{
  "best_metric": 4.183354377746582,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.15939430165371588,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0007969715082685794,
      "grad_norm": 0.3416813611984253,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 4.5312,
      "step": 1
    },
    {
      "epoch": 0.0007969715082685794,
      "eval_loss": 4.786489963531494,
      "eval_runtime": 49.1755,
      "eval_samples_per_second": 42.989,
      "eval_steps_per_second": 21.494,
      "step": 1
    },
    {
      "epoch": 0.0015939430165371588,
      "grad_norm": 0.3616485893726349,
      "learning_rate": 6.666666666666667e-06,
      "loss": 4.4007,
      "step": 2
    },
    {
      "epoch": 0.002390914524805738,
      "grad_norm": 0.3799089789390564,
      "learning_rate": 1e-05,
      "loss": 4.5966,
      "step": 3
    },
    {
      "epoch": 0.0031878860330743176,
      "grad_norm": 0.3765293061733246,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 4.5636,
      "step": 4
    },
    {
      "epoch": 0.003984857541342897,
      "grad_norm": 0.41452813148498535,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 4.7637,
      "step": 5
    },
    {
      "epoch": 0.004781829049611476,
      "grad_norm": 0.4016856253147125,
      "learning_rate": 2e-05,
      "loss": 4.5039,
      "step": 6
    },
    {
      "epoch": 0.005578800557880056,
      "grad_norm": 0.411148339509964,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 4.6633,
      "step": 7
    },
    {
      "epoch": 0.006375772066148635,
      "grad_norm": 0.41650766134262085,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 4.7055,
      "step": 8
    },
    {
      "epoch": 0.007172743574417215,
      "grad_norm": 0.4098231792449951,
      "learning_rate": 3e-05,
      "loss": 4.6016,
      "step": 9
    },
    {
      "epoch": 0.007969715082685794,
      "grad_norm": 0.4001476466655731,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 4.6999,
      "step": 10
    },
    {
      "epoch": 0.008766686590954373,
      "grad_norm": 0.41843491792678833,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 4.6633,
      "step": 11
    },
    {
      "epoch": 0.009563658099222952,
      "grad_norm": 0.4096589982509613,
      "learning_rate": 4e-05,
      "loss": 4.7193,
      "step": 12
    },
    {
      "epoch": 0.010360629607491532,
      "grad_norm": 0.4085921049118042,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 4.83,
      "step": 13
    },
    {
      "epoch": 0.011157601115760111,
      "grad_norm": 0.38747555017471313,
      "learning_rate": 4.666666666666667e-05,
      "loss": 4.5337,
      "step": 14
    },
    {
      "epoch": 0.011954572624028692,
      "grad_norm": 0.36407265067100525,
      "learning_rate": 5e-05,
      "loss": 4.5965,
      "step": 15
    },
    {
      "epoch": 0.01275154413229727,
      "grad_norm": 0.3861772119998932,
      "learning_rate": 5.333333333333333e-05,
      "loss": 4.6283,
      "step": 16
    },
    {
      "epoch": 0.01354851564056585,
      "grad_norm": 0.3796427547931671,
      "learning_rate": 5.666666666666667e-05,
      "loss": 4.6653,
      "step": 17
    },
    {
      "epoch": 0.01434548714883443,
      "grad_norm": 0.3623189926147461,
      "learning_rate": 6e-05,
      "loss": 4.4623,
      "step": 18
    },
    {
      "epoch": 0.015142458657103008,
      "grad_norm": 0.34241071343421936,
      "learning_rate": 6.333333333333333e-05,
      "loss": 4.4674,
      "step": 19
    },
    {
      "epoch": 0.01593943016537159,
      "grad_norm": 0.34254637360572815,
      "learning_rate": 6.666666666666667e-05,
      "loss": 4.7388,
      "step": 20
    },
    {
      "epoch": 0.016736401673640166,
      "grad_norm": 0.32041770219802856,
      "learning_rate": 7e-05,
      "loss": 4.589,
      "step": 21
    },
    {
      "epoch": 0.017533373181908746,
      "grad_norm": 0.31564781069755554,
      "learning_rate": 7.333333333333333e-05,
      "loss": 4.3615,
      "step": 22
    },
    {
      "epoch": 0.018330344690177327,
      "grad_norm": 0.31121671199798584,
      "learning_rate": 7.666666666666667e-05,
      "loss": 4.618,
      "step": 23
    },
    {
      "epoch": 0.019127316198445904,
      "grad_norm": 0.3136454224586487,
      "learning_rate": 8e-05,
      "loss": 4.5973,
      "step": 24
    },
    {
      "epoch": 0.019924287706714484,
      "grad_norm": 0.3090231716632843,
      "learning_rate": 8.333333333333334e-05,
      "loss": 4.6476,
      "step": 25
    },
    {
      "epoch": 0.020721259214983065,
      "grad_norm": 0.33078134059906006,
      "learning_rate": 8.666666666666667e-05,
      "loss": 4.6352,
      "step": 26
    },
    {
      "epoch": 0.021518230723251645,
      "grad_norm": 0.30211010575294495,
      "learning_rate": 9e-05,
      "loss": 4.7567,
      "step": 27
    },
    {
      "epoch": 0.022315202231520222,
      "grad_norm": 0.2964538633823395,
      "learning_rate": 9.333333333333334e-05,
      "loss": 4.6998,
      "step": 28
    },
    {
      "epoch": 0.023112173739788803,
      "grad_norm": 0.321053147315979,
      "learning_rate": 9.666666666666667e-05,
      "loss": 4.5502,
      "step": 29
    },
    {
      "epoch": 0.023909145248057383,
      "grad_norm": 0.310501366853714,
      "learning_rate": 0.0001,
      "loss": 4.3409,
      "step": 30
    },
    {
      "epoch": 0.02470611675632596,
      "grad_norm": 0.3033417761325836,
      "learning_rate": 9.999146252290264e-05,
      "loss": 4.5128,
      "step": 31
    },
    {
      "epoch": 0.02550308826459454,
      "grad_norm": 0.3218533396720886,
      "learning_rate": 9.996585300715116e-05,
      "loss": 4.482,
      "step": 32
    },
    {
      "epoch": 0.02630005977286312,
      "grad_norm": 0.312926322221756,
      "learning_rate": 9.99231801983717e-05,
      "loss": 4.6661,
      "step": 33
    },
    {
      "epoch": 0.0270970312811317,
      "grad_norm": 0.2893564701080322,
      "learning_rate": 9.986345866928941e-05,
      "loss": 4.2382,
      "step": 34
    },
    {
      "epoch": 0.02789400278940028,
      "grad_norm": 0.29125991463661194,
      "learning_rate": 9.978670881475172e-05,
      "loss": 4.3398,
      "step": 35
    },
    {
      "epoch": 0.02869097429766886,
      "grad_norm": 0.3094198703765869,
      "learning_rate": 9.96929568447637e-05,
      "loss": 4.4313,
      "step": 36
    },
    {
      "epoch": 0.029487945805937436,
      "grad_norm": 0.3066606819629669,
      "learning_rate": 9.958223477553714e-05,
      "loss": 4.4231,
      "step": 37
    },
    {
      "epoch": 0.030284917314206017,
      "grad_norm": 0.33009397983551025,
      "learning_rate": 9.94545804185573e-05,
      "loss": 4.4132,
      "step": 38
    },
    {
      "epoch": 0.031081888822474597,
      "grad_norm": 0.3135601282119751,
      "learning_rate": 9.931003736767013e-05,
      "loss": 4.5012,
      "step": 39
    },
    {
      "epoch": 0.03187886033074318,
      "grad_norm": 0.3163459300994873,
      "learning_rate": 9.91486549841951e-05,
      "loss": 4.5177,
      "step": 40
    },
    {
      "epoch": 0.03267583183901176,
      "grad_norm": 0.3239397406578064,
      "learning_rate": 9.89704883800683e-05,
      "loss": 4.4014,
      "step": 41
    },
    {
      "epoch": 0.03347280334728033,
      "grad_norm": 0.31371423602104187,
      "learning_rate": 9.877559839902184e-05,
      "loss": 4.4195,
      "step": 42
    },
    {
      "epoch": 0.03426977485554891,
      "grad_norm": 0.3381875157356262,
      "learning_rate": 9.85640515958057e-05,
      "loss": 4.3083,
      "step": 43
    },
    {
      "epoch": 0.03506674636381749,
      "grad_norm": 0.33326292037963867,
      "learning_rate": 9.833592021345937e-05,
      "loss": 4.3258,
      "step": 44
    },
    {
      "epoch": 0.03586371787208607,
      "grad_norm": 0.35837581753730774,
      "learning_rate": 9.809128215864097e-05,
      "loss": 4.2037,
      "step": 45
    },
    {
      "epoch": 0.036660689380354654,
      "grad_norm": 0.3688834309577942,
      "learning_rate": 9.783022097502204e-05,
      "loss": 4.3541,
      "step": 46
    },
    {
      "epoch": 0.037457660888623234,
      "grad_norm": 0.4095333516597748,
      "learning_rate": 9.755282581475769e-05,
      "loss": 4.4061,
      "step": 47
    },
    {
      "epoch": 0.03825463239689181,
      "grad_norm": 0.4309322237968445,
      "learning_rate": 9.725919140804099e-05,
      "loss": 4.0002,
      "step": 48
    },
    {
      "epoch": 0.03905160390516039,
      "grad_norm": 0.4441930949687958,
      "learning_rate": 9.694941803075283e-05,
      "loss": 4.1638,
      "step": 49
    },
    {
      "epoch": 0.03984857541342897,
      "grad_norm": 0.6353949308395386,
      "learning_rate": 9.662361147021779e-05,
      "loss": 4.0067,
      "step": 50
    },
    {
      "epoch": 0.03984857541342897,
      "eval_loss": 4.380072593688965,
      "eval_runtime": 49.2911,
      "eval_samples_per_second": 42.888,
      "eval_steps_per_second": 21.444,
      "step": 50
    },
    {
      "epoch": 0.04064554692169755,
      "grad_norm": 0.624276340007782,
      "learning_rate": 9.628188298907782e-05,
      "loss": 4.6153,
      "step": 51
    },
    {
      "epoch": 0.04144251842996613,
      "grad_norm": 0.5702478289604187,
      "learning_rate": 9.592434928729616e-05,
      "loss": 4.5537,
      "step": 52
    },
    {
      "epoch": 0.04223948993823471,
      "grad_norm": 0.5568172931671143,
      "learning_rate": 9.555113246230442e-05,
      "loss": 4.3303,
      "step": 53
    },
    {
      "epoch": 0.04303646144650329,
      "grad_norm": 0.5124043226242065,
      "learning_rate": 9.516235996730645e-05,
      "loss": 4.4411,
      "step": 54
    },
    {
      "epoch": 0.043833432954771864,
      "grad_norm": 0.4467545747756958,
      "learning_rate": 9.475816456775313e-05,
      "loss": 4.3182,
      "step": 55
    },
    {
      "epoch": 0.044630404463040445,
      "grad_norm": 0.3773902654647827,
      "learning_rate": 9.43386842960031e-05,
      "loss": 4.3299,
      "step": 56
    },
    {
      "epoch": 0.045427375971309025,
      "grad_norm": 0.32421058416366577,
      "learning_rate": 9.39040624041849e-05,
      "loss": 4.5109,
      "step": 57
    },
    {
      "epoch": 0.046224347479577606,
      "grad_norm": 0.297109454870224,
      "learning_rate": 9.345444731527642e-05,
      "loss": 4.4707,
      "step": 58
    },
    {
      "epoch": 0.047021318987846186,
      "grad_norm": 0.2981337308883667,
      "learning_rate": 9.298999257241863e-05,
      "loss": 4.1532,
      "step": 59
    },
    {
      "epoch": 0.04781829049611477,
      "grad_norm": 0.3211049437522888,
      "learning_rate": 9.251085678648072e-05,
      "loss": 4.3803,
      "step": 60
    },
    {
      "epoch": 0.04861526200438334,
      "grad_norm": 0.2942092716693878,
      "learning_rate": 9.201720358189464e-05,
      "loss": 4.1606,
      "step": 61
    },
    {
      "epoch": 0.04941223351265192,
      "grad_norm": 0.2600642144680023,
      "learning_rate": 9.150920154077754e-05,
      "loss": 4.3328,
      "step": 62
    },
    {
      "epoch": 0.0502092050209205,
      "grad_norm": 0.2833946943283081,
      "learning_rate": 9.098702414536107e-05,
      "loss": 4.3028,
      "step": 63
    },
    {
      "epoch": 0.05100617652918908,
      "grad_norm": 0.2821059226989746,
      "learning_rate": 9.045084971874738e-05,
      "loss": 4.2651,
      "step": 64
    },
    {
      "epoch": 0.05180314803745766,
      "grad_norm": 0.24731014668941498,
      "learning_rate": 8.9900861364012e-05,
      "loss": 4.4582,
      "step": 65
    },
    {
      "epoch": 0.05260011954572624,
      "grad_norm": 0.28231242299079895,
      "learning_rate": 8.933724690167417e-05,
      "loss": 4.3979,
      "step": 66
    },
    {
      "epoch": 0.053397091053994816,
      "grad_norm": 0.286496102809906,
      "learning_rate": 8.876019880555649e-05,
      "loss": 4.2518,
      "step": 67
    },
    {
      "epoch": 0.0541940625622634,
      "grad_norm": 0.28141117095947266,
      "learning_rate": 8.816991413705516e-05,
      "loss": 4.3857,
      "step": 68
    },
    {
      "epoch": 0.05499103407053198,
      "grad_norm": 0.3005343973636627,
      "learning_rate": 8.756659447784368e-05,
      "loss": 4.2947,
      "step": 69
    },
    {
      "epoch": 0.05578800557880056,
      "grad_norm": 0.2838599383831024,
      "learning_rate": 8.695044586103296e-05,
      "loss": 4.3911,
      "step": 70
    },
    {
      "epoch": 0.05658497708706914,
      "grad_norm": 0.2848851978778839,
      "learning_rate": 8.632167870081121e-05,
      "loss": 4.1967,
      "step": 71
    },
    {
      "epoch": 0.05738194859533772,
      "grad_norm": 0.26664552092552185,
      "learning_rate": 8.568050772058762e-05,
      "loss": 4.1475,
      "step": 72
    },
    {
      "epoch": 0.0581789201036063,
      "grad_norm": 0.24472041428089142,
      "learning_rate": 8.502715187966455e-05,
      "loss": 4.0612,
      "step": 73
    },
    {
      "epoch": 0.05897589161187487,
      "grad_norm": 0.2784044146537781,
      "learning_rate": 8.436183429846313e-05,
      "loss": 4.3801,
      "step": 74
    },
    {
      "epoch": 0.05977286312014345,
      "grad_norm": 0.25716060400009155,
      "learning_rate": 8.368478218232787e-05,
      "loss": 4.2731,
      "step": 75
    },
    {
      "epoch": 0.060569834628412034,
      "grad_norm": 0.2857103943824768,
      "learning_rate": 8.299622674393614e-05,
      "loss": 4.2079,
      "step": 76
    },
    {
      "epoch": 0.061366806136680614,
      "grad_norm": 0.27076512575149536,
      "learning_rate": 8.229640312433937e-05,
      "loss": 4.3116,
      "step": 77
    },
    {
      "epoch": 0.062163777644949195,
      "grad_norm": 0.28855860233306885,
      "learning_rate": 8.158555031266254e-05,
      "loss": 4.3531,
      "step": 78
    },
    {
      "epoch": 0.06296074915321777,
      "grad_norm": 0.25299739837646484,
      "learning_rate": 8.086391106448965e-05,
      "loss": 4.2343,
      "step": 79
    },
    {
      "epoch": 0.06375772066148636,
      "grad_norm": 0.2638589143753052,
      "learning_rate": 8.013173181896283e-05,
      "loss": 4.1138,
      "step": 80
    },
    {
      "epoch": 0.06455469216975493,
      "grad_norm": 0.26530274748802185,
      "learning_rate": 7.938926261462366e-05,
      "loss": 4.2572,
      "step": 81
    },
    {
      "epoch": 0.06535166367802352,
      "grad_norm": 0.26153290271759033,
      "learning_rate": 7.863675700402526e-05,
      "loss": 4.1664,
      "step": 82
    },
    {
      "epoch": 0.06614863518629209,
      "grad_norm": 0.2668301463127136,
      "learning_rate": 7.787447196714427e-05,
      "loss": 4.1899,
      "step": 83
    },
    {
      "epoch": 0.06694560669456066,
      "grad_norm": 0.2707118093967438,
      "learning_rate": 7.710266782362247e-05,
      "loss": 4.0062,
      "step": 84
    },
    {
      "epoch": 0.06774257820282925,
      "grad_norm": 0.26167628169059753,
      "learning_rate": 7.63216081438678e-05,
      "loss": 4.1416,
      "step": 85
    },
    {
      "epoch": 0.06853954971109782,
      "grad_norm": 0.28571200370788574,
      "learning_rate": 7.553155965904535e-05,
      "loss": 4.1765,
      "step": 86
    },
    {
      "epoch": 0.06933652121936641,
      "grad_norm": 0.2865961492061615,
      "learning_rate": 7.473279216998895e-05,
      "loss": 4.0639,
      "step": 87
    },
    {
      "epoch": 0.07013349272763499,
      "grad_norm": 0.3028181493282318,
      "learning_rate": 7.392557845506432e-05,
      "loss": 4.2758,
      "step": 88
    },
    {
      "epoch": 0.07093046423590357,
      "grad_norm": 0.2957330346107483,
      "learning_rate": 7.311019417701566e-05,
      "loss": 4.0516,
      "step": 89
    },
    {
      "epoch": 0.07172743574417215,
      "grad_norm": 0.2807944118976593,
      "learning_rate": 7.228691778882693e-05,
      "loss": 3.9975,
      "step": 90
    },
    {
      "epoch": 0.07252440725244072,
      "grad_norm": 0.28562647104263306,
      "learning_rate": 7.145603043863045e-05,
      "loss": 3.9309,
      "step": 91
    },
    {
      "epoch": 0.07332137876070931,
      "grad_norm": 0.3081153333187103,
      "learning_rate": 7.061781587369519e-05,
      "loss": 4.1601,
      "step": 92
    },
    {
      "epoch": 0.07411835026897788,
      "grad_norm": 0.2935575842857361,
      "learning_rate": 6.977256034352712e-05,
      "loss": 4.1423,
      "step": 93
    },
    {
      "epoch": 0.07491532177724647,
      "grad_norm": 0.31277942657470703,
      "learning_rate": 6.892055250211552e-05,
      "loss": 4.2067,
      "step": 94
    },
    {
      "epoch": 0.07571229328551504,
      "grad_norm": 0.36462724208831787,
      "learning_rate": 6.806208330935766e-05,
      "loss": 4.132,
      "step": 95
    },
    {
      "epoch": 0.07650926479378362,
      "grad_norm": 0.3712453842163086,
      "learning_rate": 6.719744593169641e-05,
      "loss": 4.1583,
      "step": 96
    },
    {
      "epoch": 0.0773062363020522,
      "grad_norm": 0.4156951606273651,
      "learning_rate": 6.632693564200416e-05,
      "loss": 4.1631,
      "step": 97
    },
    {
      "epoch": 0.07810320781032078,
      "grad_norm": 0.542849600315094,
      "learning_rate": 6.545084971874738e-05,
      "loss": 4.3338,
      "step": 98
    },
    {
      "epoch": 0.07890017931858936,
      "grad_norm": 0.6674630045890808,
      "learning_rate": 6.456948734446624e-05,
      "loss": 3.8856,
      "step": 99
    },
    {
      "epoch": 0.07969715082685794,
      "grad_norm": 1.0902327299118042,
      "learning_rate": 6.368314950360415e-05,
      "loss": 4.1106,
      "step": 100
    },
    {
      "epoch": 0.07969715082685794,
      "eval_loss": 4.2464494705200195,
      "eval_runtime": 49.1082,
      "eval_samples_per_second": 43.048,
      "eval_steps_per_second": 21.524,
      "step": 100
    },
    {
      "epoch": 0.08049412233512653,
      "grad_norm": 0.4536077678203583,
      "learning_rate": 6.279213887972179e-05,
      "loss": 4.4386,
      "step": 101
    },
    {
      "epoch": 0.0812910938433951,
      "grad_norm": 0.49079081416130066,
      "learning_rate": 6.189675975213094e-05,
      "loss": 4.1592,
      "step": 102
    },
    {
      "epoch": 0.08208806535166367,
      "grad_norm": 0.5224800109863281,
      "learning_rate": 6.099731789198344e-05,
      "loss": 4.4185,
      "step": 103
    },
    {
      "epoch": 0.08288503685993226,
      "grad_norm": 0.4378278851509094,
      "learning_rate": 6.009412045785051e-05,
      "loss": 4.1982,
      "step": 104
    },
    {
      "epoch": 0.08368200836820083,
      "grad_norm": 0.42964717745780945,
      "learning_rate": 5.918747589082853e-05,
      "loss": 4.3514,
      "step": 105
    },
    {
      "epoch": 0.08447897987646942,
      "grad_norm": 0.4140433967113495,
      "learning_rate": 5.82776938092065e-05,
      "loss": 4.2388,
      "step": 106
    },
    {
      "epoch": 0.085275951384738,
      "grad_norm": 0.3792392909526825,
      "learning_rate": 5.736508490273188e-05,
      "loss": 4.2905,
      "step": 107
    },
    {
      "epoch": 0.08607292289300658,
      "grad_norm": 0.3357432782649994,
      "learning_rate": 5.644996082651017e-05,
      "loss": 4.1563,
      "step": 108
    },
    {
      "epoch": 0.08686989440127516,
      "grad_norm": 0.3008951246738434,
      "learning_rate": 5.553263409457504e-05,
      "loss": 4.1949,
      "step": 109
    },
    {
      "epoch": 0.08766686590954373,
      "grad_norm": 0.26610758900642395,
      "learning_rate": 5.4613417973165106e-05,
      "loss": 4.2839,
      "step": 110
    },
    {
      "epoch": 0.08846383741781232,
      "grad_norm": 0.25112292170524597,
      "learning_rate": 5.3692626373743706e-05,
      "loss": 4.273,
      "step": 111
    },
    {
      "epoch": 0.08926080892608089,
      "grad_norm": 0.25932425260543823,
      "learning_rate": 5.27705737457985e-05,
      "loss": 4.2267,
      "step": 112
    },
    {
      "epoch": 0.09005778043434948,
      "grad_norm": 0.2820190191268921,
      "learning_rate": 5.184757496945726e-05,
      "loss": 4.1222,
      "step": 113
    },
    {
      "epoch": 0.09085475194261805,
      "grad_norm": 0.2938423156738281,
      "learning_rate": 5.092394524795649e-05,
      "loss": 4.3083,
      "step": 114
    },
    {
      "epoch": 0.09165172345088662,
      "grad_norm": 0.2608668804168701,
      "learning_rate": 5e-05,
      "loss": 4.1598,
      "step": 115
    },
    {
      "epoch": 0.09244869495915521,
      "grad_norm": 0.29658424854278564,
      "learning_rate": 4.907605475204352e-05,
      "loss": 4.3735,
      "step": 116
    },
    {
      "epoch": 0.09324566646742379,
      "grad_norm": 0.3041045069694519,
      "learning_rate": 4.8152425030542766e-05,
      "loss": 4.239,
      "step": 117
    },
    {
      "epoch": 0.09404263797569237,
      "grad_norm": 0.32803621888160706,
      "learning_rate": 4.72294262542015e-05,
      "loss": 4.4035,
      "step": 118
    },
    {
      "epoch": 0.09483960948396095,
      "grad_norm": 0.27936574816703796,
      "learning_rate": 4.6307373626256306e-05,
      "loss": 3.9513,
      "step": 119
    },
    {
      "epoch": 0.09563658099222953,
      "grad_norm": 0.27713343501091003,
      "learning_rate": 4.5386582026834906e-05,
      "loss": 4.5389,
      "step": 120
    },
    {
      "epoch": 0.09643355250049811,
      "grad_norm": 0.26673904061317444,
      "learning_rate": 4.446736590542497e-05,
      "loss": 4.2087,
      "step": 121
    },
    {
      "epoch": 0.09723052400876668,
      "grad_norm": 0.26535698771476746,
      "learning_rate": 4.3550039173489845e-05,
      "loss": 4.1479,
      "step": 122
    },
    {
      "epoch": 0.09802749551703527,
      "grad_norm": 0.2699700593948364,
      "learning_rate": 4.2634915097268115e-05,
      "loss": 4.1494,
      "step": 123
    },
    {
      "epoch": 0.09882446702530384,
      "grad_norm": 0.27292296290397644,
      "learning_rate": 4.1722306190793495e-05,
      "loss": 4.223,
      "step": 124
    },
    {
      "epoch": 0.09962143853357243,
      "grad_norm": 0.274880588054657,
      "learning_rate": 4.0812524109171476e-05,
      "loss": 4.1813,
      "step": 125
    },
    {
      "epoch": 0.100418410041841,
      "grad_norm": 0.2678457200527191,
      "learning_rate": 3.99058795421495e-05,
      "loss": 4.026,
      "step": 126
    },
    {
      "epoch": 0.10121538155010959,
      "grad_norm": 0.2603027820587158,
      "learning_rate": 3.9002682108016585e-05,
      "loss": 4.0395,
      "step": 127
    },
    {
      "epoch": 0.10201235305837816,
      "grad_norm": 0.29638221859931946,
      "learning_rate": 3.8103240247869075e-05,
      "loss": 4.0431,
      "step": 128
    },
    {
      "epoch": 0.10280932456664674,
      "grad_norm": 0.27740830183029175,
      "learning_rate": 3.720786112027822e-05,
      "loss": 3.9725,
      "step": 129
    },
    {
      "epoch": 0.10360629607491532,
      "grad_norm": 0.2580147683620453,
      "learning_rate": 3.631685049639586e-05,
      "loss": 4.1768,
      "step": 130
    },
    {
      "epoch": 0.1044032675831839,
      "grad_norm": 0.2813362181186676,
      "learning_rate": 3.543051265553377e-05,
      "loss": 4.1545,
      "step": 131
    },
    {
      "epoch": 0.10520023909145249,
      "grad_norm": 0.26297634840011597,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 4.273,
      "step": 132
    },
    {
      "epoch": 0.10599721059972106,
      "grad_norm": 0.26882943511009216,
      "learning_rate": 3.367306435799584e-05,
      "loss": 4.1523,
      "step": 133
    },
    {
      "epoch": 0.10679418210798963,
      "grad_norm": 0.25994184613227844,
      "learning_rate": 3.2802554068303596e-05,
      "loss": 4.1418,
      "step": 134
    },
    {
      "epoch": 0.10759115361625822,
      "grad_norm": 0.2707768678665161,
      "learning_rate": 3.1937916690642356e-05,
      "loss": 3.9873,
      "step": 135
    },
    {
      "epoch": 0.1083881251245268,
      "grad_norm": 0.26885348558425903,
      "learning_rate": 3.107944749788449e-05,
      "loss": 4.1367,
      "step": 136
    },
    {
      "epoch": 0.10918509663279538,
      "grad_norm": 0.28563767671585083,
      "learning_rate": 3.0227439656472877e-05,
      "loss": 3.9702,
      "step": 137
    },
    {
      "epoch": 0.10998206814106395,
      "grad_norm": 0.2777661681175232,
      "learning_rate": 2.9382184126304834e-05,
      "loss": 4.0914,
      "step": 138
    },
    {
      "epoch": 0.11077903964933254,
      "grad_norm": 0.27080315351486206,
      "learning_rate": 2.8543969561369556e-05,
      "loss": 4.124,
      "step": 139
    },
    {
      "epoch": 0.11157601115760112,
      "grad_norm": 0.29918569326400757,
      "learning_rate": 2.771308221117309e-05,
      "loss": 4.1961,
      "step": 140
    },
    {
      "epoch": 0.11237298266586969,
      "grad_norm": 0.2737353444099426,
      "learning_rate": 2.688980582298435e-05,
      "loss": 3.9326,
      "step": 141
    },
    {
      "epoch": 0.11316995417413828,
      "grad_norm": 0.3046242296695709,
      "learning_rate": 2.607442154493568e-05,
      "loss": 4.2884,
      "step": 142
    },
    {
      "epoch": 0.11396692568240685,
      "grad_norm": 0.2995275557041168,
      "learning_rate": 2.5267207830011068e-05,
      "loss": 4.0685,
      "step": 143
    },
    {
      "epoch": 0.11476389719067544,
      "grad_norm": 0.3147706091403961,
      "learning_rate": 2.446844034095466e-05,
      "loss": 4.1747,
      "step": 144
    },
    {
      "epoch": 0.11556086869894401,
      "grad_norm": 0.31899207830429077,
      "learning_rate": 2.3678391856132204e-05,
      "loss": 4.0118,
      "step": 145
    },
    {
      "epoch": 0.1163578402072126,
      "grad_norm": 0.3425534963607788,
      "learning_rate": 2.2897332176377528e-05,
      "loss": 4.0294,
      "step": 146
    },
    {
      "epoch": 0.11715481171548117,
      "grad_norm": 0.39953556656837463,
      "learning_rate": 2.2125528032855724e-05,
      "loss": 4.0949,
      "step": 147
    },
    {
      "epoch": 0.11795178322374975,
      "grad_norm": 0.4550447463989258,
      "learning_rate": 2.136324299597474e-05,
      "loss": 3.8362,
      "step": 148
    },
    {
      "epoch": 0.11874875473201833,
      "grad_norm": 0.5356332659721375,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.2568,
      "step": 149
    },
    {
      "epoch": 0.1195457262402869,
      "grad_norm": 0.9250167608261108,
      "learning_rate": 1.9868268181037185e-05,
      "loss": 4.1155,
      "step": 150
    },
    {
      "epoch": 0.1195457262402869,
      "eval_loss": 4.1894659996032715,
      "eval_runtime": 49.1569,
      "eval_samples_per_second": 43.005,
      "eval_steps_per_second": 21.503,
      "step": 150
    },
    {
      "epoch": 0.1203426977485555,
      "grad_norm": 0.2653471529483795,
      "learning_rate": 1.9136088935510362e-05,
      "loss": 4.3249,
      "step": 151
    },
    {
      "epoch": 0.12113966925682407,
      "grad_norm": 0.30772659182548523,
      "learning_rate": 1.8414449687337464e-05,
      "loss": 4.3166,
      "step": 152
    },
    {
      "epoch": 0.12193664076509265,
      "grad_norm": 0.30526572465896606,
      "learning_rate": 1.7703596875660645e-05,
      "loss": 4.3137,
      "step": 153
    },
    {
      "epoch": 0.12273361227336123,
      "grad_norm": 0.2967657744884491,
      "learning_rate": 1.700377325606388e-05,
      "loss": 4.4396,
      "step": 154
    },
    {
      "epoch": 0.1235305837816298,
      "grad_norm": 0.3225037157535553,
      "learning_rate": 1.631521781767214e-05,
      "loss": 4.2345,
      "step": 155
    },
    {
      "epoch": 0.12432755528989839,
      "grad_norm": 0.3195071816444397,
      "learning_rate": 1.5638165701536868e-05,
      "loss": 4.3621,
      "step": 156
    },
    {
      "epoch": 0.12512452679816696,
      "grad_norm": 0.31256967782974243,
      "learning_rate": 1.4972848120335453e-05,
      "loss": 4.4553,
      "step": 157
    },
    {
      "epoch": 0.12592149830643554,
      "grad_norm": 0.32809165120124817,
      "learning_rate": 1.4319492279412388e-05,
      "loss": 4.2234,
      "step": 158
    },
    {
      "epoch": 0.12671846981470414,
      "grad_norm": 0.2773759365081787,
      "learning_rate": 1.3678321299188801e-05,
      "loss": 4.277,
      "step": 159
    },
    {
      "epoch": 0.1275154413229727,
      "grad_norm": 0.2871479094028473,
      "learning_rate": 1.3049554138967051e-05,
      "loss": 4.1204,
      "step": 160
    },
    {
      "epoch": 0.12831241283124128,
      "grad_norm": 0.29100486636161804,
      "learning_rate": 1.2433405522156332e-05,
      "loss": 4.2218,
      "step": 161
    },
    {
      "epoch": 0.12910938433950986,
      "grad_norm": 0.2950289845466614,
      "learning_rate": 1.183008586294485e-05,
      "loss": 4.3895,
      "step": 162
    },
    {
      "epoch": 0.12990635584777843,
      "grad_norm": 0.2688523530960083,
      "learning_rate": 1.1239801194443506e-05,
      "loss": 4.078,
      "step": 163
    },
    {
      "epoch": 0.13070332735604703,
      "grad_norm": 0.26284605264663696,
      "learning_rate": 1.066275309832584e-05,
      "loss": 4.1274,
      "step": 164
    },
    {
      "epoch": 0.1315002988643156,
      "grad_norm": 0.2650034725666046,
      "learning_rate": 1.0099138635988026e-05,
      "loss": 4.2154,
      "step": 165
    },
    {
      "epoch": 0.13229727037258418,
      "grad_norm": 0.25614309310913086,
      "learning_rate": 9.549150281252633e-06,
      "loss": 4.3905,
      "step": 166
    },
    {
      "epoch": 0.13309424188085275,
      "grad_norm": 0.23772862553596497,
      "learning_rate": 9.012975854638949e-06,
      "loss": 4.0156,
      "step": 167
    },
    {
      "epoch": 0.13389121338912133,
      "grad_norm": 0.2594856023788452,
      "learning_rate": 8.490798459222476e-06,
      "loss": 4.1311,
      "step": 168
    },
    {
      "epoch": 0.13468818489738993,
      "grad_norm": 0.2590104937553406,
      "learning_rate": 7.982796418105371e-06,
      "loss": 4.2615,
      "step": 169
    },
    {
      "epoch": 0.1354851564056585,
      "grad_norm": 0.25985562801361084,
      "learning_rate": 7.489143213519301e-06,
      "loss": 4.0188,
      "step": 170
    },
    {
      "epoch": 0.13628212791392708,
      "grad_norm": 0.2538761794567108,
      "learning_rate": 7.010007427581378e-06,
      "loss": 4.2036,
      "step": 171
    },
    {
      "epoch": 0.13707909942219565,
      "grad_norm": 0.25580894947052,
      "learning_rate": 6.5455526847235825e-06,
      "loss": 4.2554,
      "step": 172
    },
    {
      "epoch": 0.13787607093046422,
      "grad_norm": 0.24640843272209167,
      "learning_rate": 6.0959375958151045e-06,
      "loss": 3.9291,
      "step": 173
    },
    {
      "epoch": 0.13867304243873282,
      "grad_norm": 0.2420421540737152,
      "learning_rate": 5.6613157039969055e-06,
      "loss": 4.2622,
      "step": 174
    },
    {
      "epoch": 0.1394700139470014,
      "grad_norm": 0.2577573359012604,
      "learning_rate": 5.241835432246889e-06,
      "loss": 3.9891,
      "step": 175
    },
    {
      "epoch": 0.14026698545526997,
      "grad_norm": 0.2526918649673462,
      "learning_rate": 4.837640032693558e-06,
      "loss": 4.04,
      "step": 176
    },
    {
      "epoch": 0.14106395696353854,
      "grad_norm": 0.2588857412338257,
      "learning_rate": 4.448867537695578e-06,
      "loss": 4.1807,
      "step": 177
    },
    {
      "epoch": 0.14186092847180715,
      "grad_norm": 0.24917973577976227,
      "learning_rate": 4.075650712703849e-06,
      "loss": 4.2169,
      "step": 178
    },
    {
      "epoch": 0.14265789998007572,
      "grad_norm": 0.25603026151657104,
      "learning_rate": 3.71811701092219e-06,
      "loss": 3.9137,
      "step": 179
    },
    {
      "epoch": 0.1434548714883443,
      "grad_norm": 0.26696962118148804,
      "learning_rate": 3.376388529782215e-06,
      "loss": 4.2429,
      "step": 180
    },
    {
      "epoch": 0.14425184299661287,
      "grad_norm": 0.2555984556674957,
      "learning_rate": 3.0505819692471792e-06,
      "loss": 4.2023,
      "step": 181
    },
    {
      "epoch": 0.14504881450488144,
      "grad_norm": 0.27253782749176025,
      "learning_rate": 2.7408085919590264e-06,
      "loss": 4.0074,
      "step": 182
    },
    {
      "epoch": 0.14584578601315004,
      "grad_norm": 0.27349719405174255,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 4.274,
      "step": 183
    },
    {
      "epoch": 0.14664275752141862,
      "grad_norm": 0.31064373254776,
      "learning_rate": 2.1697790249779636e-06,
      "loss": 4.4011,
      "step": 184
    },
    {
      "epoch": 0.1474397290296872,
      "grad_norm": 0.2815253436565399,
      "learning_rate": 1.908717841359048e-06,
      "loss": 4.1062,
      "step": 185
    },
    {
      "epoch": 0.14823670053795576,
      "grad_norm": 0.27812743186950684,
      "learning_rate": 1.6640797865406288e-06,
      "loss": 4.0384,
      "step": 186
    },
    {
      "epoch": 0.14903367204622434,
      "grad_norm": 0.2721196711063385,
      "learning_rate": 1.4359484041943038e-06,
      "loss": 4.0728,
      "step": 187
    },
    {
      "epoch": 0.14983064355449294,
      "grad_norm": 0.31800466775894165,
      "learning_rate": 1.2244016009781701e-06,
      "loss": 4.102,
      "step": 188
    },
    {
      "epoch": 0.1506276150627615,
      "grad_norm": 0.3215753734111786,
      "learning_rate": 1.0295116199317057e-06,
      "loss": 4.1268,
      "step": 189
    },
    {
      "epoch": 0.15142458657103008,
      "grad_norm": 0.29382383823394775,
      "learning_rate": 8.513450158049108e-07,
      "loss": 4.0609,
      "step": 190
    },
    {
      "epoch": 0.15222155807929866,
      "grad_norm": 0.2936742305755615,
      "learning_rate": 6.899626323298713e-07,
      "loss": 4.2683,
      "step": 191
    },
    {
      "epoch": 0.15301852958756723,
      "grad_norm": 0.2987104058265686,
      "learning_rate": 5.454195814427021e-07,
      "loss": 4.0856,
      "step": 192
    },
    {
      "epoch": 0.15381550109583583,
      "grad_norm": 0.31946101784706116,
      "learning_rate": 4.177652244628627e-07,
      "loss": 4.1578,
      "step": 193
    },
    {
      "epoch": 0.1546124726041044,
      "grad_norm": 0.2962910532951355,
      "learning_rate": 3.0704315523631953e-07,
      "loss": 4.0711,
      "step": 194
    },
    {
      "epoch": 0.15540944411237298,
      "grad_norm": 0.3355628550052643,
      "learning_rate": 2.1329118524827662e-07,
      "loss": 4.1497,
      "step": 195
    },
    {
      "epoch": 0.15620641562064155,
      "grad_norm": 0.3637641966342926,
      "learning_rate": 1.3654133071059893e-07,
      "loss": 4.1806,
      "step": 196
    },
    {
      "epoch": 0.15700338712891015,
      "grad_norm": 0.35605573654174805,
      "learning_rate": 7.681980162830282e-08,
      "loss": 3.9998,
      "step": 197
    },
    {
      "epoch": 0.15780035863717873,
      "grad_norm": 0.4277958869934082,
      "learning_rate": 3.4146992848854695e-08,
      "loss": 4.1548,
      "step": 198
    },
    {
      "epoch": 0.1585973301454473,
      "grad_norm": 0.5168968439102173,
      "learning_rate": 8.537477097364522e-09,
      "loss": 4.1243,
      "step": 199
    },
    {
      "epoch": 0.15939430165371588,
      "grad_norm": 0.9556027054786682,
      "learning_rate": 0.0,
      "loss": 4.2316,
      "step": 200
    },
    {
      "epoch": 0.15939430165371588,
      "eval_loss": 4.183354377746582,
      "eval_runtime": 49.1904,
      "eval_samples_per_second": 42.976,
      "eval_steps_per_second": 21.488,
      "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": 4944352090521600.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}