{
  "best_metric": 0.15735851228237152,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.5344827586206895,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007662835249042145,
      "grad_norm": 0.8423433303833008,
      "learning_rate": 6e-06,
      "loss": 0.3606,
      "step": 1
    },
    {
      "epoch": 0.007662835249042145,
      "eval_loss": 0.3226861357688904,
      "eval_runtime": 3.4747,
      "eval_samples_per_second": 63.316,
      "eval_steps_per_second": 15.829,
      "step": 1
    },
    {
      "epoch": 0.01532567049808429,
      "grad_norm": 0.735463559627533,
      "learning_rate": 1.2e-05,
      "loss": 0.4546,
      "step": 2
    },
    {
      "epoch": 0.022988505747126436,
      "grad_norm": 0.8963501453399658,
      "learning_rate": 1.8e-05,
      "loss": 0.4311,
      "step": 3
    },
    {
      "epoch": 0.03065134099616858,
      "grad_norm": 0.7320389151573181,
      "learning_rate": 2.4e-05,
      "loss": 0.3043,
      "step": 4
    },
    {
      "epoch": 0.038314176245210725,
      "grad_norm": 0.7728171348571777,
      "learning_rate": 3e-05,
      "loss": 0.4091,
      "step": 5
    },
    {
      "epoch": 0.04597701149425287,
      "grad_norm": 0.5618572235107422,
      "learning_rate": 3.6e-05,
      "loss": 0.2989,
      "step": 6
    },
    {
      "epoch": 0.05363984674329502,
      "grad_norm": 0.4079539477825165,
      "learning_rate": 4.2e-05,
      "loss": 0.3102,
      "step": 7
    },
    {
      "epoch": 0.06130268199233716,
      "grad_norm": 0.2701549232006073,
      "learning_rate": 4.8e-05,
      "loss": 0.2142,
      "step": 8
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 0.2180781215429306,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 0.2481,
      "step": 9
    },
    {
      "epoch": 0.07662835249042145,
      "grad_norm": 0.21011799573898315,
      "learning_rate": 6e-05,
      "loss": 0.1748,
      "step": 10
    },
    {
      "epoch": 0.0842911877394636,
      "grad_norm": 0.17047284543514252,
      "learning_rate": 5.999589914977407e-05,
      "loss": 0.1497,
      "step": 11
    },
    {
      "epoch": 0.09195402298850575,
      "grad_norm": 0.16233938932418823,
      "learning_rate": 5.998359772022778e-05,
      "loss": 0.186,
      "step": 12
    },
    {
      "epoch": 0.09961685823754789,
      "grad_norm": 0.171164870262146,
      "learning_rate": 5.996309907444915e-05,
      "loss": 0.1978,
      "step": 13
    },
    {
      "epoch": 0.10727969348659004,
      "grad_norm": 0.1632058173418045,
      "learning_rate": 5.9934408816563236e-05,
      "loss": 0.2963,
      "step": 14
    },
    {
      "epoch": 0.11494252873563218,
      "grad_norm": 0.18653063476085663,
      "learning_rate": 5.98975347902001e-05,
      "loss": 0.1962,
      "step": 15
    },
    {
      "epoch": 0.12260536398467432,
      "grad_norm": 0.16268739104270935,
      "learning_rate": 5.9852487076350345e-05,
      "loss": 0.1808,
      "step": 16
    },
    {
      "epoch": 0.13026819923371646,
      "grad_norm": 0.17408287525177002,
      "learning_rate": 5.979927799060915e-05,
      "loss": 0.1807,
      "step": 17
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 0.16513501107692719,
      "learning_rate": 5.9737922079809257e-05,
      "loss": 0.1695,
      "step": 18
    },
    {
      "epoch": 0.14559386973180077,
      "grad_norm": 0.1651427000761032,
      "learning_rate": 5.9668436118044054e-05,
      "loss": 0.1695,
      "step": 19
    },
    {
      "epoch": 0.1532567049808429,
      "grad_norm": 0.1825142204761505,
      "learning_rate": 5.959083910208167e-05,
      "loss": 0.1944,
      "step": 20
    },
    {
      "epoch": 0.16091954022988506,
      "grad_norm": 0.1646604835987091,
      "learning_rate": 5.9505152246171474e-05,
      "loss": 0.1823,
      "step": 21
    },
    {
      "epoch": 0.1685823754789272,
      "grad_norm": 0.18354526162147522,
      "learning_rate": 5.941139897624428e-05,
      "loss": 0.1919,
      "step": 22
    },
    {
      "epoch": 0.17624521072796934,
      "grad_norm": 0.1610330045223236,
      "learning_rate": 5.9309604923507984e-05,
      "loss": 0.2239,
      "step": 23
    },
    {
      "epoch": 0.1839080459770115,
      "grad_norm": 0.16655805706977844,
      "learning_rate": 5.9199797917440176e-05,
      "loss": 0.181,
      "step": 24
    },
    {
      "epoch": 0.19157088122605365,
      "grad_norm": 0.17879553139209747,
      "learning_rate": 5.908200797817991e-05,
      "loss": 0.2252,
      "step": 25
    },
    {
      "epoch": 0.19923371647509577,
      "grad_norm": 0.20575474202632904,
      "learning_rate": 5.895626730832046e-05,
      "loss": 0.2242,
      "step": 26
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 0.20907919108867645,
      "learning_rate": 5.882261028410545e-05,
      "loss": 0.2382,
      "step": 27
    },
    {
      "epoch": 0.21455938697318008,
      "grad_norm": 0.23485180735588074,
      "learning_rate": 5.8681073446030734e-05,
      "loss": 0.2638,
      "step": 28
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.23030665516853333,
      "learning_rate": 5.853169548885461e-05,
      "loss": 0.2498,
      "step": 29
    },
    {
      "epoch": 0.22988505747126436,
      "grad_norm": 0.26557809114456177,
      "learning_rate": 5.8374517251019035e-05,
      "loss": 0.2819,
      "step": 30
    },
    {
      "epoch": 0.23754789272030652,
      "grad_norm": 0.32848531007766724,
      "learning_rate": 5.820958170348484e-05,
      "loss": 0.3778,
      "step": 31
    },
    {
      "epoch": 0.24521072796934865,
      "grad_norm": 0.39857420325279236,
      "learning_rate": 5.8036933937983825e-05,
      "loss": 0.3802,
      "step": 32
    },
    {
      "epoch": 0.25287356321839083,
      "grad_norm": 0.24353501200675964,
      "learning_rate": 5.7856621154691217e-05,
      "loss": 0.3193,
      "step": 33
    },
    {
      "epoch": 0.26053639846743293,
      "grad_norm": 0.2523253858089447,
      "learning_rate": 5.766869264932154e-05,
      "loss": 0.3455,
      "step": 34
    },
    {
      "epoch": 0.2681992337164751,
      "grad_norm": 0.2798840403556824,
      "learning_rate": 5.747319979965172e-05,
      "loss": 0.362,
      "step": 35
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 0.22852621972560883,
      "learning_rate": 5.727019605147488e-05,
      "loss": 0.246,
      "step": 36
    },
    {
      "epoch": 0.2835249042145594,
      "grad_norm": 0.194475457072258,
      "learning_rate": 5.7059736903988775e-05,
      "loss": 0.2618,
      "step": 37
    },
    {
      "epoch": 0.29118773946360155,
      "grad_norm": 0.22199846804141998,
      "learning_rate": 5.684187989462291e-05,
      "loss": 0.3181,
      "step": 38
    },
    {
      "epoch": 0.2988505747126437,
      "grad_norm": 0.218927264213562,
      "learning_rate": 5.661668458330836e-05,
      "loss": 0.2443,
      "step": 39
    },
    {
      "epoch": 0.3065134099616858,
      "grad_norm": 0.297415167093277,
      "learning_rate": 5.638421253619467e-05,
      "loss": 0.2155,
      "step": 40
    },
    {
      "epoch": 0.31417624521072796,
      "grad_norm": 0.22468075156211853,
      "learning_rate": 5.614452730881832e-05,
      "loss": 0.1881,
      "step": 41
    },
    {
      "epoch": 0.3218390804597701,
      "grad_norm": 0.25146886706352234,
      "learning_rate": 5.589769442872722e-05,
      "loss": 0.1693,
      "step": 42
    },
    {
      "epoch": 0.32950191570881227,
      "grad_norm": 0.1930306851863861,
      "learning_rate": 5.5643781377566175e-05,
      "loss": 0.1175,
      "step": 43
    },
    {
      "epoch": 0.3371647509578544,
      "grad_norm": 0.14655643701553345,
      "learning_rate": 5.538285757262806e-05,
      "loss": 0.1227,
      "step": 44
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 0.1660422831773758,
      "learning_rate": 5.5114994347875856e-05,
      "loss": 0.121,
      "step": 45
    },
    {
      "epoch": 0.3524904214559387,
      "grad_norm": 0.1438768059015274,
      "learning_rate": 5.48402649344406e-05,
      "loss": 0.1305,
      "step": 46
    },
    {
      "epoch": 0.36015325670498083,
      "grad_norm": 0.1403757929801941,
      "learning_rate": 5.455874444060078e-05,
      "loss": 0.1166,
      "step": 47
    },
    {
      "epoch": 0.367816091954023,
      "grad_norm": 0.12318344414234161,
      "learning_rate": 5.427050983124843e-05,
      "loss": 0.1231,
      "step": 48
    },
    {
      "epoch": 0.37547892720306514,
      "grad_norm": 0.12252578139305115,
      "learning_rate": 5.397563990684774e-05,
      "loss": 0.1241,
      "step": 49
    },
    {
      "epoch": 0.3831417624521073,
      "grad_norm": 0.11253486573696136,
      "learning_rate": 5.367421528189181e-05,
      "loss": 0.1376,
      "step": 50
    },
    {
      "epoch": 0.3831417624521073,
      "eval_loss": 0.19786472618579865,
      "eval_runtime": 3.4047,
      "eval_samples_per_second": 64.616,
      "eval_steps_per_second": 16.154,
      "step": 50
    },
    {
      "epoch": 0.39080459770114945,
      "grad_norm": 0.12123022228479385,
      "learning_rate": 5.336631836286338e-05,
      "loss": 0.2249,
      "step": 51
    },
    {
      "epoch": 0.39846743295019155,
      "grad_norm": 0.10601665079593658,
      "learning_rate": 5.3052033325705774e-05,
      "loss": 0.1189,
      "step": 52
    },
    {
      "epoch": 0.4061302681992337,
      "grad_norm": 0.10251379013061523,
      "learning_rate": 5.2731446092810044e-05,
      "loss": 0.1377,
      "step": 53
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 0.1318877786397934,
      "learning_rate": 5.240464430952462e-05,
      "loss": 0.1783,
      "step": 54
    },
    {
      "epoch": 0.421455938697318,
      "grad_norm": 0.12670627236366272,
      "learning_rate": 5.207171732019395e-05,
      "loss": 0.1428,
      "step": 55
    },
    {
      "epoch": 0.42911877394636017,
      "grad_norm": 0.11889408528804779,
      "learning_rate": 5.1732756143732675e-05,
      "loss": 0.1258,
      "step": 56
    },
    {
      "epoch": 0.4367816091954023,
      "grad_norm": 0.1430068165063858,
      "learning_rate": 5.1387853448741916e-05,
      "loss": 0.3064,
      "step": 57
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.17051441967487335,
      "learning_rate": 5.103710352817465e-05,
      "loss": 0.1519,
      "step": 58
    },
    {
      "epoch": 0.4521072796934866,
      "grad_norm": 0.144890695810318,
      "learning_rate": 5.068060227355698e-05,
      "loss": 0.1364,
      "step": 59
    },
    {
      "epoch": 0.45977011494252873,
      "grad_norm": 0.15416496992111206,
      "learning_rate": 5.0318447148772234e-05,
      "loss": 0.2479,
      "step": 60
    },
    {
      "epoch": 0.4674329501915709,
      "grad_norm": 0.1834353357553482,
      "learning_rate": 4.995073716341545e-05,
      "loss": 0.1766,
      "step": 61
    },
    {
      "epoch": 0.47509578544061304,
      "grad_norm": 0.22412484884262085,
      "learning_rate": 4.957757284572506e-05,
      "loss": 0.2471,
      "step": 62
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 0.23686091601848602,
      "learning_rate": 4.91990562150995e-05,
      "loss": 0.2362,
      "step": 63
    },
    {
      "epoch": 0.4904214559386973,
      "grad_norm": 0.36825889348983765,
      "learning_rate": 4.881529075420611e-05,
      "loss": 0.3376,
      "step": 64
    },
    {
      "epoch": 0.49808429118773945,
      "grad_norm": 0.31192702054977417,
      "learning_rate": 4.8426381380690036e-05,
      "loss": 0.3341,
      "step": 65
    },
    {
      "epoch": 0.5057471264367817,
      "grad_norm": 0.26902249455451965,
      "learning_rate": 4.8032434418490753e-05,
      "loss": 0.3692,
      "step": 66
    },
    {
      "epoch": 0.5134099616858238,
      "grad_norm": 0.2696472406387329,
      "learning_rate": 4.7633557568774194e-05,
      "loss": 0.3385,
      "step": 67
    },
    {
      "epoch": 0.5210727969348659,
      "grad_norm": 0.26114562153816223,
      "learning_rate": 4.722985988048831e-05,
      "loss": 0.3358,
      "step": 68
    },
    {
      "epoch": 0.5287356321839081,
      "grad_norm": 0.2191227674484253,
      "learning_rate": 4.6821451720550184e-05,
      "loss": 0.3419,
      "step": 69
    },
    {
      "epoch": 0.5363984674329502,
      "grad_norm": 0.1417953372001648,
      "learning_rate": 4.640844474367282e-05,
      "loss": 0.3049,
      "step": 70
    },
    {
      "epoch": 0.5440613026819924,
      "grad_norm": 0.08716930449008942,
      "learning_rate": 4.5990951861839815e-05,
      "loss": 0.2303,
      "step": 71
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.10233739018440247,
      "learning_rate": 4.5569087213436455e-05,
      "loss": 0.2153,
      "step": 72
    },
    {
      "epoch": 0.5593869731800766,
      "grad_norm": 0.11295162886381149,
      "learning_rate": 4.514296613204532e-05,
      "loss": 0.1065,
      "step": 73
    },
    {
      "epoch": 0.5670498084291188,
      "grad_norm": 0.15026111900806427,
      "learning_rate": 4.471270511491525e-05,
      "loss": 0.1309,
      "step": 74
    },
    {
      "epoch": 0.5747126436781609,
      "grad_norm": 0.1282494068145752,
      "learning_rate": 4.427842179111221e-05,
      "loss": 0.1219,
      "step": 75
    },
    {
      "epoch": 0.5823754789272031,
      "grad_norm": 0.12559747695922852,
      "learning_rate": 4.3840234889360634e-05,
      "loss": 0.1231,
      "step": 76
    },
    {
      "epoch": 0.5900383141762452,
      "grad_norm": 0.13436265289783478,
      "learning_rate": 4.33982642055842e-05,
      "loss": 0.3163,
      "step": 77
    },
    {
      "epoch": 0.5977011494252874,
      "grad_norm": 0.10354889184236526,
      "learning_rate": 4.2952630570154785e-05,
      "loss": 0.1332,
      "step": 78
    },
    {
      "epoch": 0.6053639846743295,
      "grad_norm": 0.12167490273714066,
      "learning_rate": 4.250345581485871e-05,
      "loss": 0.0873,
      "step": 79
    },
    {
      "epoch": 0.6130268199233716,
      "grad_norm": 0.10158289968967438,
      "learning_rate": 4.205086273958909e-05,
      "loss": 0.0921,
      "step": 80
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 0.125481516122818,
      "learning_rate": 4.1594975078773565e-05,
      "loss": 0.0975,
      "step": 81
    },
    {
      "epoch": 0.6283524904214559,
      "grad_norm": 0.09864188730716705,
      "learning_rate": 4.113591746754662e-05,
      "loss": 0.1043,
      "step": 82
    },
    {
      "epoch": 0.6360153256704981,
      "grad_norm": 0.10632136464118958,
      "learning_rate": 4.06738154076755e-05,
      "loss": 0.0891,
      "step": 83
    },
    {
      "epoch": 0.6436781609195402,
      "grad_norm": 0.09485295414924622,
      "learning_rate": 4.020879523324929e-05,
      "loss": 0.1128,
      "step": 84
    },
    {
      "epoch": 0.6513409961685823,
      "grad_norm": 0.10994688421487808,
      "learning_rate": 3.974098407614051e-05,
      "loss": 0.0962,
      "step": 85
    },
    {
      "epoch": 0.6590038314176245,
      "grad_norm": 0.10552027821540833,
      "learning_rate": 3.927050983124842e-05,
      "loss": 0.1235,
      "step": 86
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.09930335730314255,
      "learning_rate": 3.8797501121533946e-05,
      "loss": 0.0976,
      "step": 87
    },
    {
      "epoch": 0.6743295019157088,
      "grad_norm": 0.10263009369373322,
      "learning_rate": 3.832208726285534e-05,
      "loss": 0.1238,
      "step": 88
    },
    {
      "epoch": 0.6819923371647509,
      "grad_norm": 0.09369200468063354,
      "learning_rate": 3.784439822861459e-05,
      "loss": 0.1652,
      "step": 89
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 0.0915440246462822,
      "learning_rate": 3.7364564614223976e-05,
      "loss": 0.1186,
      "step": 90
    },
    {
      "epoch": 0.6973180076628352,
      "grad_norm": 0.10551782697439194,
      "learning_rate": 3.688271760140255e-05,
      "loss": 0.1385,
      "step": 91
    },
    {
      "epoch": 0.7049808429118773,
      "grad_norm": 0.1463235318660736,
      "learning_rate": 3.6398988922312406e-05,
      "loss": 0.1676,
      "step": 92
    },
    {
      "epoch": 0.7126436781609196,
      "grad_norm": 0.15321095287799835,
      "learning_rate": 3.591351082354441e-05,
      "loss": 0.1269,
      "step": 93
    },
    {
      "epoch": 0.7203065134099617,
      "grad_norm": 0.21555227041244507,
      "learning_rate": 3.54264160299633e-05,
      "loss": 0.1737,
      "step": 94
    },
    {
      "epoch": 0.7279693486590039,
      "grad_norm": 0.19832918047904968,
      "learning_rate": 3.493783770842202e-05,
      "loss": 0.1858,
      "step": 95
    },
    {
      "epoch": 0.735632183908046,
      "grad_norm": 0.33908650279045105,
      "learning_rate": 3.444790943135526e-05,
      "loss": 0.2544,
      "step": 96
    },
    {
      "epoch": 0.7432950191570882,
      "grad_norm": 0.13350608944892883,
      "learning_rate": 3.3956765140262074e-05,
      "loss": 0.3619,
      "step": 97
    },
    {
      "epoch": 0.7509578544061303,
      "grad_norm": 0.08063369244337082,
      "learning_rate": 3.346453910908759e-05,
      "loss": 0.2084,
      "step": 98
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 0.12420119345188141,
      "learning_rate": 3.297136590751389e-05,
      "loss": 0.3331,
      "step": 99
    },
    {
      "epoch": 0.7662835249042146,
      "grad_norm": 0.1258733719587326,
      "learning_rate": 3.247738036416998e-05,
      "loss": 0.4197,
      "step": 100
    },
    {
      "epoch": 0.7662835249042146,
      "eval_loss": 0.16816693544387817,
      "eval_runtime": 3.4087,
      "eval_samples_per_second": 64.541,
      "eval_steps_per_second": 16.135,
      "step": 100
    },
    {
      "epoch": 0.7739463601532567,
      "grad_norm": 0.14441469311714172,
      "learning_rate": 3.1982717529770985e-05,
      "loss": 0.4444,
      "step": 101
    },
    {
      "epoch": 0.7816091954022989,
      "grad_norm": 0.11869792640209198,
      "learning_rate": 3.148751264019667e-05,
      "loss": 0.313,
      "step": 102
    },
    {
      "epoch": 0.789272030651341,
      "grad_norm": 0.09966205805540085,
      "learning_rate": 3.099190107951924e-05,
      "loss": 0.2333,
      "step": 103
    },
    {
      "epoch": 0.7969348659003831,
      "grad_norm": 0.12245635688304901,
      "learning_rate": 3.049601834299076e-05,
      "loss": 0.1806,
      "step": 104
    },
    {
      "epoch": 0.8045977011494253,
      "grad_norm": 0.12711887061595917,
      "learning_rate": 3e-05,
      "loss": 0.2199,
      "step": 105
    },
    {
      "epoch": 0.8122605363984674,
      "grad_norm": 0.12456270307302475,
      "learning_rate": 2.9503981657009246e-05,
      "loss": 0.0923,
      "step": 106
    },
    {
      "epoch": 0.8199233716475096,
      "grad_norm": 0.11281975358724594,
      "learning_rate": 2.9008098920480752e-05,
      "loss": 0.1066,
      "step": 107
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 0.07618711143732071,
      "learning_rate": 2.851248735980333e-05,
      "loss": 0.074,
      "step": 108
    },
    {
      "epoch": 0.8352490421455939,
      "grad_norm": 0.10129279643297195,
      "learning_rate": 2.801728247022902e-05,
      "loss": 0.0754,
      "step": 109
    },
    {
      "epoch": 0.842911877394636,
      "grad_norm": 0.09363987296819687,
      "learning_rate": 2.7522619635830034e-05,
      "loss": 0.0853,
      "step": 110
    },
    {
      "epoch": 0.8505747126436781,
      "grad_norm": 0.08725081384181976,
      "learning_rate": 2.702863409248612e-05,
      "loss": 0.2004,
      "step": 111
    },
    {
      "epoch": 0.8582375478927203,
      "grad_norm": 0.0783943384885788,
      "learning_rate": 2.6535460890912416e-05,
      "loss": 0.1047,
      "step": 112
    },
    {
      "epoch": 0.8659003831417624,
      "grad_norm": 0.10503990948200226,
      "learning_rate": 2.604323485973793e-05,
      "loss": 0.1262,
      "step": 113
    },
    {
      "epoch": 0.8735632183908046,
      "grad_norm": 0.09686201065778732,
      "learning_rate": 2.555209056864474e-05,
      "loss": 0.0969,
      "step": 114
    },
    {
      "epoch": 0.8812260536398467,
      "grad_norm": 0.0980193242430687,
      "learning_rate": 2.5062162291577978e-05,
      "loss": 0.1034,
      "step": 115
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.08893204480409622,
      "learning_rate": 2.4573583970036712e-05,
      "loss": 0.0968,
      "step": 116
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 0.12043558061122894,
      "learning_rate": 2.4086489176455595e-05,
      "loss": 0.0911,
      "step": 117
    },
    {
      "epoch": 0.9042145593869731,
      "grad_norm": 0.09192109853029251,
      "learning_rate": 2.36010110776876e-05,
      "loss": 0.0937,
      "step": 118
    },
    {
      "epoch": 0.9118773946360154,
      "grad_norm": 0.08706682920455933,
      "learning_rate": 2.3117282398597456e-05,
      "loss": 0.1428,
      "step": 119
    },
    {
      "epoch": 0.9195402298850575,
      "grad_norm": 0.11525483429431915,
      "learning_rate": 2.263543538577603e-05,
      "loss": 0.1219,
      "step": 120
    },
    {
      "epoch": 0.9272030651340997,
      "grad_norm": 0.09489944577217102,
      "learning_rate": 2.215560177138541e-05,
      "loss": 0.1083,
      "step": 121
    },
    {
      "epoch": 0.9348659003831418,
      "grad_norm": 0.10396278649568558,
      "learning_rate": 2.167791273714467e-05,
      "loss": 0.1037,
      "step": 122
    },
    {
      "epoch": 0.9425287356321839,
      "grad_norm": 0.12295960634946823,
      "learning_rate": 2.1202498878466062e-05,
      "loss": 0.1395,
      "step": 123
    },
    {
      "epoch": 0.9501915708812261,
      "grad_norm": 0.15319755673408508,
      "learning_rate": 2.072949016875158e-05,
      "loss": 0.1663,
      "step": 124
    },
    {
      "epoch": 0.9578544061302682,
      "grad_norm": 0.1255790889263153,
      "learning_rate": 2.0259015923859498e-05,
      "loss": 0.1011,
      "step": 125
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 0.17556820809841156,
      "learning_rate": 1.979120476675071e-05,
      "loss": 0.2089,
      "step": 126
    },
    {
      "epoch": 0.9731800766283525,
      "grad_norm": 0.15263380110263824,
      "learning_rate": 1.9326184592324503e-05,
      "loss": 0.1327,
      "step": 127
    },
    {
      "epoch": 0.9808429118773946,
      "grad_norm": 0.29129284620285034,
      "learning_rate": 1.8864082532453373e-05,
      "loss": 0.2664,
      "step": 128
    },
    {
      "epoch": 0.9885057471264368,
      "grad_norm": 0.07602576911449432,
      "learning_rate": 1.840502492122644e-05,
      "loss": 0.1506,
      "step": 129
    },
    {
      "epoch": 0.9961685823754789,
      "grad_norm": 0.07331366091966629,
      "learning_rate": 1.7949137260410924e-05,
      "loss": 0.0866,
      "step": 130
    },
    {
      "epoch": 1.0057471264367817,
      "grad_norm": 0.3374364674091339,
      "learning_rate": 1.7496544185141295e-05,
      "loss": 0.4057,
      "step": 131
    },
    {
      "epoch": 1.0134099616858236,
      "grad_norm": 0.10590671747922897,
      "learning_rate": 1.7047369429845216e-05,
      "loss": 0.1654,
      "step": 132
    },
    {
      "epoch": 1.0210727969348659,
      "grad_norm": 0.12875299155712128,
      "learning_rate": 1.6601735794415806e-05,
      "loss": 0.2853,
      "step": 133
    },
    {
      "epoch": 1.028735632183908,
      "grad_norm": 0.13337695598602295,
      "learning_rate": 1.615976511063937e-05,
      "loss": 0.3602,
      "step": 134
    },
    {
      "epoch": 1.0363984674329503,
      "grad_norm": 0.175828754901886,
      "learning_rate": 1.5721578208887793e-05,
      "loss": 0.4249,
      "step": 135
    },
    {
      "epoch": 1.0440613026819923,
      "grad_norm": 0.1130770593881607,
      "learning_rate": 1.5287294885084766e-05,
      "loss": 0.2504,
      "step": 136
    },
    {
      "epoch": 1.0517241379310345,
      "grad_norm": 0.09660311043262482,
      "learning_rate": 1.4857033867954697e-05,
      "loss": 0.1961,
      "step": 137
    },
    {
      "epoch": 1.0593869731800767,
      "grad_norm": 0.0801113024353981,
      "learning_rate": 1.4430912786563554e-05,
      "loss": 0.1567,
      "step": 138
    },
    {
      "epoch": 1.0670498084291187,
      "grad_norm": 0.09412167966365814,
      "learning_rate": 1.4009048138160195e-05,
      "loss": 0.1346,
      "step": 139
    },
    {
      "epoch": 1.0747126436781609,
      "grad_norm": 0.09223540127277374,
      "learning_rate": 1.3591555256327199e-05,
      "loss": 0.0735,
      "step": 140
    },
    {
      "epoch": 1.082375478927203,
      "grad_norm": 0.06864771246910095,
      "learning_rate": 1.3178548279449822e-05,
      "loss": 0.0827,
      "step": 141
    },
    {
      "epoch": 1.0900383141762453,
      "grad_norm": 0.09543950110673904,
      "learning_rate": 1.2770140119511693e-05,
      "loss": 0.0824,
      "step": 142
    },
    {
      "epoch": 1.0977011494252873,
      "grad_norm": 0.07598932087421417,
      "learning_rate": 1.2366442431225809e-05,
      "loss": 0.0992,
      "step": 143
    },
    {
      "epoch": 1.1053639846743295,
      "grad_norm": 0.090644970536232,
      "learning_rate": 1.1967565581509248e-05,
      "loss": 0.0885,
      "step": 144
    },
    {
      "epoch": 1.1130268199233717,
      "grad_norm": 0.06884025782346725,
      "learning_rate": 1.1573618619309965e-05,
      "loss": 0.0793,
      "step": 145
    },
    {
      "epoch": 1.1206896551724137,
      "grad_norm": 0.09408024698495865,
      "learning_rate": 1.1184709245793889e-05,
      "loss": 0.1176,
      "step": 146
    },
    {
      "epoch": 1.128352490421456,
      "grad_norm": 0.07865190505981445,
      "learning_rate": 1.0800943784900502e-05,
      "loss": 0.1063,
      "step": 147
    },
    {
      "epoch": 1.1360153256704981,
      "grad_norm": 0.07905281335115433,
      "learning_rate": 1.042242715427494e-05,
      "loss": 0.1115,
      "step": 148
    },
    {
      "epoch": 1.1436781609195403,
      "grad_norm": 0.10434318333864212,
      "learning_rate": 1.004926283658455e-05,
      "loss": 0.0815,
      "step": 149
    },
    {
      "epoch": 1.1513409961685823,
      "grad_norm": 0.08591567724943161,
      "learning_rate": 9.681552851227774e-06,
      "loss": 0.0945,
      "step": 150
    },
    {
      "epoch": 1.1513409961685823,
      "eval_loss": 0.16181236505508423,
      "eval_runtime": 3.4216,
      "eval_samples_per_second": 64.297,
      "eval_steps_per_second": 16.074,
      "step": 150
    },
    {
      "epoch": 1.1590038314176245,
      "grad_norm": 0.09190652519464493,
      "learning_rate": 9.319397726443026e-06,
      "loss": 0.08,
      "step": 151
    },
    {
      "epoch": 1.1666666666666667,
      "grad_norm": 0.08560865372419357,
      "learning_rate": 8.962896471825342e-06,
      "loss": 0.1018,
      "step": 152
    },
    {
      "epoch": 1.1743295019157087,
      "grad_norm": 0.09749860316514969,
      "learning_rate": 8.61214655125809e-06,
      "loss": 0.0997,
      "step": 153
    },
    {
      "epoch": 1.181992337164751,
      "grad_norm": 0.0863884836435318,
      "learning_rate": 8.267243856267331e-06,
      "loss": 0.1043,
      "step": 154
    },
    {
      "epoch": 1.1896551724137931,
      "grad_norm": 0.09674271196126938,
      "learning_rate": 7.928282679806052e-06,
      "loss": 0.0869,
      "step": 155
    },
    {
      "epoch": 1.1973180076628354,
      "grad_norm": 0.09813065826892853,
      "learning_rate": 7.595355690475393e-06,
      "loss": 0.115,
      "step": 156
    },
    {
      "epoch": 1.2049808429118773,
      "grad_norm": 0.11123661696910858,
      "learning_rate": 7.268553907189964e-06,
      "loss": 0.1153,
      "step": 157
    },
    {
      "epoch": 1.2126436781609196,
      "grad_norm": 0.1099216416478157,
      "learning_rate": 6.947966674294236e-06,
      "loss": 0.1313,
      "step": 158
    },
    {
      "epoch": 1.2203065134099618,
      "grad_norm": 0.12104592472314835,
      "learning_rate": 6.6336816371366305e-06,
      "loss": 0.175,
      "step": 159
    },
    {
      "epoch": 1.2279693486590038,
      "grad_norm": 0.14585037529468536,
      "learning_rate": 6.325784718108196e-06,
      "loss": 0.1442,
      "step": 160
    },
    {
      "epoch": 1.235632183908046,
      "grad_norm": 0.162072092294693,
      "learning_rate": 6.0243600931522595e-06,
      "loss": 0.1529,
      "step": 161
    },
    {
      "epoch": 1.2432950191570882,
      "grad_norm": 0.22925086319446564,
      "learning_rate": 5.72949016875158e-06,
      "loss": 0.1984,
      "step": 162
    },
    {
      "epoch": 1.2509578544061304,
      "grad_norm": 0.13616731762886047,
      "learning_rate": 5.44125555939923e-06,
      "loss": 0.2254,
      "step": 163
    },
    {
      "epoch": 1.2586206896551724,
      "grad_norm": 0.14947763085365295,
      "learning_rate": 5.159735065559399e-06,
      "loss": 0.379,
      "step": 164
    },
    {
      "epoch": 1.2662835249042146,
      "grad_norm": 0.10282624512910843,
      "learning_rate": 4.885005652124144e-06,
      "loss": 0.3126,
      "step": 165
    },
    {
      "epoch": 1.2739463601532566,
      "grad_norm": 0.12892718613147736,
      "learning_rate": 4.617142427371934e-06,
      "loss": 0.4403,
      "step": 166
    },
    {
      "epoch": 1.2816091954022988,
      "grad_norm": 0.09580893069505692,
      "learning_rate": 4.3562186224338265e-06,
      "loss": 0.2804,
      "step": 167
    },
    {
      "epoch": 1.289272030651341,
      "grad_norm": 0.07744543999433517,
      "learning_rate": 4.102305571272783e-06,
      "loss": 0.2067,
      "step": 168
    },
    {
      "epoch": 1.2969348659003832,
      "grad_norm": 0.08755867183208466,
      "learning_rate": 3.855472691181678e-06,
      "loss": 0.2504,
      "step": 169
    },
    {
      "epoch": 1.3045977011494254,
      "grad_norm": 0.07829572260379791,
      "learning_rate": 3.615787463805331e-06,
      "loss": 0.1686,
      "step": 170
    },
    {
      "epoch": 1.3122605363984674,
      "grad_norm": 0.06435937434434891,
      "learning_rate": 3.383315416691646e-06,
      "loss": 0.1323,
      "step": 171
    },
    {
      "epoch": 1.3199233716475096,
      "grad_norm": 0.11443574726581573,
      "learning_rate": 3.158120105377096e-06,
      "loss": 0.1042,
      "step": 172
    },
    {
      "epoch": 1.3275862068965516,
      "grad_norm": 0.0844988003373146,
      "learning_rate": 2.940263096011233e-06,
      "loss": 0.1191,
      "step": 173
    },
    {
      "epoch": 1.3352490421455938,
      "grad_norm": 0.07520398497581482,
      "learning_rate": 2.729803948525125e-06,
      "loss": 0.0742,
      "step": 174
    },
    {
      "epoch": 1.342911877394636,
      "grad_norm": 0.08877892047166824,
      "learning_rate": 2.526800200348275e-06,
      "loss": 0.1036,
      "step": 175
    },
    {
      "epoch": 1.3505747126436782,
      "grad_norm": 0.09079942852258682,
      "learning_rate": 2.3313073506784575e-06,
      "loss": 0.1974,
      "step": 176
    },
    {
      "epoch": 1.3582375478927204,
      "grad_norm": 0.06422315537929535,
      "learning_rate": 2.143378845308791e-06,
      "loss": 0.0846,
      "step": 177
    },
    {
      "epoch": 1.3659003831417624,
      "grad_norm": 0.08929227292537689,
      "learning_rate": 1.9630660620161777e-06,
      "loss": 0.2096,
      "step": 178
    },
    {
      "epoch": 1.3735632183908046,
      "grad_norm": 0.0649232268333435,
      "learning_rate": 1.790418296515165e-06,
      "loss": 0.0711,
      "step": 179
    },
    {
      "epoch": 1.3812260536398466,
      "grad_norm": 0.06585374474525452,
      "learning_rate": 1.625482748980961e-06,
      "loss": 0.0756,
      "step": 180
    },
    {
      "epoch": 1.3888888888888888,
      "grad_norm": 0.08487678319215775,
      "learning_rate": 1.4683045111453942e-06,
      "loss": 0.1589,
      "step": 181
    },
    {
      "epoch": 1.396551724137931,
      "grad_norm": 0.08722589910030365,
      "learning_rate": 1.3189265539692707e-06,
      "loss": 0.0999,
      "step": 182
    },
    {
      "epoch": 1.4042145593869733,
      "grad_norm": 0.066057488322258,
      "learning_rate": 1.1773897158945557e-06,
      "loss": 0.0952,
      "step": 183
    },
    {
      "epoch": 1.4118773946360152,
      "grad_norm": 0.07314711809158325,
      "learning_rate": 1.0437326916795432e-06,
      "loss": 0.0918,
      "step": 184
    },
    {
      "epoch": 1.4195402298850575,
      "grad_norm": 0.08497738838195801,
      "learning_rate": 9.179920218200888e-07,
      "loss": 0.0882,
      "step": 185
    },
    {
      "epoch": 1.4272030651340997,
      "grad_norm": 0.08262482285499573,
      "learning_rate": 8.002020825598277e-07,
      "loss": 0.0782,
      "step": 186
    },
    {
      "epoch": 1.4348659003831417,
      "grad_norm": 0.10725460946559906,
      "learning_rate": 6.90395076492022e-07,
      "loss": 0.1653,
      "step": 187
    },
    {
      "epoch": 1.4425287356321839,
      "grad_norm": 0.0954778790473938,
      "learning_rate": 5.886010237557194e-07,
      "loss": 0.1031,
      "step": 188
    },
    {
      "epoch": 1.450191570881226,
      "grad_norm": 0.08982761204242706,
      "learning_rate": 4.94847753828529e-07,
      "loss": 0.0956,
      "step": 189
    },
    {
      "epoch": 1.4578544061302683,
      "grad_norm": 0.10045281052589417,
      "learning_rate": 4.091608979183303e-07,
      "loss": 0.1225,
      "step": 190
    },
    {
      "epoch": 1.4655172413793103,
      "grad_norm": 0.11967014521360397,
      "learning_rate": 3.315638819559452e-07,
      "loss": 0.1291,
      "step": 191
    },
    {
      "epoch": 1.4731800766283525,
      "grad_norm": 0.15871833264827728,
      "learning_rate": 2.6207792019074414e-07,
      "loss": 0.1136,
      "step": 192
    },
    {
      "epoch": 1.4808429118773947,
      "grad_norm": 0.15147604048252106,
      "learning_rate": 2.0072200939085573e-07,
      "loss": 0.1825,
      "step": 193
    },
    {
      "epoch": 1.4885057471264367,
      "grad_norm": 0.19062329828739166,
      "learning_rate": 1.475129236496575e-07,
      "loss": 0.176,
      "step": 194
    },
    {
      "epoch": 1.496168582375479,
      "grad_norm": 0.10360642522573471,
      "learning_rate": 1.0246520979990459e-07,
      "loss": 0.2787,
      "step": 195
    },
    {
      "epoch": 1.503831417624521,
      "grad_norm": 0.09995601326227188,
      "learning_rate": 6.559118343676396e-08,
      "loss": 0.2528,
      "step": 196
    },
    {
      "epoch": 1.5114942528735633,
      "grad_norm": 0.11199713498353958,
      "learning_rate": 3.690092555085789e-08,
      "loss": 0.2772,
      "step": 197
    },
    {
      "epoch": 1.5191570881226055,
      "grad_norm": 0.09361888468265533,
      "learning_rate": 1.640227977221853e-08,
      "loss": 0.2773,
      "step": 198
    },
    {
      "epoch": 1.5268199233716475,
      "grad_norm": 0.09406208992004395,
      "learning_rate": 4.1008502259298755e-09,
      "loss": 0.2662,
      "step": 199
    },
    {
      "epoch": 1.5344827586206895,
      "grad_norm": 0.10898585617542267,
      "learning_rate": 0.0,
      "loss": 0.3079,
      "step": 200
    },
    {
      "epoch": 1.5344827586206895,
      "eval_loss": 0.15735851228237152,
      "eval_runtime": 3.4239,
      "eval_samples_per_second": 64.254,
      "eval_steps_per_second": 16.064,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 4,
        "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": 7132227890577408.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}