132,
      "step": 49
    },
    {
      "epoch": 0.0029549954197570995,
      "grad_norm": 3.9852514266967773,
      "learning_rate": 9.662361147021779e-05,
      "loss": 1.6873,
      "step": 50
    },
    {
      "epoch": 0.0029549954197570995,
      "eval_loss": 2.1213467121124268,
      "eval_runtime": 683.4663,
      "eval_samples_per_second": 41.696,
      "eval_steps_per_second": 20.848,
      "step": 50
    },
    {
      "epoch": 0.0030140953281522415,
      "grad_norm": 5.36259126663208,
      "learning_rate": 9.628188298907782e-05,
      "loss": 2.25,
      "step": 51
    },
    {
      "epoch": 0.003073195236547383,
      "grad_norm": 4.400323867797852,
      "learning_rate": 9.592434928729616e-05,
      "loss": 2.3144,
      "step": 52
    },
    {
      "epoch": 0.003132295144942525,
      "grad_norm": 4.434739112854004,
      "learning_rate": 9.555113246230442e-05,
      "loss": 2.2359,
      "step": 53
    },
    {
      "epoch": 0.0031913950533376673,
      "grad_norm": 3.8840668201446533,
      "learning_rate": 9.516235996730645e-05,
      "loss": 2.1431,
      "step": 54
    },
    {
      "epoch": 0.0032504949617328093,
      "grad_norm": 3.4554171562194824,
      "learning_rate": 9.475816456775313e-05,
      "loss": 2.3377,
      "step": 55
    },
    {
      "epoch": 0.0033095948701279514,
      "grad_norm": 3.1983530521392822,
      "learning_rate": 9.43386842960031e-05,
      "loss": 2.0718,
      "step": 56
    },
    {
      "epoch": 0.0033686947785230934,
      "grad_norm": 3.158194065093994,
      "learning_rate": 9.39040624041849e-05,
      "loss": 2.1517,
      "step": 57
    },
    {
      "epoch": 0.0034277946869182355,
      "grad_norm": 3.890373468399048,
      "learning_rate": 9.345444731527642e-05,
      "loss": 2.441,
      "step": 58
    },
    {
      "epoch": 0.003486894595313377,
      "grad_norm": 3.167813777923584,
      "learning_rate": 9.298999257241863e-05,
      "loss": 2.2091,
      "step": 59
    },
    {
      "epoch": 0.003545994503708519,
      "grad_norm": 3.1658525466918945,
      "learning_rate": 9.251085678648072e-05,
      "loss": 2.0947,
      "step": 60
    },
    {
      "epoch": 0.003605094412103661,
      "grad_norm": 3.408245325088501,
      "learning_rate": 9.201720358189464e-05,
      "loss": 2.2448,
      "step": 61
    },
    {
      "epoch": 0.0036641943204988032,
      "grad_norm": 3.817218542098999,
      "learning_rate": 9.150920154077754e-05,
      "loss": 2.2507,
      "step": 62
    },
    {
      "epoch": 0.0037232942288939453,
      "grad_norm": 3.1942789554595947,
      "learning_rate": 9.098702414536107e-05,
      "loss": 1.9487,
      "step": 63
    },
    {
      "epoch": 0.0037823941372890873,
      "grad_norm": 4.15613317489624,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.2011,
      "step": 64
    },
    {
      "epoch": 0.0038414940456842294,
      "grad_norm": 3.4071335792541504,
      "learning_rate": 8.9900861364012e-05,
      "loss": 1.9582,
      "step": 65
    },
    {
      "epoch": 0.003900593954079371,
      "grad_norm": 3.021066665649414,
      "learning_rate": 8.933724690167417e-05,
      "loss": 2.3522,
      "step": 66
    },
    {
      "epoch": 0.003959693862474513,
      "grad_norm": 3.616442918777466,
      "learning_rate": 8.876019880555649e-05,
      "loss": 2.098,
      "step": 67
    },
    {
      "epoch": 0.004018793770869655,
      "grad_norm": 2.8962440490722656,
      "learning_rate": 8.816991413705516e-05,
      "loss": 2.0984,
      "step": 68
    },
    {
      "epoch": 0.004077893679264797,
      "grad_norm": 3.579745292663574,
      "learning_rate": 8.756659447784368e-05,
      "loss": 2.0251,
      "step": 69
    },
    {
      "epoch": 0.004136993587659939,
      "grad_norm": 3.775783061981201,
      "learning_rate": 8.695044586103296e-05,
      "loss": 2.119,
      "step": 70
    },
    {
      "epoch": 0.004196093496055081,
      "grad_norm": 3.080296754837036,
      "learning_rate": 8.632167870081121e-05,
      "loss": 2.2573,
      "step": 71
    },
    {
      "epoch": 0.004255193404450223,
      "grad_norm": 2.8051774501800537,
      "learning_rate": 8.568050772058762e-05,
      "loss": 1.9605,
      "step": 72
    },
    {
      "epoch": 0.004314293312845365,
      "grad_norm": 2.691725730895996,
      "learning_rate": 8.502715187966455e-05,
      "loss": 1.9226,
      "step": 73
    },
    {
      "epoch": 0.004373393221240507,
      "grad_norm": 2.9309446811676025,
      "learning_rate": 8.436183429846313e-05,
      "loss": 1.8635,
      "step": 74
    },
    {
      "epoch": 0.0044324931296356495,
      "grad_norm": 3.0920376777648926,
      "learning_rate": 8.368478218232787e-05,
      "loss": 2.3189,
      "step": 75
    },
    {
      "epoch": 0.004491593038030791,
      "grad_norm": 3.608562707901001,
      "learning_rate": 8.299622674393614e-05,
      "loss": 2.1573,
      "step": 76
    },
    {
      "epoch": 0.004550692946425933,
      "grad_norm": 3.4357681274414062,
      "learning_rate": 8.229640312433937e-05,
      "loss": 2.3833,
      "step": 77
    },
    {
      "epoch": 0.004609792854821075,
      "grad_norm": 3.2600772380828857,
      "learning_rate": 8.158555031266254e-05,
      "loss": 1.8943,
      "step": 78
    },
    {
      "epoch": 0.004668892763216217,
      "grad_norm": 3.176687002182007,
      "learning_rate": 8.086391106448965e-05,
      "loss": 1.8515,
      "step": 79
    },
    {
      "epoch": 0.004727992671611359,
      "grad_norm": 3.3577382564544678,
      "learning_rate": 8.013173181896283e-05,
      "loss": 1.8858,
      "step": 80
    },
    {
      "epoch": 0.004787092580006501,
      "grad_norm": 3.1467320919036865,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.0871,
      "step": 81
    },
    {
      "epoch": 0.004846192488401643,
      "grad_norm": 2.7516732215881348,
      "learning_rate": 7.863675700402526e-05,
      "loss": 1.8316,
      "step": 82
    },
    {
      "epoch": 0.004905292396796785,
      "grad_norm": 3.2467401027679443,
      "learning_rate": 7.787447196714427e-05,
      "loss": 2.0702,
      "step": 83
    },
    {
      "epoch": 0.004964392305191927,
      "grad_norm": 3.381272315979004,
      "learning_rate": 7.710266782362247e-05,
      "loss": 2.2569,
      "step": 84
    },
    {
      "epoch": 0.005023492213587069,
      "grad_norm": 2.826927661895752,
      "learning_rate": 7.63216081438678e-05,
      "loss": 2.1583,
      "step": 85
    },
    {
      "epoch": 0.005082592121982211,
      "grad_norm": 3.259561061859131,
      "learning_rate": 7.553155965904535e-05,
      "loss": 1.9269,
      "step": 86
    },
    {
      "epoch": 0.005141692030377353,
      "grad_norm": 3.7768051624298096,
      "learning_rate": 7.473279216998895e-05,
      "loss": 2.362,
      "step": 87
    },
    {
      "epoch": 0.005200791938772495,
      "grad_norm": 2.951033592224121,
      "learning_rate": 7.392557845506432e-05,
      "loss": 2.0554,
      "step": 88
    },
    {
      "epoch": 0.005259891847167637,
      "grad_norm": 2.699878454208374,
      "learning_rate": 7.311019417701566e-05,
      "loss": 1.6497,
      "step": 89
    },
    {
      "epoch": 0.0053189917555627785,
      "grad_norm": 2.780122756958008,
      "learning_rate": 7.228691778882693e-05,
      "loss": 1.904,
      "step": 90
    },
    {
      "epoch": 0.0053780916639579205,
      "grad_norm": 2.7070605754852295,
      "learning_rate": 7.145603043863045e-05,
      "loss": 1.8025,
      "step": 91
    },
    {
      "epoch": 0.005437191572353063,
      "grad_norm": 2.740710973739624,
      "learning_rate": 7.061781587369519e-05,
      "loss": 1.9662,
      "step": 92
    },
    {
      "epoch": 0.005496291480748205,
      "grad_norm": 3.0559332370758057,
      "learning_rate": 6.977256034352712e-05,
      "loss": 2.0185,
      "step": 93
    },
    {
      "epoch": 0.005555391389143347,
      "grad_norm": 2.9242119789123535,
      "learning_rate": 6.892055250211552e-05,
      "loss": 2.0248,
      "step": 94
    },
    {
      "epoch": 0.005614491297538489,
      "grad_norm": 2.762571334838867,
      "learning_rate": 6.806208330935766e-05,
      "loss": 1.7682,
      "step": 95
    },
    {
      "epoch": 0.005673591205933631,
      "grad_norm": 2.9518630504608154,
      "learning_rate": 6.719744593169641e-05,
      "loss": 1.88,
      "step": 96
    },
    {
      "epoch": 0.005732691114328773,
      "grad_norm": 2.976564645767212,
      "learning_rate": 6.632693564200416e-05,
      "loss": 1.8969,
      "step": 97
    },
    {
      "epoch": 0.005791791022723915,
      "grad_norm": 2.815509796142578,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.972,
      "step": 98
    },
    {
      "epoch": 0.005850890931119057,
      "grad_norm": 3.5287201404571533,
      "learning_rate": 6.456948734446624e-05,
      "loss": 2.125,
      "step": 99
    },
    {
      "epoch": 0.005909990839514199,
      "grad_norm": 3.9216277599334717,
      "learning_rate": 6.368314950360415e-05,
      "loss": 1.9601,
      "step": 100
    },
    {
      "epoch": 0.005909990839514199,
      "eval_loss": 2.066779136657715,
      "eval_runtime": 683.3126,
      "eval_samples_per_second": 41.706,
      "eval_steps_per_second": 20.853,
      "step": 100
    },
    {
      "epoch": 0.005969090747909341,
      "grad_norm": 4.6523542404174805,
      "learning_rate": 6.279213887972179e-05,
      "loss": 2.4316,
      "step": 101
    },
    {
      "epoch": 0.006028190656304483,
      "grad_norm": 3.620638847351074,
      "learning_rate": 6.189675975213094e-05,
      "loss": 2.1098,
      "step": 102
    },
    {
      "epoch": 0.006087290564699625,
      "grad_norm": 3.4210116863250732,
      "learning_rate": 6.099731789198344e-05,
      "loss": 2.4399,
      "step": 103
    },
    {
      "epoch": 0.006146390473094766,
      "grad_norm": 3.1519083976745605,
      "learning_rate": 6.009412045785051e-05,
      "loss": 2.0912,
      "step": 104
    },
    {
      "epoch": 0.006205490381489908,
      "grad_norm": 2.7077889442443848,
      "learning_rate": 5.918747589082853e-05,
      "loss": 2.0573,
      "step": 105
    },
    {
      "epoch": 0.00626459028988505,
      "grad_norm": 3.0147182941436768,
      "learning_rate": 5.82776938092065e-05,
      "loss": 2.0726,
      "step": 106
    },
    {
      "epoch": 0.0063236901982801925,
      "grad_norm": 3.317253351211548,
      "learning_rate": 5.736508490273188e-05,
      "loss": 1.97,
      "step": 107
    },
    {
      "epoch": 0.0063827901066753345,
      "grad_norm": 3.3307950496673584,
      "learning_rate": 5.644996082651017e-05,
      "loss": 2.5252,
      "step": 108
    },
    {
      "epoch": 0.006441890015070477,
      "grad_norm": 3.0444374084472656,
      "learning_rate": 5.553263409457504e-05,
      "loss": 2.2743,
      "step": 109
    },
    {
      "epoch": 0.006500989923465619,
      "grad_norm": 2.982682704925537,
      "learning_rate": 5.4613417973165106e-05,
      "loss": 2.3016,
      "step": 110
    },
    {
      "epoch": 0.006560089831860761,
      "grad_norm": 2.6371850967407227,
      "learning_rate": 5.3692626373743706e-05,
      "loss": 2.0164,
      "step": 111
    },
    {
      "epoch": 0.006619189740255903,
      "grad_norm": 2.7763266563415527,
      "learning_rate": 5.27705737457985e-05,
      "loss": 2.0413,
      "step": 112
    },
    {
      "epoch": 0.006678289648651045,
      "grad_norm": 3.169642448425293,
      "learning_rate": 5.184757496945726e-05,
      "loss": 1.9156,
      "step": 113
    },
    {
      "epoch": 0.006737389557046187,
      "grad_norm": 2.8326120376586914,
      "learning_rate": 5.092394524795649e-05,
      "loss": 1.9181,
      "step": 114
    },
    {
      "epoch": 0.006796489465441329,
      "grad_norm": 2.9549639225006104,
      "learning_rate": 5e-05,
      "loss": 2.2077,
      "step": 115
    },
    {
      "epoch": 0.006855589373836471,
      "grad_norm": 2.7871503829956055,
      "learning_rate": 4.907605475204352e-05,
      "loss": 1.8016,
      "step": 116
    },
    {
      "epoch": 0.006914689282231612,
      "grad_norm": 2.778618812561035,
      "learning_rate": 4.8152425030542766e-05,
      "loss": 2.1039,
      "step": 117
    },
    {
      "epoch": 0.006973789190626754,
      "grad_norm": 3.4482669830322266,
      "learning_rate": 4.72294262542015e-05,
      "loss": 2.248,
      "step": 118
    },
    {
      "epoch": 0.007032889099021896,
      "grad_norm": 2.6852567195892334,
      "learning_rate": 4.6307373626256306e-05,
      "loss": 1.8628,
      "step": 119
    },
    {
      "epoch": 0.007091989007417038,
      "grad_norm": 2.594987154006958,
      "learning_rate": 4.5386582026834906e-05,
      "loss": 1.975,
      "step": 120
    },
    {
      "epoch": 0.00715108891581218,
      "grad_norm": 3.144756555557251,
      "learning_rate": 4.446736590542497e-05,
      "loss": 2.2441,
      "step": 121
    },
    {
      "epoch": 0.007210188824207322,
      "grad_norm": 2.6541213989257812,
      "learning_rate": 4.3550039173489845e-05,
      "loss": 2.1649,
      "step": 122
    },
    {
      "epoch": 0.007269288732602464,
      "grad_norm": 2.779205083847046,
      "learning_rate": 4.2634915097268115e-05,
      "loss": 1.8455,
      "step": 123
    },
    {
      "epoch": 0.0073283886409976065,
      "grad_norm": 2.9109785556793213,
      "learning_rate": 4.1722306190793495e-05,
      "loss": 1.7731,
      "step": 124
    },
    {
      "epoch": 0.0073874885493927485,
      "grad_norm": 3.254918098449707,
      "learning_rate": 4.0812524109171476e-05,
      "loss": 2.2217,
      "step": 125
    },
    {
      "epoch": 0.007446588457787891,
      "grad_norm": 3.0662970542907715,
      "learning_rate": 3.99058795421495e-05,
      "loss": 2.0891,
      "step": 126
    },
    {
      "epoch": 0.007505688366183033,
      "grad_norm": 3.069082021713257,
      "learning_rate": 3.9002682108016585e-05,
      "loss": 1.9835,
      "step": 127
    },
    {
      "epoch": 0.007564788274578175,
      "grad_norm": 2.814826726913452,
      "learning_rate": 3.8103240247869075e-05,
      "loss": 2.012,
      "step": 128
    },
    {
      "epoch": 0.007623888182973317,
      "grad_norm": 2.723910093307495,
      "learning_rate": 3.720786112027822e-05,
      "loss": 2.1048,
      "step": 129
    },
    {
      "epoch": 0.007682988091368459,
      "grad_norm": 2.600010395050049,
      "learning_rate": 3.631685049639586e-05,
      "loss": 1.6493,
      "step": 130
    },
    {
      "epoch": 0.0077420879997636,
      "grad_norm": 4.922942638397217,
      "learning_rate": 3.543051265553377e-05,
      "loss": 2.025,
      "step": 131
    },
    {
      "epoch": 0.007801187908158742,
      "grad_norm": 3.4721429347991943,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.492,
      "step": 132
    },
    {
      "epoch": 0.007860287816553884,
      "grad_norm": 2.9339911937713623,
      "learning_rate": 3.367306435799584e-05,
      "loss": 1.8878,
      "step": 133
    },
    {
      "epoch": 0.007919387724949026,
      "grad_norm": 2.507004976272583,
      "learning_rate": 3.2802554068303596e-05,
      "loss": 1.6928,
      "step": 134
    },
    {
      "epoch": 0.007978487633344168,
      "grad_norm": 2.646677017211914,
      "learning_rate": 3.1937916690642356e-05,
      "loss": 1.9123,
      "step": 135
    },
    {
      "epoch": 0.00803758754173931,
      "grad_norm": 3.230515956878662,
      "learning_rate": 3.107944749788449e-05,
      "loss": 2.1234,
      "step": 136
    },
    {
      "epoch": 0.008096687450134452,
      "grad_norm": 2.8240675926208496,
      "learning_rate": 3.0227439656472877e-05,
      "loss": 2.0608,
      "step": 137
    },
    {
      "epoch": 0.008155787358529594,
      "grad_norm": 3.0936813354492188,
      "learning_rate": 2.9382184126304834e-05,
      "loss": 1.9963,
      "step": 138
    },
    {
      "epoch": 0.008214887266924736,
      "grad_norm": 2.895601272583008,
      "learning_rate": 2.8543969561369556e-05,
      "loss": 1.9524,
      "step": 139
    },
    {
      "epoch": 0.008273987175319878,
      "grad_norm": 2.7313430309295654,
      "learning_rate": 2.771308221117309e-05,
      "loss": 1.848,
      "step": 140
    },
    {
      "epoch": 0.00833308708371502,
      "grad_norm": 2.8361904621124268,
      "learning_rate": 2.688980582298435e-05,
      "loss": 1.9554,
      "step": 141
    },
    {
      "epoch": 0.008392186992110163,
      "grad_norm": 2.7524335384368896,
      "learning_rate": 2.607442154493568e-05,
      "loss": 1.9239,
      "step": 142
    },
    {
      "epoch": 0.008451286900505305,
      "grad_norm": 3.083930492401123,
      "learning_rate": 2.5267207830011068e-05,
      "loss": 1.9522,
      "step": 143
    },
    {
      "epoch": 0.008510386808900447,
      "grad_norm": 2.8543131351470947,
      "learning_rate": 2.446844034095466e-05,
      "loss": 1.9891,
      "step": 144
    },
    {
      "epoch": 0.008569486717295589,
      "grad_norm": 2.8486130237579346,
      "learning_rate": 2.3678391856132204e-05,
      "loss": 2.052,
      "step": 145
    },
    {
      "epoch": 0.00862858662569073,
      "grad_norm": 2.6929166316986084,
      "learning_rate": 2.2897332176377528e-05,
      "loss": 1.9629,
      "step": 146
    },
    {
      "epoch": 0.008687686534085873,
      "grad_norm": 3.058091163635254,
      "learning_rate": 2.2125528032855724e-05,
      "loss": 1.9577,
      "step": 147
    },
    {
      "epoch": 0.008746786442481015,
      "grad_norm": 3.2105419635772705,
      "learning_rate": 2.136324299597474e-05,
      "loss": 2.0027,
      "step": 148
    },
    {
      "epoch": 0.008805886350876157,
      "grad_norm": 2.93302845954895,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.977,
      "step": 149
    },
    {
      "epoch": 0.008864986259271299,
      "grad_norm": 3.1761159896850586,
      "learning_rate": 1.9868268181037185e-05,
      "loss": 1.8015,
      "step": 150
    },
    {
      "epoch": 0.008864986259271299,
      "eval_loss": 2.0016629695892334,
      "eval_runtime": 684.7232,
      "eval_samples_per_second": 41.62,
      "eval_steps_per_second": 20.81,
      "step": 150
    },
    {
      "epoch": 0.008924086167666441,
      "grad_norm": 3.083522081375122,
      "learning_rate": 1.9136088935510362e-05,
      "loss": 1.8471,
      "step": 151
    },
    {
      "epoch": 0.008983186076061581,
      "grad_norm": 3.249023914337158,
      "learning_rate": 1.8414449687337464e-05,
      "loss": 1.9572,
      "step": 152
    },
    {
      "epoch": 0.009042285984456723,
      "grad_norm": 2.9980008602142334,
      "learning_rate": 1.7703596875660645e-05,
      "loss": 2.1264,
      "step": 153
    },
    {
      "epoch": 0.009101385892851865,
      "grad_norm": 2.857516050338745,
      "learning_rate": 1.700377325606388e-05,
      "loss": 1.901,
      "step": 154
    },
    {
      "epoch": 0.009160485801247007,
      "grad_norm": 3.0055689811706543,
      "learning_rate": 1.631521781767214e-05,
      "loss": 2.192,
      "step": 155
    },
    {
      "epoch": 0.00921958570964215,
      "grad_norm": 2.5224833488464355,
      "learning_rate": 1.5638165701536868e-05,
      "loss": 1.9115,
      "step": 156
    },
    {
      "epoch": 0.009278685618037292,
      "grad_norm": 2.467513084411621,
      "learning_rate": 1.4972848120335453e-05,
      "loss": 1.7274,
      "step": 157
    },
    {
      "epoch": 0.009337785526432434,
      "grad_norm": 3.3824963569641113,
      "learning_rate": 1.4319492279412388e-05,
      "loss": 2.2169,
      "step": 158
    },
    {
      "epoch": 0.009396885434827576,
      "grad_norm": 2.666517496109009,
      "learning_rate": 1.3678321299188801e-05,
      "loss": 2.1059,
      "step": 159
    },
    {
      "epoch": 0.009455985343222718,
      "grad_norm": 3.0350444316864014,
      "learning_rate": 1.3049554138967051e-05,
      "loss": 2.3778,
      "step": 160
    },
    {
      "epoch": 0.00951508525161786,
      "grad_norm": 2.6140732765197754,
      "learning_rate": 1.2433405522156332e-05,
      "loss": 1.8671,
      "step": 161
    },
    {
      "epoch": 0.009574185160013002,
      "grad_norm": 2.8309125900268555,
      "learning_rate": 1.183008586294485e-05,
      "loss": 1.977,
      "step": 162
    },
    {
      "epoch": 0.009633285068408144,
      "grad_norm": 3.1472816467285156,
      "learning_rate": 1.1239801194443506e-05,
      "loss": 2.4342,
      "step": 163
    },
    {
      "epoch": 0.009692384976803286,
      "grad_norm": 2.7688024044036865,
      "learning_rate": 1.066275309832584e-05,
      "loss": 1.9787,
      "step": 164
    },
    {
      "epoch": 0.009751484885198428,
      "grad_norm": 3.0642006397247314,
      "learning_rate": 1.0099138635988026e-05,
      "loss": 2.1242,
      "step": 165
    },
    {
      "epoch": 0.00981058479359357,
      "grad_norm": 2.5731711387634277,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.0205,
      "step": 166
    },
    {
      "epoch": 0.009869684701988712,
      "grad_norm": 2.6096303462982178,
      "learning_rate": 9.012975854638949e-06,
      "loss": 1.9473,
      "step": 167
    },
    {
      "epoch": 0.009928784610383854,
      "grad_norm": 2.6382505893707275,
      "learning_rate": 8.490798459222476e-06,
      "loss": 1.9061,
      "step": 168
    },
    {
      "epoch": 0.009987884518778996,
      "grad_norm": 3.020261526107788,
      "learning_rate": 7.982796418105371e-06,
      "loss": 2.3045,
      "step": 169
    },
    {
      "epoch": 0.010046984427174138,
      "grad_norm": 2.3413009643554688,
      "learning_rate": 7.489143213519301e-06,
      "loss": 1.7087,
      "step": 170
    },
    {
      "epoch": 0.01010608433556928,
      "grad_norm": 2.669962167739868,
      "learning_rate": 7.010007427581378e-06,
      "loss": 2.0177,
      "step": 171
    },
    {
      "epoch": 0.010165184243964422,
      "grad_norm": 2.9112627506256104,
      "learning_rate": 6.5455526847235825e-06,
      "loss": 2.2063,
      "step": 172
    },
    {
      "epoch": 0.010224284152359564,
      "grad_norm": 2.7621371746063232,
      "learning_rate": 6.0959375958151045e-06,
      "loss": 2.1102,
      "step": 173
    },
    {
      "epoch": 0.010283384060754706,
      "grad_norm": 2.7470543384552,
      "learning_rate": 5.6613157039969055e-06,
      "loss": 1.9059,
      "step": 174
    },
    {
      "epoch": 0.010342483969149848,
      "grad_norm": 2.68990159034729,
      "learning_rate": 5.241835432246889e-06,
      "loss": 2.0888,
      "step": 175
    },
    {
      "epoch": 0.01040158387754499,
      "grad_norm": 2.371671676635742,
      "learning_rate": 4.837640032693558e-06,
      "loss": 1.6942,
      "step": 176
    },
    {
      "epoch": 0.010460683785940133,
      "grad_norm": 2.7487049102783203,
      "learning_rate": 4.448867537695578e-06,
      "loss": 1.9798,
      "step": 177
    },
    {
      "epoch": 0.010519783694335275,
      "grad_norm": 2.5755209922790527,
      "learning_rate": 4.075650712703849e-06,
      "loss": 1.8144,
      "step": 178
    },
    {
      "epoch": 0.010578883602730415,
      "grad_norm": 2.740054130554199,
      "learning_rate": 3.71811701092219e-06,
      "loss": 1.9632,
      "step": 179
    },
    {
      "epoch": 0.010637983511125557,
      "grad_norm": 2.762751340866089,
      "learning_rate": 3.376388529782215e-06,
      "loss": 2.0491,
      "step": 180
    },
    {
      "epoch": 0.010697083419520699,
      "grad_norm": 2.6476433277130127,
      "learning_rate": 3.0505819692471792e-06,
      "loss": 2.0671,
      "step": 181
    },
    {
      "epoch": 0.010756183327915841,
      "grad_norm": 3.5005996227264404,
      "learning_rate": 2.7408085919590264e-06,
      "loss": 2.2838,
      "step": 182
    },
    {
      "epoch": 0.010815283236310983,
      "grad_norm": 2.557985305786133,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.8317,
      "step": 183
    },
    {
      "epoch": 0.010874383144706125,
      "grad_norm": 2.7409958839416504,
      "learning_rate": 2.1697790249779636e-06,
      "loss": 1.9403,
      "step": 184
    },
    {
      "epoch": 0.010933483053101267,
      "grad_norm": 2.9567854404449463,
      "learning_rate": 1.908717841359048e-06,
      "loss": 2.0511,
      "step": 185
    },
    {
      "epoch": 0.01099258296149641,
      "grad_norm": 3.5258986949920654,
      "learning_rate": 1.6640797865406288e-06,
      "loss": 2.3455,
      "step": 186
    },
    {
      "epoch": 0.011051682869891551,
      "grad_norm": 2.719268560409546,
      "learning_rate": 1.4359484041943038e-06,
      "loss": 1.9524,
      "step": 187
    },
    {
      "epoch": 0.011110782778286693,
      "grad_norm": 3.0413248538970947,
      "learning_rate": 1.2244016009781701e-06,
      "loss": 2.0141,
      "step": 188
    },
    {
      "epoch": 0.011169882686681835,
      "grad_norm": 3.6312408447265625,
      "learning_rate": 1.0295116199317057e-06,
      "loss": 1.986,
      "step": 189
    },
    {
      "epoch": 0.011228982595076977,
      "grad_norm": 2.7848613262176514,
      "learning_rate": 8.513450158049108e-07,
      "loss": 1.7297,
      "step": 190
    },
    {
      "epoch": 0.01128808250347212,
      "grad_norm": 2.962815523147583,
      "learning_rate": 6.899626323298713e-07,
      "loss": 2.0131,
      "step": 191
    },
    {
      "epoch": 0.011347182411867262,
      "grad_norm": 3.577907085418701,
      "learning_rate": 5.454195814427021e-07,
      "loss": 2.4659,
      "step": 192
    },
    {
      "epoch": 0.011406282320262404,
      "grad_norm": 3.7039952278137207,
      "learning_rate": 4.177652244628627e-07,
      "loss": 2.0184,
      "step": 193
    },
    {
      "epoch": 0.011465382228657546,
      "grad_norm": 3.0766096115112305,
      "learning_rate": 3.0704315523631953e-07,
      "loss": 1.9093,
      "step": 194
    },
    {
      "epoch": 0.011524482137052688,
      "grad_norm": 2.9538846015930176,
      "learning_rate": 2.1329118524827662e-07,
      "loss": 1.9372,
      "step": 195
    },
    {
      "epoch": 0.01158358204544783,
      "grad_norm": 3.276865005493164,
      "learning_rate": 1.3654133071059893e-07,
      "loss": 2.1478,
      "step": 196
    },
    {
      "epoch": 0.011642681953842972,
      "grad_norm": 2.88531231880188,
      "learning_rate": 7.681980162830282e-08,
      "loss": 1.9631,
      "step": 197
    },
    {
      "epoch": 0.011701781862238114,
      "grad_norm": 2.8093371391296387,
      "learning_rate": 3.4146992848854695e-08,
      "loss": 1.7216,
      "step": 198
    },
    {
      "epoch": 0.011760881770633256,
      "grad_norm": 2.859862804412842,
      "learning_rate": 8.537477097364522e-09,
      "loss": 1.8161,
      "step": 199
    },
    {
      "epoch": 0.011819981679028398,
      "grad_norm": 3.6518397331237793,
      "learning_rate": 0.0,
      "loss": 2.0347,
      "step": 200
    },
    {
      "epoch": 0.011819981679028398,
      "eval_loss": 1.9855155944824219,
      "eval_runtime": 684.4118,
      "eval_samples_per_second": 41.639,
      "eval_steps_per_second": 20.819,
      "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": 1.54569648635904e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}