{
  "best_metric": 1.1092175245285034,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3784295175023652,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001892147587511826,
      "grad_norm": 3.9962542057037354,
      "learning_rate": 1e-05,
      "loss": 3.1641,
      "step": 1
    },
    {
      "epoch": 0.001892147587511826,
      "eval_loss": 3.7552740573883057,
      "eval_runtime": 65.7317,
      "eval_samples_per_second": 13.54,
      "eval_steps_per_second": 3.393,
      "step": 1
    },
    {
      "epoch": 0.003784295175023652,
      "grad_norm": 4.148435115814209,
      "learning_rate": 2e-05,
      "loss": 3.2134,
      "step": 2
    },
    {
      "epoch": 0.005676442762535478,
      "grad_norm": 4.270897388458252,
      "learning_rate": 3e-05,
      "loss": 3.1615,
      "step": 3
    },
    {
      "epoch": 0.007568590350047304,
      "grad_norm": 4.7075700759887695,
      "learning_rate": 4e-05,
      "loss": 3.4663,
      "step": 4
    },
    {
      "epoch": 0.00946073793755913,
      "grad_norm": 4.862111568450928,
      "learning_rate": 5e-05,
      "loss": 3.3649,
      "step": 5
    },
    {
      "epoch": 0.011352885525070956,
      "grad_norm": 4.50938081741333,
      "learning_rate": 6e-05,
      "loss": 3.0731,
      "step": 6
    },
    {
      "epoch": 0.013245033112582781,
      "grad_norm": 4.179661273956299,
      "learning_rate": 7e-05,
      "loss": 2.9059,
      "step": 7
    },
    {
      "epoch": 0.015137180700094607,
      "grad_norm": 3.4024016857147217,
      "learning_rate": 8e-05,
      "loss": 2.4961,
      "step": 8
    },
    {
      "epoch": 0.017029328287606435,
      "grad_norm": 3.079073429107666,
      "learning_rate": 9e-05,
      "loss": 2.0441,
      "step": 9
    },
    {
      "epoch": 0.01892147587511826,
      "grad_norm": 4.025096416473389,
      "learning_rate": 0.0001,
      "loss": 2.079,
      "step": 10
    },
    {
      "epoch": 0.020813623462630087,
      "grad_norm": 4.364260673522949,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.2205,
      "step": 11
    },
    {
      "epoch": 0.02270577105014191,
      "grad_norm": 3.2194790840148926,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.7834,
      "step": 12
    },
    {
      "epoch": 0.02459791863765374,
      "grad_norm": 2.8641135692596436,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.6726,
      "step": 13
    },
    {
      "epoch": 0.026490066225165563,
      "grad_norm": 3.4023993015289307,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.8643,
      "step": 14
    },
    {
      "epoch": 0.02838221381267739,
      "grad_norm": 2.7386951446533203,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.8503,
      "step": 15
    },
    {
      "epoch": 0.030274361400189215,
      "grad_norm": 2.6271438598632812,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.4376,
      "step": 16
    },
    {
      "epoch": 0.03216650898770104,
      "grad_norm": 2.5256102085113525,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.6511,
      "step": 17
    },
    {
      "epoch": 0.03405865657521287,
      "grad_norm": 1.9166487455368042,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.5186,
      "step": 18
    },
    {
      "epoch": 0.03595080416272469,
      "grad_norm": 2.018867015838623,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.3928,
      "step": 19
    },
    {
      "epoch": 0.03784295175023652,
      "grad_norm": 2.063495635986328,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3574,
      "step": 20
    },
    {
      "epoch": 0.039735099337748346,
      "grad_norm": 2.194066047668457,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.3055,
      "step": 21
    },
    {
      "epoch": 0.041627246925260174,
      "grad_norm": 2.5645647048950195,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.6894,
      "step": 22
    },
    {
      "epoch": 0.043519394512771994,
      "grad_norm": 2.1246392726898193,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.407,
      "step": 23
    },
    {
      "epoch": 0.04541154210028382,
      "grad_norm": 4.374281406402588,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.6437,
      "step": 24
    },
    {
      "epoch": 0.04730368968779565,
      "grad_norm": 2.118626356124878,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.4098,
      "step": 25
    },
    {
      "epoch": 0.04919583727530748,
      "grad_norm": 2.3993191719055176,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5439,
      "step": 26
    },
    {
      "epoch": 0.0510879848628193,
      "grad_norm": 2.2297019958496094,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4976,
      "step": 27
    },
    {
      "epoch": 0.052980132450331126,
      "grad_norm": 2.101769208908081,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.5698,
      "step": 28
    },
    {
      "epoch": 0.05487228003784295,
      "grad_norm": 2.0445644855499268,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.4628,
      "step": 29
    },
    {
      "epoch": 0.05676442762535478,
      "grad_norm": 2.3136656284332275,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.5558,
      "step": 30
    },
    {
      "epoch": 0.0586565752128666,
      "grad_norm": 2.296520709991455,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4657,
      "step": 31
    },
    {
      "epoch": 0.06054872280037843,
      "grad_norm": 1.8998284339904785,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.4703,
      "step": 32
    },
    {
      "epoch": 0.06244087038789026,
      "grad_norm": 1.6950056552886963,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.4804,
      "step": 33
    },
    {
      "epoch": 0.06433301797540208,
      "grad_norm": 2.253100633621216,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.3544,
      "step": 34
    },
    {
      "epoch": 0.06622516556291391,
      "grad_norm": 2.0696325302124023,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.4927,
      "step": 35
    },
    {
      "epoch": 0.06811731315042574,
      "grad_norm": 2.27337908744812,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.7245,
      "step": 36
    },
    {
      "epoch": 0.07000946073793755,
      "grad_norm": 2.0895378589630127,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.5984,
      "step": 37
    },
    {
      "epoch": 0.07190160832544938,
      "grad_norm": 1.6925121545791626,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.3438,
      "step": 38
    },
    {
      "epoch": 0.07379375591296121,
      "grad_norm": 2.2014336585998535,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.5786,
      "step": 39
    },
    {
      "epoch": 0.07568590350047304,
      "grad_norm": 2.523989677429199,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.4379,
      "step": 40
    },
    {
      "epoch": 0.07757805108798486,
      "grad_norm": 2.0422544479370117,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.6971,
      "step": 41
    },
    {
      "epoch": 0.07947019867549669,
      "grad_norm": 2.4108798503875732,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.4946,
      "step": 42
    },
    {
      "epoch": 0.08136234626300852,
      "grad_norm": 2.110912322998047,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.383,
      "step": 43
    },
    {
      "epoch": 0.08325449385052035,
      "grad_norm": 1.9565459489822388,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.4719,
      "step": 44
    },
    {
      "epoch": 0.08514664143803216,
      "grad_norm": 2.10152530670166,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.283,
      "step": 45
    },
    {
      "epoch": 0.08703878902554399,
      "grad_norm": 3.026787042617798,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.8526,
      "step": 46
    },
    {
      "epoch": 0.08893093661305582,
      "grad_norm": 1.8629006147384644,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.392,
      "step": 47
    },
    {
      "epoch": 0.09082308420056764,
      "grad_norm": 2.531705141067505,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.602,
      "step": 48
    },
    {
      "epoch": 0.09271523178807947,
      "grad_norm": 2.122438669204712,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3731,
      "step": 49
    },
    {
      "epoch": 0.0946073793755913,
      "grad_norm": 1.8923145532608032,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.7168,
      "step": 50
    },
    {
      "epoch": 0.0946073793755913,
      "eval_loss": 1.3977649211883545,
      "eval_runtime": 66.789,
      "eval_samples_per_second": 13.326,
      "eval_steps_per_second": 3.339,
      "step": 50
    },
    {
      "epoch": 0.09649952696310313,
      "grad_norm": 2.109998941421509,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.7043,
      "step": 51
    },
    {
      "epoch": 0.09839167455061495,
      "grad_norm": 1.8613288402557373,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.3952,
      "step": 52
    },
    {
      "epoch": 0.10028382213812677,
      "grad_norm": 2.7694361209869385,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.3591,
      "step": 53
    },
    {
      "epoch": 0.1021759697256386,
      "grad_norm": 1.6803975105285645,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.3981,
      "step": 54
    },
    {
      "epoch": 0.10406811731315042,
      "grad_norm": 1.5886008739471436,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.4392,
      "step": 55
    },
    {
      "epoch": 0.10596026490066225,
      "grad_norm": 1.526458978652954,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.3948,
      "step": 56
    },
    {
      "epoch": 0.10785241248817408,
      "grad_norm": 1.8608886003494263,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.5254,
      "step": 57
    },
    {
      "epoch": 0.1097445600756859,
      "grad_norm": 1.7112435102462769,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.3341,
      "step": 58
    },
    {
      "epoch": 0.11163670766319773,
      "grad_norm": 1.69413423538208,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.3861,
      "step": 59
    },
    {
      "epoch": 0.11352885525070956,
      "grad_norm": 1.7627002000808716,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.4518,
      "step": 60
    },
    {
      "epoch": 0.11542100283822138,
      "grad_norm": 1.795143961906433,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.5952,
      "step": 61
    },
    {
      "epoch": 0.1173131504257332,
      "grad_norm": 1.811423420906067,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.5621,
      "step": 62
    },
    {
      "epoch": 0.11920529801324503,
      "grad_norm": 1.4757728576660156,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.3902,
      "step": 63
    },
    {
      "epoch": 0.12109744560075686,
      "grad_norm": 1.7193387746810913,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.229,
      "step": 64
    },
    {
      "epoch": 0.12298959318826869,
      "grad_norm": 1.6671079397201538,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.2012,
      "step": 65
    },
    {
      "epoch": 0.12488174077578051,
      "grad_norm": 1.8845690488815308,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3359,
      "step": 66
    },
    {
      "epoch": 0.12677388836329234,
      "grad_norm": 1.901315450668335,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.346,
      "step": 67
    },
    {
      "epoch": 0.12866603595080417,
      "grad_norm": 1.4740817546844482,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.0382,
      "step": 68
    },
    {
      "epoch": 0.130558183538316,
      "grad_norm": 2.0071771144866943,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4074,
      "step": 69
    },
    {
      "epoch": 0.13245033112582782,
      "grad_norm": 1.6853525638580322,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.2203,
      "step": 70
    },
    {
      "epoch": 0.13434247871333965,
      "grad_norm": 1.5709892511367798,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.2123,
      "step": 71
    },
    {
      "epoch": 0.13623462630085148,
      "grad_norm": 1.925991177558899,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.072,
      "step": 72
    },
    {
      "epoch": 0.13812677388836328,
      "grad_norm": 1.7214508056640625,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.2429,
      "step": 73
    },
    {
      "epoch": 0.1400189214758751,
      "grad_norm": 1.88029944896698,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.3282,
      "step": 74
    },
    {
      "epoch": 0.14191106906338694,
      "grad_norm": 1.7187904119491577,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.3825,
      "step": 75
    },
    {
      "epoch": 0.14380321665089876,
      "grad_norm": 1.9618759155273438,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.5929,
      "step": 76
    },
    {
      "epoch": 0.1456953642384106,
      "grad_norm": 1.5495868921279907,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.0328,
      "step": 77
    },
    {
      "epoch": 0.14758751182592242,
      "grad_norm": 1.5756014585494995,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.1529,
      "step": 78
    },
    {
      "epoch": 0.14947965941343425,
      "grad_norm": 1.7391507625579834,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3958,
      "step": 79
    },
    {
      "epoch": 0.15137180700094607,
      "grad_norm": 1.7217364311218262,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.2126,
      "step": 80
    },
    {
      "epoch": 0.1532639545884579,
      "grad_norm": 1.5925014019012451,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.3255,
      "step": 81
    },
    {
      "epoch": 0.15515610217596973,
      "grad_norm": 1.6853570938110352,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.2167,
      "step": 82
    },
    {
      "epoch": 0.15704824976348156,
      "grad_norm": 1.5792571306228638,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2772,
      "step": 83
    },
    {
      "epoch": 0.15894039735099338,
      "grad_norm": 1.9697550535202026,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.3278,
      "step": 84
    },
    {
      "epoch": 0.1608325449385052,
      "grad_norm": 1.7860850095748901,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.311,
      "step": 85
    },
    {
      "epoch": 0.16272469252601704,
      "grad_norm": 2.0435984134674072,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.3314,
      "step": 86
    },
    {
      "epoch": 0.16461684011352887,
      "grad_norm": 1.7616769075393677,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.3505,
      "step": 87
    },
    {
      "epoch": 0.1665089877010407,
      "grad_norm": 1.7681667804718018,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.2317,
      "step": 88
    },
    {
      "epoch": 0.1684011352885525,
      "grad_norm": 2.0126516819000244,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.1626,
      "step": 89
    },
    {
      "epoch": 0.17029328287606432,
      "grad_norm": 2.282297372817993,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.6093,
      "step": 90
    },
    {
      "epoch": 0.17218543046357615,
      "grad_norm": 1.9442919492721558,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.42,
      "step": 91
    },
    {
      "epoch": 0.17407757805108798,
      "grad_norm": 1.7442634105682373,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2481,
      "step": 92
    },
    {
      "epoch": 0.1759697256385998,
      "grad_norm": 2.0548510551452637,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.3327,
      "step": 93
    },
    {
      "epoch": 0.17786187322611163,
      "grad_norm": 1.814758062362671,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.3098,
      "step": 94
    },
    {
      "epoch": 0.17975402081362346,
      "grad_norm": 2.1227543354034424,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.1124,
      "step": 95
    },
    {
      "epoch": 0.1816461684011353,
      "grad_norm": 1.6692239046096802,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.3132,
      "step": 96
    },
    {
      "epoch": 0.18353831598864712,
      "grad_norm": 1.683793544769287,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.2399,
      "step": 97
    },
    {
      "epoch": 0.18543046357615894,
      "grad_norm": 2.022306203842163,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.3763,
      "step": 98
    },
    {
      "epoch": 0.18732261116367077,
      "grad_norm": 1.9978488683700562,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.2873,
      "step": 99
    },
    {
      "epoch": 0.1892147587511826,
      "grad_norm": 2.042080879211426,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.2375,
      "step": 100
    },
    {
      "epoch": 0.1892147587511826,
      "eval_loss": 1.2365391254425049,
      "eval_runtime": 66.7273,
      "eval_samples_per_second": 13.338,
      "eval_steps_per_second": 3.342,
      "step": 100
    },
    {
      "epoch": 0.19110690633869443,
      "grad_norm": 1.7195600271224976,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4397,
      "step": 101
    },
    {
      "epoch": 0.19299905392620625,
      "grad_norm": 1.8695496320724487,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2714,
      "step": 102
    },
    {
      "epoch": 0.19489120151371808,
      "grad_norm": 1.579904317855835,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.3455,
      "step": 103
    },
    {
      "epoch": 0.1967833491012299,
      "grad_norm": 1.7603436708450317,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.1797,
      "step": 104
    },
    {
      "epoch": 0.1986754966887417,
      "grad_norm": 1.83244788646698,
      "learning_rate": 5e-05,
      "loss": 1.2,
      "step": 105
    },
    {
      "epoch": 0.20056764427625354,
      "grad_norm": 2.0718467235565186,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.138,
      "step": 106
    },
    {
      "epoch": 0.20245979186376536,
      "grad_norm": 1.8487212657928467,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2829,
      "step": 107
    },
    {
      "epoch": 0.2043519394512772,
      "grad_norm": 2.3473074436187744,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.1983,
      "step": 108
    },
    {
      "epoch": 0.20624408703878902,
      "grad_norm": 1.7346230745315552,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.5133,
      "step": 109
    },
    {
      "epoch": 0.20813623462630085,
      "grad_norm": 1.7954139709472656,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.273,
      "step": 110
    },
    {
      "epoch": 0.21002838221381268,
      "grad_norm": 1.9921718835830688,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.3704,
      "step": 111
    },
    {
      "epoch": 0.2119205298013245,
      "grad_norm": 2.621481418609619,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.2292,
      "step": 112
    },
    {
      "epoch": 0.21381267738883633,
      "grad_norm": 2.2035317420959473,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.1071,
      "step": 113
    },
    {
      "epoch": 0.21570482497634816,
      "grad_norm": 2.2035317420959473,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.044,
      "step": 114
    },
    {
      "epoch": 0.21759697256385999,
      "grad_norm": 2.2800393104553223,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1611,
      "step": 115
    },
    {
      "epoch": 0.2194891201513718,
      "grad_norm": 1.8511682748794556,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.1103,
      "step": 116
    },
    {
      "epoch": 0.22138126773888364,
      "grad_norm": 1.8721153736114502,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.1115,
      "step": 117
    },
    {
      "epoch": 0.22327341532639547,
      "grad_norm": 1.5265511274337769,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.0757,
      "step": 118
    },
    {
      "epoch": 0.2251655629139073,
      "grad_norm": 1.8025033473968506,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.1608,
      "step": 119
    },
    {
      "epoch": 0.22705771050141912,
      "grad_norm": 1.7695872783660889,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.0201,
      "step": 120
    },
    {
      "epoch": 0.22894985808893092,
      "grad_norm": 1.7956249713897705,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.1264,
      "step": 121
    },
    {
      "epoch": 0.23084200567644275,
      "grad_norm": 1.8696130514144897,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2482,
      "step": 122
    },
    {
      "epoch": 0.23273415326395458,
      "grad_norm": 1.7897480726242065,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2374,
      "step": 123
    },
    {
      "epoch": 0.2346263008514664,
      "grad_norm": 1.8421881198883057,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.0501,
      "step": 124
    },
    {
      "epoch": 0.23651844843897823,
      "grad_norm": 1.7816740274429321,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.1444,
      "step": 125
    },
    {
      "epoch": 0.23841059602649006,
      "grad_norm": 1.6408940553665161,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.1843,
      "step": 126
    },
    {
      "epoch": 0.2403027436140019,
      "grad_norm": 1.799723505973816,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.2607,
      "step": 127
    },
    {
      "epoch": 0.24219489120151372,
      "grad_norm": 1.7503219842910767,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.3451,
      "step": 128
    },
    {
      "epoch": 0.24408703878902555,
      "grad_norm": 1.8366929292678833,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.0965,
      "step": 129
    },
    {
      "epoch": 0.24597918637653737,
      "grad_norm": 1.570515751838684,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.9602,
      "step": 130
    },
    {
      "epoch": 0.2478713339640492,
      "grad_norm": 1.8711854219436646,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1383,
      "step": 131
    },
    {
      "epoch": 0.24976348155156103,
      "grad_norm": 1.7590998411178589,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.8858,
      "step": 132
    },
    {
      "epoch": 0.25165562913907286,
      "grad_norm": 1.879822015762329,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.2554,
      "step": 133
    },
    {
      "epoch": 0.2535477767265847,
      "grad_norm": 1.780387282371521,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.9556,
      "step": 134
    },
    {
      "epoch": 0.2554399243140965,
      "grad_norm": 2.600294351577759,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2438,
      "step": 135
    },
    {
      "epoch": 0.25733207190160834,
      "grad_norm": 1.9569140672683716,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.1873,
      "step": 136
    },
    {
      "epoch": 0.25922421948912017,
      "grad_norm": 1.753614902496338,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.0875,
      "step": 137
    },
    {
      "epoch": 0.261116367076632,
      "grad_norm": 1.82419753074646,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.2137,
      "step": 138
    },
    {
      "epoch": 0.2630085146641438,
      "grad_norm": 1.7329962253570557,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.998,
      "step": 139
    },
    {
      "epoch": 0.26490066225165565,
      "grad_norm": 1.8971621990203857,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.2746,
      "step": 140
    },
    {
      "epoch": 0.2667928098391675,
      "grad_norm": 2.097468852996826,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2942,
      "step": 141
    },
    {
      "epoch": 0.2686849574266793,
      "grad_norm": 1.7963608503341675,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.3434,
      "step": 142
    },
    {
      "epoch": 0.27057710501419113,
      "grad_norm": 1.7190752029418945,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1352,
      "step": 143
    },
    {
      "epoch": 0.27246925260170296,
      "grad_norm": 1.9588944911956787,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2619,
      "step": 144
    },
    {
      "epoch": 0.27436140018921473,
      "grad_norm": 2.0283164978027344,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.4427,
      "step": 145
    },
    {
      "epoch": 0.27625354777672656,
      "grad_norm": 1.748177409172058,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.088,
      "step": 146
    },
    {
      "epoch": 0.2781456953642384,
      "grad_norm": 2.112205743789673,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.2643,
      "step": 147
    },
    {
      "epoch": 0.2800378429517502,
      "grad_norm": 2.0163683891296387,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1468,
      "step": 148
    },
    {
      "epoch": 0.28192999053926204,
      "grad_norm": 1.7784696817398071,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.9371,
      "step": 149
    },
    {
      "epoch": 0.28382213812677387,
      "grad_norm": 2.1307387351989746,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.1531,
      "step": 150
    },
    {
      "epoch": 0.28382213812677387,
      "eval_loss": 1.1395671367645264,
      "eval_runtime": 66.7852,
      "eval_samples_per_second": 13.326,
      "eval_steps_per_second": 3.339,
      "step": 150
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 1.8473310470581055,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.1967,
      "step": 151
    },
    {
      "epoch": 0.2876064333017975,
      "grad_norm": 2.269474506378174,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.3963,
      "step": 152
    },
    {
      "epoch": 0.28949858088930935,
      "grad_norm": 1.6505825519561768,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.0957,
      "step": 153
    },
    {
      "epoch": 0.2913907284768212,
      "grad_norm": 2.0257890224456787,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.0526,
      "step": 154
    },
    {
      "epoch": 0.293282876064333,
      "grad_norm": 1.578539490699768,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.1102,
      "step": 155
    },
    {
      "epoch": 0.29517502365184484,
      "grad_norm": 1.731192708015442,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.0575,
      "step": 156
    },
    {
      "epoch": 0.29706717123935666,
      "grad_norm": 1.6806793212890625,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.9428,
      "step": 157
    },
    {
      "epoch": 0.2989593188268685,
      "grad_norm": 2.630728244781494,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.508,
      "step": 158
    },
    {
      "epoch": 0.3008514664143803,
      "grad_norm": 1.693023681640625,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.0684,
      "step": 159
    },
    {
      "epoch": 0.30274361400189215,
      "grad_norm": 1.8537516593933105,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.0644,
      "step": 160
    },
    {
      "epoch": 0.304635761589404,
      "grad_norm": 2.042389392852783,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2159,
      "step": 161
    },
    {
      "epoch": 0.3065279091769158,
      "grad_norm": 1.7469531297683716,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0371,
      "step": 162
    },
    {
      "epoch": 0.30842005676442763,
      "grad_norm": 1.8574581146240234,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0863,
      "step": 163
    },
    {
      "epoch": 0.31031220435193946,
      "grad_norm": 1.6021952629089355,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.9645,
      "step": 164
    },
    {
      "epoch": 0.3122043519394513,
      "grad_norm": 1.8458691835403442,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.0904,
      "step": 165
    },
    {
      "epoch": 0.3140964995269631,
      "grad_norm": 2.2032949924468994,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.2311,
      "step": 166
    },
    {
      "epoch": 0.31598864711447494,
      "grad_norm": 1.703037142753601,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.9176,
      "step": 167
    },
    {
      "epoch": 0.31788079470198677,
      "grad_norm": 1.7083073854446411,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.0886,
      "step": 168
    },
    {
      "epoch": 0.3197729422894986,
      "grad_norm": 1.8962111473083496,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.1489,
      "step": 169
    },
    {
      "epoch": 0.3216650898770104,
      "grad_norm": 2.1403186321258545,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.0412,
      "step": 170
    },
    {
      "epoch": 0.32355723746452225,
      "grad_norm": 1.9753446578979492,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.021,
      "step": 171
    },
    {
      "epoch": 0.3254493850520341,
      "grad_norm": 2.290349006652832,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.1794,
      "step": 172
    },
    {
      "epoch": 0.3273415326395459,
      "grad_norm": 1.8455179929733276,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.0265,
      "step": 173
    },
    {
      "epoch": 0.32923368022705773,
      "grad_norm": 2.217878580093384,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.1141,
      "step": 174
    },
    {
      "epoch": 0.33112582781456956,
      "grad_norm": 2.248539686203003,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.3454,
      "step": 175
    },
    {
      "epoch": 0.3330179754020814,
      "grad_norm": 1.8536025285720825,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9759,
      "step": 176
    },
    {
      "epoch": 0.33491012298959316,
      "grad_norm": 2.457880973815918,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.402,
      "step": 177
    },
    {
      "epoch": 0.336802270577105,
      "grad_norm": 2.3613903522491455,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.1297,
      "step": 178
    },
    {
      "epoch": 0.3386944181646168,
      "grad_norm": 1.802681565284729,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.0184,
      "step": 179
    },
    {
      "epoch": 0.34058656575212864,
      "grad_norm": 1.5964237451553345,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.0135,
      "step": 180
    },
    {
      "epoch": 0.3424787133396405,
      "grad_norm": 1.4919226169586182,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.9119,
      "step": 181
    },
    {
      "epoch": 0.3443708609271523,
      "grad_norm": 1.7520695924758911,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.0671,
      "step": 182
    },
    {
      "epoch": 0.34626300851466413,
      "grad_norm": 1.930341362953186,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.1606,
      "step": 183
    },
    {
      "epoch": 0.34815515610217596,
      "grad_norm": 2.2083332538604736,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.0059,
      "step": 184
    },
    {
      "epoch": 0.3500473036896878,
      "grad_norm": 1.7441879510879517,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.1052,
      "step": 185
    },
    {
      "epoch": 0.3519394512771996,
      "grad_norm": 2.17475962638855,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.212,
      "step": 186
    },
    {
      "epoch": 0.35383159886471144,
      "grad_norm": 1.94939124584198,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0692,
      "step": 187
    },
    {
      "epoch": 0.35572374645222327,
      "grad_norm": 1.8644752502441406,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.0136,
      "step": 188
    },
    {
      "epoch": 0.3576158940397351,
      "grad_norm": 1.5284093618392944,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.9695,
      "step": 189
    },
    {
      "epoch": 0.3595080416272469,
      "grad_norm": 1.8485616445541382,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.1734,
      "step": 190
    },
    {
      "epoch": 0.36140018921475875,
      "grad_norm": 1.6790764331817627,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.0258,
      "step": 191
    },
    {
      "epoch": 0.3632923368022706,
      "grad_norm": 2.109971761703491,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.351,
      "step": 192
    },
    {
      "epoch": 0.3651844843897824,
      "grad_norm": 1.7791240215301514,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.0149,
      "step": 193
    },
    {
      "epoch": 0.36707663197729423,
      "grad_norm": 2.3764374256134033,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.3265,
      "step": 194
    },
    {
      "epoch": 0.36896877956480606,
      "grad_norm": 2.1635193824768066,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.1823,
      "step": 195
    },
    {
      "epoch": 0.3708609271523179,
      "grad_norm": 1.810774564743042,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2866,
      "step": 196
    },
    {
      "epoch": 0.3727530747398297,
      "grad_norm": 1.9317963123321533,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.2588,
      "step": 197
    },
    {
      "epoch": 0.37464522232734154,
      "grad_norm": 2.1256799697875977,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.2946,
      "step": 198
    },
    {
      "epoch": 0.37653736991485337,
      "grad_norm": 1.6562749147415161,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.101,
      "step": 199
    },
    {
      "epoch": 0.3784295175023652,
      "grad_norm": 1.6716514825820923,
      "learning_rate": 0.0,
      "loss": 1.134,
      "step": 200
    },
    {
      "epoch": 0.3784295175023652,
      "eval_loss": 1.1092175245285034,
      "eval_runtime": 66.7828,
      "eval_samples_per_second": 13.327,
      "eval_steps_per_second": 3.339,
      "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": 3.017027157491712e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}