{
  "best_metric": 0.6712756752967834,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.006745874476140685,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.372937238070342e-05,
      "grad_norm": 4.9675726890563965,
      "learning_rate": 1e-05,
      "loss": 2.5173,
      "step": 1
    },
    {
      "epoch": 3.372937238070342e-05,
      "eval_loss": 0.9100027084350586,
      "eval_runtime": 3859.0916,
      "eval_samples_per_second": 12.939,
      "eval_steps_per_second": 3.235,
      "step": 1
    },
    {
      "epoch": 6.745874476140685e-05,
      "grad_norm": 4.62200403213501,
      "learning_rate": 2e-05,
      "loss": 2.4784,
      "step": 2
    },
    {
      "epoch": 0.00010118811714211028,
      "grad_norm": 3.8608202934265137,
      "learning_rate": 3e-05,
      "loss": 2.2684,
      "step": 3
    },
    {
      "epoch": 0.0001349174895228137,
      "grad_norm": 3.470393180847168,
      "learning_rate": 4e-05,
      "loss": 2.2497,
      "step": 4
    },
    {
      "epoch": 0.00016864686190351714,
      "grad_norm": 3.5668654441833496,
      "learning_rate": 5e-05,
      "loss": 2.5262,
      "step": 5
    },
    {
      "epoch": 0.00020237623428422057,
      "grad_norm": 3.75443959236145,
      "learning_rate": 6e-05,
      "loss": 2.7642,
      "step": 6
    },
    {
      "epoch": 0.000236105606664924,
      "grad_norm": 3.943720817565918,
      "learning_rate": 7e-05,
      "loss": 2.5689,
      "step": 7
    },
    {
      "epoch": 0.0002698349790456274,
      "grad_norm": 3.737637996673584,
      "learning_rate": 8e-05,
      "loss": 2.7614,
      "step": 8
    },
    {
      "epoch": 0.00030356435142633083,
      "grad_norm": 4.10280704498291,
      "learning_rate": 9e-05,
      "loss": 2.4234,
      "step": 9
    },
    {
      "epoch": 0.0003372937238070343,
      "grad_norm": 4.241938591003418,
      "learning_rate": 0.0001,
      "loss": 2.5635,
      "step": 10
    },
    {
      "epoch": 0.0003710230961877377,
      "grad_norm": 4.563390254974365,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.1132,
      "step": 11
    },
    {
      "epoch": 0.00040475246856844113,
      "grad_norm": 4.01047420501709,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.6356,
      "step": 12
    },
    {
      "epoch": 0.00043848184094914453,
      "grad_norm": 4.109508991241455,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.7044,
      "step": 13
    },
    {
      "epoch": 0.000472211213329848,
      "grad_norm": 4.838118553161621,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.7929,
      "step": 14
    },
    {
      "epoch": 0.0005059405857105514,
      "grad_norm": 4.246994972229004,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.6147,
      "step": 15
    },
    {
      "epoch": 0.0005396699580912548,
      "grad_norm": 4.243779182434082,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.5871,
      "step": 16
    },
    {
      "epoch": 0.0005733993304719582,
      "grad_norm": 4.342154026031494,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.3968,
      "step": 17
    },
    {
      "epoch": 0.0006071287028526617,
      "grad_norm": 4.179283142089844,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.4878,
      "step": 18
    },
    {
      "epoch": 0.0006408580752333651,
      "grad_norm": 3.999943733215332,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.4634,
      "step": 19
    },
    {
      "epoch": 0.0006745874476140686,
      "grad_norm": 3.8715505599975586,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.2205,
      "step": 20
    },
    {
      "epoch": 0.0007083168199947719,
      "grad_norm": 4.4865946769714355,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.9216,
      "step": 21
    },
    {
      "epoch": 0.0007420461923754754,
      "grad_norm": 3.726553440093994,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.512,
      "step": 22
    },
    {
      "epoch": 0.0007757755647561788,
      "grad_norm": 4.362462043762207,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.5651,
      "step": 23
    },
    {
      "epoch": 0.0008095049371368823,
      "grad_norm": 4.397223472595215,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.8405,
      "step": 24
    },
    {
      "epoch": 0.0008432343095175856,
      "grad_norm": 3.6292521953582764,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.1325,
      "step": 25
    },
    {
      "epoch": 0.0008769636818982891,
      "grad_norm": 3.9084038734436035,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.6292,
      "step": 26
    },
    {
      "epoch": 0.0009106930542789925,
      "grad_norm": 4.10239315032959,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.3033,
      "step": 27
    },
    {
      "epoch": 0.000944422426659696,
      "grad_norm": 5.176626205444336,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.9474,
      "step": 28
    },
    {
      "epoch": 0.0009781517990403994,
      "grad_norm": 4.226640224456787,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.3539,
      "step": 29
    },
    {
      "epoch": 0.0010118811714211029,
      "grad_norm": 4.918447017669678,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.6261,
      "step": 30
    },
    {
      "epoch": 0.0010456105438018063,
      "grad_norm": 4.636198997497559,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.2813,
      "step": 31
    },
    {
      "epoch": 0.0010793399161825095,
      "grad_norm": 4.674834251403809,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.4499,
      "step": 32
    },
    {
      "epoch": 0.001113069288563213,
      "grad_norm": 4.780590057373047,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.3699,
      "step": 33
    },
    {
      "epoch": 0.0011467986609439164,
      "grad_norm": 5.42938756942749,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.5696,
      "step": 34
    },
    {
      "epoch": 0.0011805280333246199,
      "grad_norm": 6.209203243255615,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.7835,
      "step": 35
    },
    {
      "epoch": 0.0012142574057053233,
      "grad_norm": 5.5038743019104,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.8474,
      "step": 36
    },
    {
      "epoch": 0.0012479867780860268,
      "grad_norm": 6.3492865562438965,
      "learning_rate": 9.509956150664796e-05,
      "loss": 3.209,
      "step": 37
    },
    {
      "epoch": 0.0012817161504667302,
      "grad_norm": 6.391679286956787,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.7486,
      "step": 38
    },
    {
      "epoch": 0.0013154455228474337,
      "grad_norm": 5.925459861755371,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.8106,
      "step": 39
    },
    {
      "epoch": 0.0013491748952281371,
      "grad_norm": 7.866671562194824,
      "learning_rate": 9.397368756032445e-05,
      "loss": 3.3968,
      "step": 40
    },
    {
      "epoch": 0.0013829042676088404,
      "grad_norm": 7.10189962387085,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.1876,
      "step": 41
    },
    {
      "epoch": 0.0014166336399895438,
      "grad_norm": 7.296741485595703,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.5158,
      "step": 42
    },
    {
      "epoch": 0.0014503630123702473,
      "grad_norm": 7.449498653411865,
      "learning_rate": 9.273963562927695e-05,
      "loss": 3.6532,
      "step": 43
    },
    {
      "epoch": 0.0014840923847509507,
      "grad_norm": 9.269072532653809,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.3671,
      "step": 44
    },
    {
      "epoch": 0.0015178217571316542,
      "grad_norm": 9.499707221984863,
      "learning_rate": 9.185832391312644e-05,
      "loss": 3.6059,
      "step": 45
    },
    {
      "epoch": 0.0015515511295123576,
      "grad_norm": 10.61764144897461,
      "learning_rate": 9.140044155740101e-05,
      "loss": 4.2079,
      "step": 46
    },
    {
      "epoch": 0.001585280501893061,
      "grad_norm": 11.174882888793945,
      "learning_rate": 9.093124073433463e-05,
      "loss": 3.5928,
      "step": 47
    },
    {
      "epoch": 0.0016190098742737645,
      "grad_norm": 16.714811325073242,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.3573,
      "step": 48
    },
    {
      "epoch": 0.001652739246654468,
      "grad_norm": 18.38225746154785,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.0649,
      "step": 49
    },
    {
      "epoch": 0.0016864686190351712,
      "grad_norm": 20.537551879882812,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.6044,
      "step": 50
    },
    {
      "epoch": 0.0016864686190351712,
      "eval_loss": 0.7880783081054688,
      "eval_runtime": 3884.7275,
      "eval_samples_per_second": 12.854,
      "eval_steps_per_second": 3.214,
      "step": 50
    },
    {
      "epoch": 0.0017201979914158747,
      "grad_norm": 4.31300687789917,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.1438,
      "step": 51
    },
    {
      "epoch": 0.0017539273637965781,
      "grad_norm": 3.882681369781494,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.3107,
      "step": 52
    },
    {
      "epoch": 0.0017876567361772816,
      "grad_norm": 2.9382972717285156,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.3852,
      "step": 53
    },
    {
      "epoch": 0.001821386108557985,
      "grad_norm": 3.1539480686187744,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.1344,
      "step": 54
    },
    {
      "epoch": 0.0018551154809386885,
      "grad_norm": 2.7618463039398193,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.3161,
      "step": 55
    },
    {
      "epoch": 0.001888844853319392,
      "grad_norm": 3.2603886127471924,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.959,
      "step": 56
    },
    {
      "epoch": 0.0019225742257000954,
      "grad_norm": 2.727137565612793,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.4184,
      "step": 57
    },
    {
      "epoch": 0.001956303598080799,
      "grad_norm": 3.054453134536743,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.8178,
      "step": 58
    },
    {
      "epoch": 0.001990032970461502,
      "grad_norm": 3.143711566925049,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.5598,
      "step": 59
    },
    {
      "epoch": 0.0020237623428422057,
      "grad_norm": 3.228459358215332,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.7599,
      "step": 60
    },
    {
      "epoch": 0.002057491715222909,
      "grad_norm": 3.522310972213745,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.5527,
      "step": 61
    },
    {
      "epoch": 0.0020912210876036126,
      "grad_norm": 3.334474563598633,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.9926,
      "step": 62
    },
    {
      "epoch": 0.002124950459984316,
      "grad_norm": 3.318838357925415,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.4928,
      "step": 63
    },
    {
      "epoch": 0.002158679832365019,
      "grad_norm": 3.3245880603790283,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.8187,
      "step": 64
    },
    {
      "epoch": 0.0021924092047457227,
      "grad_norm": 3.322962760925293,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.2101,
      "step": 65
    },
    {
      "epoch": 0.002226138577126426,
      "grad_norm": 2.930368185043335,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.3321,
      "step": 66
    },
    {
      "epoch": 0.0022598679495071296,
      "grad_norm": 3.058220148086548,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.6113,
      "step": 67
    },
    {
      "epoch": 0.002293597321887833,
      "grad_norm": 3.276414632797241,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.9165,
      "step": 68
    },
    {
      "epoch": 0.0023273266942685365,
      "grad_norm": 3.4669575691223145,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.7333,
      "step": 69
    },
    {
      "epoch": 0.0023610560666492398,
      "grad_norm": 3.1421966552734375,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.0868,
      "step": 70
    },
    {
      "epoch": 0.0023947854390299434,
      "grad_norm": 3.565483808517456,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.3425,
      "step": 71
    },
    {
      "epoch": 0.0024285148114106467,
      "grad_norm": 3.4381721019744873,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.4996,
      "step": 72
    },
    {
      "epoch": 0.00246224418379135,
      "grad_norm": 3.4929451942443848,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.5654,
      "step": 73
    },
    {
      "epoch": 0.0024959735561720536,
      "grad_norm": 4.104004859924316,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.2211,
      "step": 74
    },
    {
      "epoch": 0.002529702928552757,
      "grad_norm": 3.5221614837646484,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.3085,
      "step": 75
    },
    {
      "epoch": 0.0025634323009334605,
      "grad_norm": 3.6019487380981445,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.3206,
      "step": 76
    },
    {
      "epoch": 0.0025971616733141637,
      "grad_norm": 3.495086431503296,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.59,
      "step": 77
    },
    {
      "epoch": 0.0026308910456948674,
      "grad_norm": 3.374650478363037,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.4811,
      "step": 78
    },
    {
      "epoch": 0.0026646204180755706,
      "grad_norm": 3.4102554321289062,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.2963,
      "step": 79
    },
    {
      "epoch": 0.0026983497904562743,
      "grad_norm": 4.140578746795654,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.5812,
      "step": 80
    },
    {
      "epoch": 0.0027320791628369775,
      "grad_norm": 3.9557831287384033,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.3223,
      "step": 81
    },
    {
      "epoch": 0.0027658085352176807,
      "grad_norm": 4.113265037536621,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.749,
      "step": 82
    },
    {
      "epoch": 0.0027995379075983844,
      "grad_norm": 4.868432998657227,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.5498,
      "step": 83
    },
    {
      "epoch": 0.0028332672799790876,
      "grad_norm": 4.497052192687988,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.7512,
      "step": 84
    },
    {
      "epoch": 0.0028669966523597913,
      "grad_norm": 4.6095356941223145,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.3831,
      "step": 85
    },
    {
      "epoch": 0.0029007260247404945,
      "grad_norm": 5.214783191680908,
      "learning_rate": 6.545084971874738e-05,
      "loss": 3.2881,
      "step": 86
    },
    {
      "epoch": 0.002934455397121198,
      "grad_norm": 4.7796549797058105,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.7716,
      "step": 87
    },
    {
      "epoch": 0.0029681847695019014,
      "grad_norm": 6.0481061935424805,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.8462,
      "step": 88
    },
    {
      "epoch": 0.003001914141882605,
      "grad_norm": 5.599370002746582,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.9459,
      "step": 89
    },
    {
      "epoch": 0.0030356435142633083,
      "grad_norm": 5.589810371398926,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.7519,
      "step": 90
    },
    {
      "epoch": 0.0030693728866440116,
      "grad_norm": 6.791563510894775,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.516,
      "step": 91
    },
    {
      "epoch": 0.0031031022590247152,
      "grad_norm": 7.459616661071777,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.1134,
      "step": 92
    },
    {
      "epoch": 0.0031368316314054185,
      "grad_norm": 6.73064661026001,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.1224,
      "step": 93
    },
    {
      "epoch": 0.003170561003786122,
      "grad_norm": 7.096497535705566,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.8812,
      "step": 94
    },
    {
      "epoch": 0.0032042903761668254,
      "grad_norm": 7.977499008178711,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 3.3947,
      "step": 95
    },
    {
      "epoch": 0.003238019748547529,
      "grad_norm": 9.823198318481445,
      "learning_rate": 5.74131823855921e-05,
      "loss": 3.7102,
      "step": 96
    },
    {
      "epoch": 0.0032717491209282323,
      "grad_norm": 11.373855590820312,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 3.329,
      "step": 97
    },
    {
      "epoch": 0.003305478493308936,
      "grad_norm": 12.161859512329102,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.9455,
      "step": 98
    },
    {
      "epoch": 0.003339207865689639,
      "grad_norm": 13.403934478759766,
      "learning_rate": 5.495227651252315e-05,
      "loss": 3.3245,
      "step": 99
    },
    {
      "epoch": 0.0033729372380703424,
      "grad_norm": 18.20441246032715,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 3.8852,
      "step": 100
    },
    {
      "epoch": 0.0033729372380703424,
      "eval_loss": 0.7298057675361633,
      "eval_runtime": 3885.2829,
      "eval_samples_per_second": 12.852,
      "eval_steps_per_second": 3.213,
      "step": 100
    },
    {
      "epoch": 0.003406666610451046,
      "grad_norm": 3.111365556716919,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.4397,
      "step": 101
    },
    {
      "epoch": 0.0034403959828317493,
      "grad_norm": 2.9476871490478516,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.0504,
      "step": 102
    },
    {
      "epoch": 0.003474125355212453,
      "grad_norm": 3.170501470565796,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.9921,
      "step": 103
    },
    {
      "epoch": 0.0035078547275931562,
      "grad_norm": 2.601001739501953,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.141,
      "step": 104
    },
    {
      "epoch": 0.00354158409997386,
      "grad_norm": 2.606767177581787,
      "learning_rate": 5e-05,
      "loss": 2.7474,
      "step": 105
    },
    {
      "epoch": 0.003575313472354563,
      "grad_norm": 2.6180262565612793,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.7362,
      "step": 106
    },
    {
      "epoch": 0.003609042844735267,
      "grad_norm": 2.7048604488372803,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.2784,
      "step": 107
    },
    {
      "epoch": 0.00364277221711597,
      "grad_norm": 2.458204507827759,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.3686,
      "step": 108
    },
    {
      "epoch": 0.0036765015894966733,
      "grad_norm": 3.105318307876587,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.9872,
      "step": 109
    },
    {
      "epoch": 0.003710230961877377,
      "grad_norm": 2.9087324142456055,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.5763,
      "step": 110
    },
    {
      "epoch": 0.00374396033425808,
      "grad_norm": 3.3524820804595947,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.5951,
      "step": 111
    },
    {
      "epoch": 0.003777689706638784,
      "grad_norm": 3.0980982780456543,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.4949,
      "step": 112
    },
    {
      "epoch": 0.003811419079019487,
      "grad_norm": 2.9690330028533936,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.6124,
      "step": 113
    },
    {
      "epoch": 0.0038451484514001907,
      "grad_norm": 2.8932576179504395,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.5953,
      "step": 114
    },
    {
      "epoch": 0.003878877823780894,
      "grad_norm": 3.3136117458343506,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.3985,
      "step": 115
    },
    {
      "epoch": 0.003912607196161598,
      "grad_norm": 3.411710262298584,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.7487,
      "step": 116
    },
    {
      "epoch": 0.003946336568542301,
      "grad_norm": 3.244392156600952,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.6008,
      "step": 117
    },
    {
      "epoch": 0.003980065940923004,
      "grad_norm": 3.0537939071655273,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.5601,
      "step": 118
    },
    {
      "epoch": 0.004013795313303707,
      "grad_norm": 2.933382034301758,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.6705,
      "step": 119
    },
    {
      "epoch": 0.004047524685684411,
      "grad_norm": 3.036072015762329,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.166,
      "step": 120
    },
    {
      "epoch": 0.004081254058065115,
      "grad_norm": 3.252941608428955,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.0373,
      "step": 121
    },
    {
      "epoch": 0.004114983430445818,
      "grad_norm": 2.9841160774230957,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.2531,
      "step": 122
    },
    {
      "epoch": 0.004148712802826521,
      "grad_norm": 3.5263969898223877,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.7147,
      "step": 123
    },
    {
      "epoch": 0.004182442175207225,
      "grad_norm": 3.2517738342285156,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.377,
      "step": 124
    },
    {
      "epoch": 0.0042161715475879285,
      "grad_norm": 3.267925977706909,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.0617,
      "step": 125
    },
    {
      "epoch": 0.004249900919968632,
      "grad_norm": 3.7262022495269775,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.6081,
      "step": 126
    },
    {
      "epoch": 0.004283630292349335,
      "grad_norm": 3.4712376594543457,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.4889,
      "step": 127
    },
    {
      "epoch": 0.004317359664730038,
      "grad_norm": 3.6753430366516113,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.5376,
      "step": 128
    },
    {
      "epoch": 0.004351089037110742,
      "grad_norm": 3.4343435764312744,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.2512,
      "step": 129
    },
    {
      "epoch": 0.0043848184094914455,
      "grad_norm": 4.201005935668945,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.1652,
      "step": 130
    },
    {
      "epoch": 0.004418547781872149,
      "grad_norm": 4.135227680206299,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.0392,
      "step": 131
    },
    {
      "epoch": 0.004452277154252852,
      "grad_norm": 4.8751044273376465,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.9728,
      "step": 132
    },
    {
      "epoch": 0.004486006526633556,
      "grad_norm": 4.290978908538818,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.1564,
      "step": 133
    },
    {
      "epoch": 0.004519735899014259,
      "grad_norm": 4.898470878601074,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.3571,
      "step": 134
    },
    {
      "epoch": 0.0045534652713949625,
      "grad_norm": 4.462129592895508,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.3667,
      "step": 135
    },
    {
      "epoch": 0.004587194643775666,
      "grad_norm": 4.757960319519043,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.6919,
      "step": 136
    },
    {
      "epoch": 0.004620924016156369,
      "grad_norm": 4.8541154861450195,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.6552,
      "step": 137
    },
    {
      "epoch": 0.004654653388537073,
      "grad_norm": 5.870574474334717,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.091,
      "step": 138
    },
    {
      "epoch": 0.004688382760917776,
      "grad_norm": 5.609764575958252,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.515,
      "step": 139
    },
    {
      "epoch": 0.0047221121332984796,
      "grad_norm": 5.161116123199463,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.5369,
      "step": 140
    },
    {
      "epoch": 0.004755841505679183,
      "grad_norm": 8.233118057250977,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.0189,
      "step": 141
    },
    {
      "epoch": 0.004789570878059887,
      "grad_norm": 6.386716365814209,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.9592,
      "step": 142
    },
    {
      "epoch": 0.00482330025044059,
      "grad_norm": 7.640812873840332,
      "learning_rate": 2.061073738537635e-05,
      "loss": 3.1168,
      "step": 143
    },
    {
      "epoch": 0.004857029622821293,
      "grad_norm": 8.133594512939453,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 3.0273,
      "step": 144
    },
    {
      "epoch": 0.004890758995201997,
      "grad_norm": 8.013106346130371,
      "learning_rate": 1.928936436551661e-05,
      "loss": 4.2874,
      "step": 145
    },
    {
      "epoch": 0.0049244883675827,
      "grad_norm": 8.809735298156738,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.9128,
      "step": 146
    },
    {
      "epoch": 0.004958217739963404,
      "grad_norm": 9.66916275024414,
      "learning_rate": 1.800157297483417e-05,
      "loss": 3.5167,
      "step": 147
    },
    {
      "epoch": 0.004991947112344107,
      "grad_norm": 10.982966423034668,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 3.11,
      "step": 148
    },
    {
      "epoch": 0.00502567648472481,
      "grad_norm": 11.521635055541992,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.9937,
      "step": 149
    },
    {
      "epoch": 0.005059405857105514,
      "grad_norm": 20.632587432861328,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.7432,
      "step": 150
    },
    {
      "epoch": 0.005059405857105514,
      "eval_loss": 0.6794349551200867,
      "eval_runtime": 3882.3594,
      "eval_samples_per_second": 12.862,
      "eval_steps_per_second": 3.216,
      "step": 150
    },
    {
      "epoch": 0.005093135229486218,
      "grad_norm": 2.4981322288513184,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.0919,
      "step": 151
    },
    {
      "epoch": 0.005126864601866921,
      "grad_norm": 2.3475005626678467,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.4463,
      "step": 152
    },
    {
      "epoch": 0.005160593974247624,
      "grad_norm": 2.4792892932891846,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.1469,
      "step": 153
    },
    {
      "epoch": 0.005194323346628327,
      "grad_norm": 2.2998950481414795,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.2438,
      "step": 154
    },
    {
      "epoch": 0.005228052719009031,
      "grad_norm": 2.3543636798858643,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.2871,
      "step": 155
    },
    {
      "epoch": 0.005261782091389735,
      "grad_norm": 2.587918996810913,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.6881,
      "step": 156
    },
    {
      "epoch": 0.005295511463770438,
      "grad_norm": 2.631528854370117,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.3526,
      "step": 157
    },
    {
      "epoch": 0.005329240836151141,
      "grad_norm": 2.696045398712158,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.6122,
      "step": 158
    },
    {
      "epoch": 0.0053629702085318445,
      "grad_norm": 3.2631983757019043,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.6751,
      "step": 159
    },
    {
      "epoch": 0.0053966995809125486,
      "grad_norm": 3.2978782653808594,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.3895,
      "step": 160
    },
    {
      "epoch": 0.005430428953293252,
      "grad_norm": 3.0057790279388428,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.6181,
      "step": 161
    },
    {
      "epoch": 0.005464158325673955,
      "grad_norm": 2.948392152786255,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.5046,
      "step": 162
    },
    {
      "epoch": 0.005497887698054658,
      "grad_norm": 3.015632152557373,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.9409,
      "step": 163
    },
    {
      "epoch": 0.0055316170704353615,
      "grad_norm": 2.977125406265259,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.3303,
      "step": 164
    },
    {
      "epoch": 0.005565346442816066,
      "grad_norm": 3.0287766456604004,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.5209,
      "step": 165
    },
    {
      "epoch": 0.005599075815196769,
      "grad_norm": 2.773470163345337,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.0332,
      "step": 166
    },
    {
      "epoch": 0.005632805187577472,
      "grad_norm": 3.0762088298797607,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.9397,
      "step": 167
    },
    {
      "epoch": 0.005666534559958175,
      "grad_norm": 3.1546998023986816,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.2505,
      "step": 168
    },
    {
      "epoch": 0.005700263932338879,
      "grad_norm": 3.2390730381011963,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.5438,
      "step": 169
    },
    {
      "epoch": 0.005733993304719583,
      "grad_norm": 2.80165696144104,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.894,
      "step": 170
    },
    {
      "epoch": 0.005767722677100286,
      "grad_norm": 3.5087597370147705,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 3.1929,
      "step": 171
    },
    {
      "epoch": 0.005801452049480989,
      "grad_norm": 3.068382978439331,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.3744,
      "step": 172
    },
    {
      "epoch": 0.005835181421861692,
      "grad_norm": 2.8875439167022705,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.9178,
      "step": 173
    },
    {
      "epoch": 0.005868910794242396,
      "grad_norm": 3.2268171310424805,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.2995,
      "step": 174
    },
    {
      "epoch": 0.0059026401666231,
      "grad_norm": 3.2308332920074463,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.2007,
      "step": 175
    },
    {
      "epoch": 0.005936369539003803,
      "grad_norm": 3.4872491359710693,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.2694,
      "step": 176
    },
    {
      "epoch": 0.005970098911384506,
      "grad_norm": 3.521763324737549,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.3273,
      "step": 177
    },
    {
      "epoch": 0.00600382828376521,
      "grad_norm": 3.2070682048797607,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.2202,
      "step": 178
    },
    {
      "epoch": 0.0060375576561459135,
      "grad_norm": 3.4529716968536377,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.4655,
      "step": 179
    },
    {
      "epoch": 0.006071287028526617,
      "grad_norm": 3.7131290435791016,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.6974,
      "step": 180
    },
    {
      "epoch": 0.00610501640090732,
      "grad_norm": 3.5592501163482666,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.0068,
      "step": 181
    },
    {
      "epoch": 0.006138745773288023,
      "grad_norm": 4.195763111114502,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.81,
      "step": 182
    },
    {
      "epoch": 0.006172475145668727,
      "grad_norm": 4.668822288513184,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.3004,
      "step": 183
    },
    {
      "epoch": 0.0062062045180494305,
      "grad_norm": 4.005098342895508,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.2855,
      "step": 184
    },
    {
      "epoch": 0.006239933890430134,
      "grad_norm": 4.315607070922852,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.3335,
      "step": 185
    },
    {
      "epoch": 0.006273663262810837,
      "grad_norm": 4.2397379875183105,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.5642,
      "step": 186
    },
    {
      "epoch": 0.006307392635191541,
      "grad_norm": 5.775464057922363,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.8801,
      "step": 187
    },
    {
      "epoch": 0.006341122007572244,
      "grad_norm": 4.949732303619385,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.0565,
      "step": 188
    },
    {
      "epoch": 0.0063748513799529475,
      "grad_norm": 5.895129680633545,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.7886,
      "step": 189
    },
    {
      "epoch": 0.006408580752333651,
      "grad_norm": 6.09521484375,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.3757,
      "step": 190
    },
    {
      "epoch": 0.006442310124714354,
      "grad_norm": 6.1414475440979,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.2645,
      "step": 191
    },
    {
      "epoch": 0.006476039497095058,
      "grad_norm": 6.252451419830322,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.6678,
      "step": 192
    },
    {
      "epoch": 0.006509768869475761,
      "grad_norm": 6.535440444946289,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.5947,
      "step": 193
    },
    {
      "epoch": 0.0065434982418564646,
      "grad_norm": 7.74290132522583,
      "learning_rate": 2.458548727494292e-07,
      "loss": 3.8317,
      "step": 194
    },
    {
      "epoch": 0.006577227614237168,
      "grad_norm": 8.696444511413574,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 3.8377,
      "step": 195
    },
    {
      "epoch": 0.006610956986617872,
      "grad_norm": 7.943871021270752,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 3.7488,
      "step": 196
    },
    {
      "epoch": 0.006644686358998575,
      "grad_norm": 10.58278751373291,
      "learning_rate": 6.150154258476315e-08,
      "loss": 3.7804,
      "step": 197
    },
    {
      "epoch": 0.006678415731379278,
      "grad_norm": 12.448182106018066,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 3.6433,
      "step": 198
    },
    {
      "epoch": 0.006712145103759982,
      "grad_norm": 16.00191307067871,
      "learning_rate": 6.834750376549792e-09,
      "loss": 3.5287,
      "step": 199
    },
    {
      "epoch": 0.006745874476140685,
      "grad_norm": 19.31696891784668,
      "learning_rate": 0.0,
      "loss": 4.5231,
      "step": 200
    },
    {
      "epoch": 0.006745874476140685,
      "eval_loss": 0.6712756752967834,
      "eval_runtime": 3892.6543,
      "eval_samples_per_second": 12.827,
      "eval_steps_per_second": 3.207,
      "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": 2.984920369720197e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}