{
  "best_metric": 3.1328935623168945,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.07432181345224824,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003716090672612412,
      "grad_norm": 1.395404577255249,
      "learning_rate": 5e-06,
      "loss": 3.8589,
      "step": 1
    },
    {
      "epoch": 0.0003716090672612412,
      "eval_loss": 3.698608160018921,
      "eval_runtime": 110.3096,
      "eval_samples_per_second": 41.084,
      "eval_steps_per_second": 20.542,
      "step": 1
    },
    {
      "epoch": 0.0007432181345224824,
      "grad_norm": 1.495450735092163,
      "learning_rate": 1e-05,
      "loss": 3.8595,
      "step": 2
    },
    {
      "epoch": 0.0011148272017837235,
      "grad_norm": 1.708409309387207,
      "learning_rate": 1.5e-05,
      "loss": 3.8337,
      "step": 3
    },
    {
      "epoch": 0.0014864362690449647,
      "grad_norm": 1.6244080066680908,
      "learning_rate": 2e-05,
      "loss": 3.9127,
      "step": 4
    },
    {
      "epoch": 0.0018580453363062058,
      "grad_norm": 1.6346101760864258,
      "learning_rate": 2.5e-05,
      "loss": 3.7906,
      "step": 5
    },
    {
      "epoch": 0.002229654403567447,
      "grad_norm": 1.5767675638198853,
      "learning_rate": 3e-05,
      "loss": 3.5424,
      "step": 6
    },
    {
      "epoch": 0.002601263470828688,
      "grad_norm": 1.6814939975738525,
      "learning_rate": 3.5e-05,
      "loss": 3.7044,
      "step": 7
    },
    {
      "epoch": 0.0029728725380899295,
      "grad_norm": 1.6342999935150146,
      "learning_rate": 4e-05,
      "loss": 3.7866,
      "step": 8
    },
    {
      "epoch": 0.0033444816053511705,
      "grad_norm": 1.6840134859085083,
      "learning_rate": 4.5e-05,
      "loss": 3.5734,
      "step": 9
    },
    {
      "epoch": 0.0037160906726124115,
      "grad_norm": 1.8313182592391968,
      "learning_rate": 5e-05,
      "loss": 3.8096,
      "step": 10
    },
    {
      "epoch": 0.004087699739873653,
      "grad_norm": 1.8736138343811035,
      "learning_rate": 5.500000000000001e-05,
      "loss": 3.56,
      "step": 11
    },
    {
      "epoch": 0.004459308807134894,
      "grad_norm": 1.9644217491149902,
      "learning_rate": 6e-05,
      "loss": 3.6562,
      "step": 12
    },
    {
      "epoch": 0.004830917874396135,
      "grad_norm": 1.846778392791748,
      "learning_rate": 6.500000000000001e-05,
      "loss": 3.502,
      "step": 13
    },
    {
      "epoch": 0.005202526941657376,
      "grad_norm": 1.958957314491272,
      "learning_rate": 7e-05,
      "loss": 3.7459,
      "step": 14
    },
    {
      "epoch": 0.005574136008918618,
      "grad_norm": 1.9100260734558105,
      "learning_rate": 7.500000000000001e-05,
      "loss": 3.5192,
      "step": 15
    },
    {
      "epoch": 0.005945745076179859,
      "grad_norm": 1.9562067985534668,
      "learning_rate": 8e-05,
      "loss": 3.5551,
      "step": 16
    },
    {
      "epoch": 0.0063173541434411,
      "grad_norm": 2.0819551944732666,
      "learning_rate": 8.5e-05,
      "loss": 3.6895,
      "step": 17
    },
    {
      "epoch": 0.006688963210702341,
      "grad_norm": 2.0404553413391113,
      "learning_rate": 9e-05,
      "loss": 3.5621,
      "step": 18
    },
    {
      "epoch": 0.007060572277963582,
      "grad_norm": 2.0112435817718506,
      "learning_rate": 9.5e-05,
      "loss": 3.4923,
      "step": 19
    },
    {
      "epoch": 0.007432181345224823,
      "grad_norm": 2.050147771835327,
      "learning_rate": 0.0001,
      "loss": 3.7079,
      "step": 20
    },
    {
      "epoch": 0.007803790412486065,
      "grad_norm": 1.969068169593811,
      "learning_rate": 9.999238475781957e-05,
      "loss": 3.5209,
      "step": 21
    },
    {
      "epoch": 0.008175399479747306,
      "grad_norm": 2.1419968605041504,
      "learning_rate": 9.99695413509548e-05,
      "loss": 3.5605,
      "step": 22
    },
    {
      "epoch": 0.008547008547008548,
      "grad_norm": 1.9498823881149292,
      "learning_rate": 9.99314767377287e-05,
      "loss": 3.4271,
      "step": 23
    },
    {
      "epoch": 0.008918617614269788,
      "grad_norm": 2.0768706798553467,
      "learning_rate": 9.987820251299122e-05,
      "loss": 3.4113,
      "step": 24
    },
    {
      "epoch": 0.00929022668153103,
      "grad_norm": 2.0907914638519287,
      "learning_rate": 9.980973490458728e-05,
      "loss": 3.3731,
      "step": 25
    },
    {
      "epoch": 0.00966183574879227,
      "grad_norm": 2.208066463470459,
      "learning_rate": 9.972609476841367e-05,
      "loss": 3.3223,
      "step": 26
    },
    {
      "epoch": 0.010033444816053512,
      "grad_norm": 2.1362009048461914,
      "learning_rate": 9.962730758206611e-05,
      "loss": 3.1932,
      "step": 27
    },
    {
      "epoch": 0.010405053883314752,
      "grad_norm": 2.2497036457061768,
      "learning_rate": 9.951340343707852e-05,
      "loss": 3.2557,
      "step": 28
    },
    {
      "epoch": 0.010776662950575994,
      "grad_norm": 2.3101062774658203,
      "learning_rate": 9.938441702975689e-05,
      "loss": 3.2662,
      "step": 29
    },
    {
      "epoch": 0.011148272017837236,
      "grad_norm": 2.285050392150879,
      "learning_rate": 9.924038765061042e-05,
      "loss": 3.3092,
      "step": 30
    },
    {
      "epoch": 0.011519881085098476,
      "grad_norm": 2.3182809352874756,
      "learning_rate": 9.908135917238321e-05,
      "loss": 3.2587,
      "step": 31
    },
    {
      "epoch": 0.011891490152359718,
      "grad_norm": 2.44177508354187,
      "learning_rate": 9.890738003669029e-05,
      "loss": 3.2171,
      "step": 32
    },
    {
      "epoch": 0.012263099219620958,
      "grad_norm": 2.3121793270111084,
      "learning_rate": 9.871850323926177e-05,
      "loss": 3.2153,
      "step": 33
    },
    {
      "epoch": 0.0126347082868822,
      "grad_norm": 2.3896982669830322,
      "learning_rate": 9.851478631379982e-05,
      "loss": 3.3923,
      "step": 34
    },
    {
      "epoch": 0.013006317354143442,
      "grad_norm": 2.573725700378418,
      "learning_rate": 9.829629131445342e-05,
      "loss": 3.0088,
      "step": 35
    },
    {
      "epoch": 0.013377926421404682,
      "grad_norm": 2.594827175140381,
      "learning_rate": 9.806308479691595e-05,
      "loss": 3.1635,
      "step": 36
    },
    {
      "epoch": 0.013749535488665924,
      "grad_norm": 2.49944806098938,
      "learning_rate": 9.781523779815179e-05,
      "loss": 3.1225,
      "step": 37
    },
    {
      "epoch": 0.014121144555927164,
      "grad_norm": 2.5924439430236816,
      "learning_rate": 9.755282581475769e-05,
      "loss": 3.2409,
      "step": 38
    },
    {
      "epoch": 0.014492753623188406,
      "grad_norm": 2.622465133666992,
      "learning_rate": 9.727592877996585e-05,
      "loss": 2.8234,
      "step": 39
    },
    {
      "epoch": 0.014864362690449646,
      "grad_norm": 2.6936821937561035,
      "learning_rate": 9.698463103929542e-05,
      "loss": 3.1865,
      "step": 40
    },
    {
      "epoch": 0.015235971757710888,
      "grad_norm": 2.5600905418395996,
      "learning_rate": 9.667902132486009e-05,
      "loss": 3.2298,
      "step": 41
    },
    {
      "epoch": 0.01560758082497213,
      "grad_norm": 2.606717824935913,
      "learning_rate": 9.635919272833938e-05,
      "loss": 3.2275,
      "step": 42
    },
    {
      "epoch": 0.015979189892233372,
      "grad_norm": 2.7310972213745117,
      "learning_rate": 9.602524267262203e-05,
      "loss": 3.1006,
      "step": 43
    },
    {
      "epoch": 0.016350798959494612,
      "grad_norm": 2.696859836578369,
      "learning_rate": 9.567727288213005e-05,
      "loss": 3.2682,
      "step": 44
    },
    {
      "epoch": 0.016722408026755852,
      "grad_norm": 3.2989847660064697,
      "learning_rate": 9.53153893518325e-05,
      "loss": 3.4225,
      "step": 45
    },
    {
      "epoch": 0.017094017094017096,
      "grad_norm": 2.866588830947876,
      "learning_rate": 9.493970231495835e-05,
      "loss": 3.248,
      "step": 46
    },
    {
      "epoch": 0.017465626161278336,
      "grad_norm": 2.8411591053009033,
      "learning_rate": 9.45503262094184e-05,
      "loss": 2.9874,
      "step": 47
    },
    {
      "epoch": 0.017837235228539576,
      "grad_norm": 3.2578847408294678,
      "learning_rate": 9.414737964294636e-05,
      "loss": 3.248,
      "step": 48
    },
    {
      "epoch": 0.018208844295800816,
      "grad_norm": 2.9824798107147217,
      "learning_rate": 9.373098535696979e-05,
      "loss": 3.599,
      "step": 49
    },
    {
      "epoch": 0.01858045336306206,
      "grad_norm": 3.438220500946045,
      "learning_rate": 9.330127018922194e-05,
      "loss": 3.7818,
      "step": 50
    },
    {
      "epoch": 0.01858045336306206,
      "eval_loss": 3.3963160514831543,
      "eval_runtime": 110.1439,
      "eval_samples_per_second": 41.146,
      "eval_steps_per_second": 20.573,
      "step": 50
    },
    {
      "epoch": 0.0189520624303233,
      "grad_norm": 3.2213802337646484,
      "learning_rate": 9.285836503510562e-05,
      "loss": 3.7682,
      "step": 51
    },
    {
      "epoch": 0.01932367149758454,
      "grad_norm": 2.897183418273926,
      "learning_rate": 9.24024048078213e-05,
      "loss": 3.6962,
      "step": 52
    },
    {
      "epoch": 0.019695280564845784,
      "grad_norm": 2.333893299102783,
      "learning_rate": 9.193352839727121e-05,
      "loss": 3.55,
      "step": 53
    },
    {
      "epoch": 0.020066889632107024,
      "grad_norm": 2.0622718334198,
      "learning_rate": 9.145187862775209e-05,
      "loss": 3.6484,
      "step": 54
    },
    {
      "epoch": 0.020438498699368264,
      "grad_norm": 1.9058257341384888,
      "learning_rate": 9.09576022144496e-05,
      "loss": 3.5701,
      "step": 55
    },
    {
      "epoch": 0.020810107766629504,
      "grad_norm": 1.4503419399261475,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.4605,
      "step": 56
    },
    {
      "epoch": 0.021181716833890748,
      "grad_norm": 1.41510808467865,
      "learning_rate": 8.993177550236464e-05,
      "loss": 3.3843,
      "step": 57
    },
    {
      "epoch": 0.021553325901151988,
      "grad_norm": 1.5639384984970093,
      "learning_rate": 8.940053768033609e-05,
      "loss": 3.2906,
      "step": 58
    },
    {
      "epoch": 0.021924934968413228,
      "grad_norm": 1.5347603559494019,
      "learning_rate": 8.885729807284856e-05,
      "loss": 3.586,
      "step": 59
    },
    {
      "epoch": 0.022296544035674472,
      "grad_norm": 1.5613843202590942,
      "learning_rate": 8.83022221559489e-05,
      "loss": 3.3605,
      "step": 60
    },
    {
      "epoch": 0.022668153102935712,
      "grad_norm": 1.5832593441009521,
      "learning_rate": 8.773547901113862e-05,
      "loss": 3.415,
      "step": 61
    },
    {
      "epoch": 0.023039762170196952,
      "grad_norm": 1.6786587238311768,
      "learning_rate": 8.715724127386972e-05,
      "loss": 3.3806,
      "step": 62
    },
    {
      "epoch": 0.023411371237458192,
      "grad_norm": 1.6316285133361816,
      "learning_rate": 8.656768508095853e-05,
      "loss": 3.4167,
      "step": 63
    },
    {
      "epoch": 0.023782980304719436,
      "grad_norm": 1.6437264680862427,
      "learning_rate": 8.596699001693255e-05,
      "loss": 3.4412,
      "step": 64
    },
    {
      "epoch": 0.024154589371980676,
      "grad_norm": 1.5929906368255615,
      "learning_rate": 8.535533905932738e-05,
      "loss": 3.379,
      "step": 65
    },
    {
      "epoch": 0.024526198439241916,
      "grad_norm": 1.6711010932922363,
      "learning_rate": 8.473291852294987e-05,
      "loss": 3.2288,
      "step": 66
    },
    {
      "epoch": 0.02489780750650316,
      "grad_norm": 1.816035509109497,
      "learning_rate": 8.409991800312493e-05,
      "loss": 3.2581,
      "step": 67
    },
    {
      "epoch": 0.0252694165737644,
      "grad_norm": 1.5691595077514648,
      "learning_rate": 8.345653031794292e-05,
      "loss": 3.3002,
      "step": 68
    },
    {
      "epoch": 0.02564102564102564,
      "grad_norm": 1.7285726070404053,
      "learning_rate": 8.280295144952536e-05,
      "loss": 3.3853,
      "step": 69
    },
    {
      "epoch": 0.026012634708286884,
      "grad_norm": 1.6965199708938599,
      "learning_rate": 8.213938048432697e-05,
      "loss": 3.1998,
      "step": 70
    },
    {
      "epoch": 0.026384243775548124,
      "grad_norm": 1.7152042388916016,
      "learning_rate": 8.146601955249188e-05,
      "loss": 3.304,
      "step": 71
    },
    {
      "epoch": 0.026755852842809364,
      "grad_norm": 1.8366388082504272,
      "learning_rate": 8.07830737662829e-05,
      "loss": 3.1231,
      "step": 72
    },
    {
      "epoch": 0.027127461910070604,
      "grad_norm": 1.8617119789123535,
      "learning_rate": 8.009075115760243e-05,
      "loss": 3.3145,
      "step": 73
    },
    {
      "epoch": 0.027499070977331848,
      "grad_norm": 1.8905341625213623,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.3134,
      "step": 74
    },
    {
      "epoch": 0.027870680044593088,
      "grad_norm": 1.882635474205017,
      "learning_rate": 7.86788218175523e-05,
      "loss": 3.3169,
      "step": 75
    },
    {
      "epoch": 0.028242289111854328,
      "grad_norm": 2.0442984104156494,
      "learning_rate": 7.795964517353735e-05,
      "loss": 3.1649,
      "step": 76
    },
    {
      "epoch": 0.028613898179115572,
      "grad_norm": 1.940576434135437,
      "learning_rate": 7.723195175075136e-05,
      "loss": 3.1936,
      "step": 77
    },
    {
      "epoch": 0.028985507246376812,
      "grad_norm": 2.0615429878234863,
      "learning_rate": 7.649596321166024e-05,
      "loss": 3.3171,
      "step": 78
    },
    {
      "epoch": 0.029357116313638052,
      "grad_norm": 2.0494372844696045,
      "learning_rate": 7.575190374550272e-05,
      "loss": 3.2035,
      "step": 79
    },
    {
      "epoch": 0.029728725380899292,
      "grad_norm": 2.0333774089813232,
      "learning_rate": 7.500000000000001e-05,
      "loss": 3.1547,
      "step": 80
    },
    {
      "epoch": 0.030100334448160536,
      "grad_norm": 2.141954183578491,
      "learning_rate": 7.424048101231686e-05,
      "loss": 2.9143,
      "step": 81
    },
    {
      "epoch": 0.030471943515421776,
      "grad_norm": 2.22705340385437,
      "learning_rate": 7.347357813929454e-05,
      "loss": 3.1911,
      "step": 82
    },
    {
      "epoch": 0.030843552582683016,
      "grad_norm": 2.196378231048584,
      "learning_rate": 7.269952498697734e-05,
      "loss": 3.2222,
      "step": 83
    },
    {
      "epoch": 0.03121516164994426,
      "grad_norm": 2.2278859615325928,
      "learning_rate": 7.191855733945387e-05,
      "loss": 3.0329,
      "step": 84
    },
    {
      "epoch": 0.031586770717205497,
      "grad_norm": 2.3488399982452393,
      "learning_rate": 7.113091308703498e-05,
      "loss": 2.9382,
      "step": 85
    },
    {
      "epoch": 0.031958379784466744,
      "grad_norm": 2.3422460556030273,
      "learning_rate": 7.033683215379002e-05,
      "loss": 3.1308,
      "step": 86
    },
    {
      "epoch": 0.032329988851727984,
      "grad_norm": 2.4222495555877686,
      "learning_rate": 6.953655642446368e-05,
      "loss": 2.9883,
      "step": 87
    },
    {
      "epoch": 0.032701597918989224,
      "grad_norm": 2.425354242324829,
      "learning_rate": 6.873032967079561e-05,
      "loss": 3.1242,
      "step": 88
    },
    {
      "epoch": 0.033073206986250464,
      "grad_norm": 2.56612491607666,
      "learning_rate": 6.7918397477265e-05,
      "loss": 3.0771,
      "step": 89
    },
    {
      "epoch": 0.033444816053511704,
      "grad_norm": 2.566655397415161,
      "learning_rate": 6.710100716628344e-05,
      "loss": 2.9399,
      "step": 90
    },
    {
      "epoch": 0.033816425120772944,
      "grad_norm": 2.3641583919525146,
      "learning_rate": 6.627840772285784e-05,
      "loss": 2.9741,
      "step": 91
    },
    {
      "epoch": 0.03418803418803419,
      "grad_norm": 2.47420334815979,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.7321,
      "step": 92
    },
    {
      "epoch": 0.03455964325529543,
      "grad_norm": 2.7138381004333496,
      "learning_rate": 6.461858523613684e-05,
      "loss": 3.027,
      "step": 93
    },
    {
      "epoch": 0.03493125232255667,
      "grad_norm": 2.655938148498535,
      "learning_rate": 6.378186779084995e-05,
      "loss": 2.9049,
      "step": 94
    },
    {
      "epoch": 0.03530286138981791,
      "grad_norm": 2.87831974029541,
      "learning_rate": 6.294095225512603e-05,
      "loss": 3.2944,
      "step": 95
    },
    {
      "epoch": 0.03567447045707915,
      "grad_norm": 2.891333818435669,
      "learning_rate": 6.209609477998338e-05,
      "loss": 3.2566,
      "step": 96
    },
    {
      "epoch": 0.03604607952434039,
      "grad_norm": 2.8243558406829834,
      "learning_rate": 6.124755271719325e-05,
      "loss": 3.1022,
      "step": 97
    },
    {
      "epoch": 0.03641768859160163,
      "grad_norm": 3.1256723403930664,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 3.3878,
      "step": 98
    },
    {
      "epoch": 0.03678929765886288,
      "grad_norm": 3.284562349319458,
      "learning_rate": 5.9540449768827246e-05,
      "loss": 3.376,
      "step": 99
    },
    {
      "epoch": 0.03716090672612412,
      "grad_norm": 3.2456462383270264,
      "learning_rate": 5.868240888334653e-05,
      "loss": 3.7382,
      "step": 100
    },
    {
      "epoch": 0.03716090672612412,
      "eval_loss": 3.239711284637451,
      "eval_runtime": 109.8084,
      "eval_samples_per_second": 41.272,
      "eval_steps_per_second": 20.636,
      "step": 100
    },
    {
      "epoch": 0.03753251579338536,
      "grad_norm": 1.8480045795440674,
      "learning_rate": 5.782172325201155e-05,
      "loss": 3.4696,
      "step": 101
    },
    {
      "epoch": 0.0379041248606466,
      "grad_norm": 1.9059332609176636,
      "learning_rate": 5.695865504800327e-05,
      "loss": 3.5296,
      "step": 102
    },
    {
      "epoch": 0.03827573392790784,
      "grad_norm": 1.69303297996521,
      "learning_rate": 5.6093467170257374e-05,
      "loss": 3.3444,
      "step": 103
    },
    {
      "epoch": 0.03864734299516908,
      "grad_norm": 1.7022647857666016,
      "learning_rate": 5.522642316338268e-05,
      "loss": 3.4552,
      "step": 104
    },
    {
      "epoch": 0.03901895206243032,
      "grad_norm": 1.520664930343628,
      "learning_rate": 5.435778713738292e-05,
      "loss": 3.2903,
      "step": 105
    },
    {
      "epoch": 0.03939056112969157,
      "grad_norm": 1.3997753858566284,
      "learning_rate": 5.348782368720626e-05,
      "loss": 3.3293,
      "step": 106
    },
    {
      "epoch": 0.03976217019695281,
      "grad_norm": 1.373949646949768,
      "learning_rate": 5.26167978121472e-05,
      "loss": 3.3496,
      "step": 107
    },
    {
      "epoch": 0.04013377926421405,
      "grad_norm": 1.4093092679977417,
      "learning_rate": 5.174497483512506e-05,
      "loss": 3.2133,
      "step": 108
    },
    {
      "epoch": 0.04050538833147529,
      "grad_norm": 1.5684155225753784,
      "learning_rate": 5.0872620321864185e-05,
      "loss": 3.3316,
      "step": 109
    },
    {
      "epoch": 0.04087699739873653,
      "grad_norm": 1.5542587041854858,
      "learning_rate": 5e-05,
      "loss": 3.1965,
      "step": 110
    },
    {
      "epoch": 0.04124860646599777,
      "grad_norm": 1.5495750904083252,
      "learning_rate": 4.912737967813583e-05,
      "loss": 3.2084,
      "step": 111
    },
    {
      "epoch": 0.04162021553325901,
      "grad_norm": 1.596696138381958,
      "learning_rate": 4.825502516487497e-05,
      "loss": 3.3498,
      "step": 112
    },
    {
      "epoch": 0.041991824600520256,
      "grad_norm": 1.546371579170227,
      "learning_rate": 4.738320218785281e-05,
      "loss": 3.3169,
      "step": 113
    },
    {
      "epoch": 0.042363433667781496,
      "grad_norm": 1.6366443634033203,
      "learning_rate": 4.6512176312793736e-05,
      "loss": 3.1769,
      "step": 114
    },
    {
      "epoch": 0.042735042735042736,
      "grad_norm": 1.6571626663208008,
      "learning_rate": 4.564221286261709e-05,
      "loss": 3.1964,
      "step": 115
    },
    {
      "epoch": 0.043106651802303976,
      "grad_norm": 1.7170463800430298,
      "learning_rate": 4.477357683661734e-05,
      "loss": 3.2483,
      "step": 116
    },
    {
      "epoch": 0.043478260869565216,
      "grad_norm": 1.6605336666107178,
      "learning_rate": 4.390653282974264e-05,
      "loss": 3.183,
      "step": 117
    },
    {
      "epoch": 0.043849869936826456,
      "grad_norm": 1.6899641752243042,
      "learning_rate": 4.3041344951996746e-05,
      "loss": 3.1072,
      "step": 118
    },
    {
      "epoch": 0.044221479004087696,
      "grad_norm": 1.6592519283294678,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 3.1128,
      "step": 119
    },
    {
      "epoch": 0.044593088071348944,
      "grad_norm": 1.7209159135818481,
      "learning_rate": 4.131759111665349e-05,
      "loss": 3.1762,
      "step": 120
    },
    {
      "epoch": 0.044964697138610184,
      "grad_norm": 1.8177850246429443,
      "learning_rate": 4.045955023117276e-05,
      "loss": 3.2915,
      "step": 121
    },
    {
      "epoch": 0.045336306205871424,
      "grad_norm": 1.829469084739685,
      "learning_rate": 3.960441545911204e-05,
      "loss": 3.1219,
      "step": 122
    },
    {
      "epoch": 0.045707915273132664,
      "grad_norm": 1.9255375862121582,
      "learning_rate": 3.875244728280676e-05,
      "loss": 3.2112,
      "step": 123
    },
    {
      "epoch": 0.046079524340393904,
      "grad_norm": 1.9076225757598877,
      "learning_rate": 3.790390522001662e-05,
      "loss": 3.1843,
      "step": 124
    },
    {
      "epoch": 0.046451133407655144,
      "grad_norm": 1.8772251605987549,
      "learning_rate": 3.705904774487396e-05,
      "loss": 3.295,
      "step": 125
    },
    {
      "epoch": 0.046822742474916385,
      "grad_norm": 1.8381478786468506,
      "learning_rate": 3.6218132209150045e-05,
      "loss": 3.2535,
      "step": 126
    },
    {
      "epoch": 0.04719435154217763,
      "grad_norm": 1.9620457887649536,
      "learning_rate": 3.5381414763863166e-05,
      "loss": 3.1093,
      "step": 127
    },
    {
      "epoch": 0.04756596060943887,
      "grad_norm": 1.8187255859375,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.0412,
      "step": 128
    },
    {
      "epoch": 0.04793756967670011,
      "grad_norm": 2.0452768802642822,
      "learning_rate": 3.372159227714218e-05,
      "loss": 3.1595,
      "step": 129
    },
    {
      "epoch": 0.04830917874396135,
      "grad_norm": 2.0695669651031494,
      "learning_rate": 3.289899283371657e-05,
      "loss": 3.0353,
      "step": 130
    },
    {
      "epoch": 0.04868078781122259,
      "grad_norm": 1.918813705444336,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 3.0169,
      "step": 131
    },
    {
      "epoch": 0.04905239687848383,
      "grad_norm": 2.064602851867676,
      "learning_rate": 3.12696703292044e-05,
      "loss": 3.0316,
      "step": 132
    },
    {
      "epoch": 0.04942400594574508,
      "grad_norm": 2.167141914367676,
      "learning_rate": 3.046344357553632e-05,
      "loss": 3.244,
      "step": 133
    },
    {
      "epoch": 0.04979561501300632,
      "grad_norm": 2.1696243286132812,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 3.11,
      "step": 134
    },
    {
      "epoch": 0.05016722408026756,
      "grad_norm": 2.138988971710205,
      "learning_rate": 2.886908691296504e-05,
      "loss": 3.049,
      "step": 135
    },
    {
      "epoch": 0.0505388331475288,
      "grad_norm": 2.2165169715881348,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 3.0517,
      "step": 136
    },
    {
      "epoch": 0.05091044221479004,
      "grad_norm": 2.2241313457489014,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 2.9602,
      "step": 137
    },
    {
      "epoch": 0.05128205128205128,
      "grad_norm": 2.334522247314453,
      "learning_rate": 2.6526421860705473e-05,
      "loss": 3.1206,
      "step": 138
    },
    {
      "epoch": 0.05165366034931252,
      "grad_norm": 2.241197109222412,
      "learning_rate": 2.575951898768315e-05,
      "loss": 2.9884,
      "step": 139
    },
    {
      "epoch": 0.05202526941657377,
      "grad_norm": 2.4286975860595703,
      "learning_rate": 2.500000000000001e-05,
      "loss": 2.8666,
      "step": 140
    },
    {
      "epoch": 0.05239687848383501,
      "grad_norm": 2.472275495529175,
      "learning_rate": 2.4248096254497288e-05,
      "loss": 2.9763,
      "step": 141
    },
    {
      "epoch": 0.05276848755109625,
      "grad_norm": 2.541703224182129,
      "learning_rate": 2.350403678833976e-05,
      "loss": 2.8725,
      "step": 142
    },
    {
      "epoch": 0.05314009661835749,
      "grad_norm": 2.5989933013916016,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 2.7639,
      "step": 143
    },
    {
      "epoch": 0.05351170568561873,
      "grad_norm": 2.4298439025878906,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 2.8823,
      "step": 144
    },
    {
      "epoch": 0.05388331475287997,
      "grad_norm": 2.7883102893829346,
      "learning_rate": 2.132117818244771e-05,
      "loss": 2.9756,
      "step": 145
    },
    {
      "epoch": 0.05425492382014121,
      "grad_norm": 3.1214065551757812,
      "learning_rate": 2.061073738537635e-05,
      "loss": 3.1575,
      "step": 146
    },
    {
      "epoch": 0.054626532887402456,
      "grad_norm": 2.6867949962615967,
      "learning_rate": 1.9909248842397584e-05,
      "loss": 3.0654,
      "step": 147
    },
    {
      "epoch": 0.054998141954663696,
      "grad_norm": 2.903306722640991,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 3.3354,
      "step": 148
    },
    {
      "epoch": 0.055369751021924936,
      "grad_norm": 3.0160841941833496,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 3.333,
      "step": 149
    },
    {
      "epoch": 0.055741360089186176,
      "grad_norm": 3.1288528442382812,
      "learning_rate": 1.7860619515673033e-05,
      "loss": 3.5297,
      "step": 150
    },
    {
      "epoch": 0.055741360089186176,
      "eval_loss": 3.157715320587158,
      "eval_runtime": 109.7254,
      "eval_samples_per_second": 41.303,
      "eval_steps_per_second": 20.652,
      "step": 150
    },
    {
      "epoch": 0.056112969156447416,
      "grad_norm": 1.3505332469940186,
      "learning_rate": 1.7197048550474643e-05,
      "loss": 3.2944,
      "step": 151
    },
    {
      "epoch": 0.056484578223708656,
      "grad_norm": 1.6176457405090332,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 3.5568,
      "step": 152
    },
    {
      "epoch": 0.056856187290969896,
      "grad_norm": 1.506137728691101,
      "learning_rate": 1.5900081996875083e-05,
      "loss": 3.3174,
      "step": 153
    },
    {
      "epoch": 0.057227796358231144,
      "grad_norm": 1.5761620998382568,
      "learning_rate": 1.526708147705013e-05,
      "loss": 3.1891,
      "step": 154
    },
    {
      "epoch": 0.057599405425492384,
      "grad_norm": 1.5188205242156982,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 3.2784,
      "step": 155
    },
    {
      "epoch": 0.057971014492753624,
      "grad_norm": 1.4433867931365967,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 3.3127,
      "step": 156
    },
    {
      "epoch": 0.058342623560014864,
      "grad_norm": 1.559830904006958,
      "learning_rate": 1.3432314919041478e-05,
      "loss": 3.3048,
      "step": 157
    },
    {
      "epoch": 0.058714232627276104,
      "grad_norm": 1.5355303287506104,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 3.3979,
      "step": 158
    },
    {
      "epoch": 0.059085841694537344,
      "grad_norm": 1.553816795349121,
      "learning_rate": 1.22645209888614e-05,
      "loss": 3.4218,
      "step": 159
    },
    {
      "epoch": 0.059457450761798585,
      "grad_norm": 1.5942816734313965,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 3.2902,
      "step": 160
    },
    {
      "epoch": 0.05982905982905983,
      "grad_norm": 1.5093539953231812,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 3.0813,
      "step": 161
    },
    {
      "epoch": 0.06020066889632107,
      "grad_norm": 1.5273271799087524,
      "learning_rate": 1.0599462319663905e-05,
      "loss": 3.1854,
      "step": 162
    },
    {
      "epoch": 0.06057227796358231,
      "grad_norm": 1.5481042861938477,
      "learning_rate": 1.006822449763537e-05,
      "loss": 3.1412,
      "step": 163
    },
    {
      "epoch": 0.06094388703084355,
      "grad_norm": 1.5044341087341309,
      "learning_rate": 9.549150281252633e-06,
      "loss": 3.1521,
      "step": 164
    },
    {
      "epoch": 0.06131549609810479,
      "grad_norm": 1.755643606185913,
      "learning_rate": 9.042397785550405e-06,
      "loss": 3.1965,
      "step": 165
    },
    {
      "epoch": 0.06168710516536603,
      "grad_norm": 1.5893656015396118,
      "learning_rate": 8.548121372247918e-06,
      "loss": 3.2033,
      "step": 166
    },
    {
      "epoch": 0.06205871423262728,
      "grad_norm": 1.6347731351852417,
      "learning_rate": 8.066471602728803e-06,
      "loss": 3.2256,
      "step": 167
    },
    {
      "epoch": 0.06243032329988852,
      "grad_norm": 1.612523078918457,
      "learning_rate": 7.597595192178702e-06,
      "loss": 3.0524,
      "step": 168
    },
    {
      "epoch": 0.06280193236714976,
      "grad_norm": 1.6514352560043335,
      "learning_rate": 7.1416349648943894e-06,
      "loss": 3.2676,
      "step": 169
    },
    {
      "epoch": 0.06317354143441099,
      "grad_norm": 1.7786526679992676,
      "learning_rate": 6.698729810778065e-06,
      "loss": 3.0361,
      "step": 170
    },
    {
      "epoch": 0.06354515050167224,
      "grad_norm": 1.7342578172683716,
      "learning_rate": 6.269014643030213e-06,
      "loss": 3.2377,
      "step": 171
    },
    {
      "epoch": 0.06391675956893349,
      "grad_norm": 1.8072782754898071,
      "learning_rate": 5.852620357053651e-06,
      "loss": 3.2941,
      "step": 172
    },
    {
      "epoch": 0.06428836863619472,
      "grad_norm": 1.8157482147216797,
      "learning_rate": 5.449673790581611e-06,
      "loss": 3.3946,
      "step": 173
    },
    {
      "epoch": 0.06465997770345597,
      "grad_norm": 1.8105980157852173,
      "learning_rate": 5.060297685041659e-06,
      "loss": 3.2999,
      "step": 174
    },
    {
      "epoch": 0.0650315867707172,
      "grad_norm": 1.8060694932937622,
      "learning_rate": 4.684610648167503e-06,
      "loss": 3.1024,
      "step": 175
    },
    {
      "epoch": 0.06540319583797845,
      "grad_norm": 1.9588212966918945,
      "learning_rate": 4.322727117869951e-06,
      "loss": 3.2093,
      "step": 176
    },
    {
      "epoch": 0.06577480490523968,
      "grad_norm": 1.8691589832305908,
      "learning_rate": 3.974757327377981e-06,
      "loss": 3.1458,
      "step": 177
    },
    {
      "epoch": 0.06614641397250093,
      "grad_norm": 1.8959709405899048,
      "learning_rate": 3.6408072716606346e-06,
      "loss": 2.9414,
      "step": 178
    },
    {
      "epoch": 0.06651802303976218,
      "grad_norm": 2.009995460510254,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 3.1847,
      "step": 179
    },
    {
      "epoch": 0.06688963210702341,
      "grad_norm": 2.099750518798828,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 3.2094,
      "step": 180
    },
    {
      "epoch": 0.06726124117428466,
      "grad_norm": 1.9630095958709717,
      "learning_rate": 2.724071220034158e-06,
      "loss": 2.9846,
      "step": 181
    },
    {
      "epoch": 0.06763285024154589,
      "grad_norm": 2.1061551570892334,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.9945,
      "step": 182
    },
    {
      "epoch": 0.06800445930880714,
      "grad_norm": 2.263111114501953,
      "learning_rate": 2.1847622018482283e-06,
      "loss": 3.0124,
      "step": 183
    },
    {
      "epoch": 0.06837606837606838,
      "grad_norm": 2.2543697357177734,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 3.0128,
      "step": 184
    },
    {
      "epoch": 0.06874767744332962,
      "grad_norm": 2.3193070888519287,
      "learning_rate": 1.70370868554659e-06,
      "loss": 3.1496,
      "step": 185
    },
    {
      "epoch": 0.06911928651059086,
      "grad_norm": 2.193545341491699,
      "learning_rate": 1.4852136862001764e-06,
      "loss": 2.7909,
      "step": 186
    },
    {
      "epoch": 0.0694908955778521,
      "grad_norm": 2.370417833328247,
      "learning_rate": 1.2814967607382432e-06,
      "loss": 3.0471,
      "step": 187
    },
    {
      "epoch": 0.06986250464511334,
      "grad_norm": 2.1035029888153076,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 2.9547,
      "step": 188
    },
    {
      "epoch": 0.07023411371237458,
      "grad_norm": 2.3313772678375244,
      "learning_rate": 9.186408276168013e-07,
      "loss": 2.9233,
      "step": 189
    },
    {
      "epoch": 0.07060572277963582,
      "grad_norm": 2.323155641555786,
      "learning_rate": 7.596123493895991e-07,
      "loss": 3.0269,
      "step": 190
    },
    {
      "epoch": 0.07097733184689707,
      "grad_norm": 2.430950164794922,
      "learning_rate": 6.15582970243117e-07,
      "loss": 2.9647,
      "step": 191
    },
    {
      "epoch": 0.0713489409141583,
      "grad_norm": 2.574784994125366,
      "learning_rate": 4.865965629214819e-07,
      "loss": 3.0377,
      "step": 192
    },
    {
      "epoch": 0.07172054998141955,
      "grad_norm": 2.6166577339172363,
      "learning_rate": 3.7269241793390085e-07,
      "loss": 2.9381,
      "step": 193
    },
    {
      "epoch": 0.07209215904868078,
      "grad_norm": 2.817121744155884,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 2.8891,
      "step": 194
    },
    {
      "epoch": 0.07246376811594203,
      "grad_norm": 2.983161687850952,
      "learning_rate": 1.9026509541272275e-07,
      "loss": 2.9957,
      "step": 195
    },
    {
      "epoch": 0.07283537718320326,
      "grad_norm": 2.733790159225464,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 2.8709,
      "step": 196
    },
    {
      "epoch": 0.07320698625046451,
      "grad_norm": 2.8372256755828857,
      "learning_rate": 6.852326227130834e-08,
      "loss": 2.8952,
      "step": 197
    },
    {
      "epoch": 0.07357859531772576,
      "grad_norm": 2.988194227218628,
      "learning_rate": 3.04586490452119e-08,
      "loss": 3.3513,
      "step": 198
    },
    {
      "epoch": 0.07395020438498699,
      "grad_norm": 3.270862102508545,
      "learning_rate": 7.615242180436522e-09,
      "loss": 3.3235,
      "step": 199
    },
    {
      "epoch": 0.07432181345224824,
      "grad_norm": 3.2458367347717285,
      "learning_rate": 0.0,
      "loss": 3.458,
      "step": 200
    },
    {
      "epoch": 0.07432181345224824,
      "eval_loss": 3.1328935623168945,
      "eval_runtime": 109.4292,
      "eval_samples_per_second": 41.415,
      "eval_steps_per_second": 20.707,
      "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": 1.5495607275749376e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}