{
  "best_metric": 11.824332237243652,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.03653301671385514,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00018266508356927573,
      "grad_norm": 0.3465905487537384,
      "learning_rate": 1e-05,
      "loss": 11.9383,
      "step": 1
    },
    {
      "epoch": 0.00018266508356927573,
      "eval_loss": 11.941452026367188,
      "eval_runtime": 90.1914,
      "eval_samples_per_second": 102.227,
      "eval_steps_per_second": 25.557,
      "step": 1
    },
    {
      "epoch": 0.00036533016713855145,
      "grad_norm": 0.3875073492527008,
      "learning_rate": 2e-05,
      "loss": 11.9331,
      "step": 2
    },
    {
      "epoch": 0.0005479952507078272,
      "grad_norm": 0.385800838470459,
      "learning_rate": 3e-05,
      "loss": 11.9474,
      "step": 3
    },
    {
      "epoch": 0.0007306603342771029,
      "grad_norm": 0.35018494725227356,
      "learning_rate": 4e-05,
      "loss": 11.941,
      "step": 4
    },
    {
      "epoch": 0.0009133254178463787,
      "grad_norm": 0.4610854983329773,
      "learning_rate": 5e-05,
      "loss": 11.9401,
      "step": 5
    },
    {
      "epoch": 0.0010959905014156545,
      "grad_norm": 0.36862891912460327,
      "learning_rate": 6e-05,
      "loss": 11.9362,
      "step": 6
    },
    {
      "epoch": 0.0012786555849849301,
      "grad_norm": 0.4254288077354431,
      "learning_rate": 7e-05,
      "loss": 11.9351,
      "step": 7
    },
    {
      "epoch": 0.0014613206685542058,
      "grad_norm": 0.46606364846229553,
      "learning_rate": 8e-05,
      "loss": 11.9386,
      "step": 8
    },
    {
      "epoch": 0.0016439857521234817,
      "grad_norm": 0.3738418519496918,
      "learning_rate": 9e-05,
      "loss": 11.9359,
      "step": 9
    },
    {
      "epoch": 0.0018266508356927574,
      "grad_norm": 0.3796614408493042,
      "learning_rate": 0.0001,
      "loss": 11.9368,
      "step": 10
    },
    {
      "epoch": 0.002009315919262033,
      "grad_norm": 0.4579927623271942,
      "learning_rate": 9.999316524962345e-05,
      "loss": 11.9377,
      "step": 11
    },
    {
      "epoch": 0.002191981002831309,
      "grad_norm": 0.43001529574394226,
      "learning_rate": 9.997266286704631e-05,
      "loss": 11.9321,
      "step": 12
    },
    {
      "epoch": 0.0023746460864005844,
      "grad_norm": 0.4973090589046478,
      "learning_rate": 9.993849845741524e-05,
      "loss": 11.9264,
      "step": 13
    },
    {
      "epoch": 0.0025573111699698603,
      "grad_norm": 0.5061120986938477,
      "learning_rate": 9.989068136093873e-05,
      "loss": 11.9258,
      "step": 14
    },
    {
      "epoch": 0.002739976253539136,
      "grad_norm": 0.5346381664276123,
      "learning_rate": 9.98292246503335e-05,
      "loss": 11.9225,
      "step": 15
    },
    {
      "epoch": 0.0029226413371084116,
      "grad_norm": 0.550250232219696,
      "learning_rate": 9.975414512725057e-05,
      "loss": 11.9169,
      "step": 16
    },
    {
      "epoch": 0.0031053064206776875,
      "grad_norm": 0.555730938911438,
      "learning_rate": 9.966546331768191e-05,
      "loss": 11.9112,
      "step": 17
    },
    {
      "epoch": 0.0032879715042469634,
      "grad_norm": 0.5179576873779297,
      "learning_rate": 9.956320346634876e-05,
      "loss": 11.9131,
      "step": 18
    },
    {
      "epoch": 0.003470636587816239,
      "grad_norm": 0.6362577080726624,
      "learning_rate": 9.944739353007344e-05,
      "loss": 11.9082,
      "step": 19
    },
    {
      "epoch": 0.0036533016713855147,
      "grad_norm": 0.4821767807006836,
      "learning_rate": 9.931806517013612e-05,
      "loss": 11.9116,
      "step": 20
    },
    {
      "epoch": 0.00383596675495479,
      "grad_norm": 0.6245077252388,
      "learning_rate": 9.917525374361912e-05,
      "loss": 11.8964,
      "step": 21
    },
    {
      "epoch": 0.004018631838524066,
      "grad_norm": 0.6252933740615845,
      "learning_rate": 9.901899829374047e-05,
      "loss": 11.9044,
      "step": 22
    },
    {
      "epoch": 0.004201296922093342,
      "grad_norm": 0.6258057951927185,
      "learning_rate": 9.884934153917997e-05,
      "loss": 11.8952,
      "step": 23
    },
    {
      "epoch": 0.004383962005662618,
      "grad_norm": 0.6631948351860046,
      "learning_rate": 9.86663298624003e-05,
      "loss": 11.881,
      "step": 24
    },
    {
      "epoch": 0.004566627089231893,
      "grad_norm": 0.6315574049949646,
      "learning_rate": 9.847001329696653e-05,
      "loss": 11.8787,
      "step": 25
    },
    {
      "epoch": 0.004749292172801169,
      "grad_norm": 0.6357966661453247,
      "learning_rate": 9.826044551386744e-05,
      "loss": 11.8683,
      "step": 26
    },
    {
      "epoch": 0.004931957256370445,
      "grad_norm": 0.5510373711585999,
      "learning_rate": 9.803768380684242e-05,
      "loss": 11.8724,
      "step": 27
    },
    {
      "epoch": 0.0051146223399397206,
      "grad_norm": 0.5501624941825867,
      "learning_rate": 9.780178907671789e-05,
      "loss": 11.8748,
      "step": 28
    },
    {
      "epoch": 0.0052972874235089965,
      "grad_norm": 0.6480464339256287,
      "learning_rate": 9.755282581475769e-05,
      "loss": 11.8677,
      "step": 29
    },
    {
      "epoch": 0.005479952507078272,
      "grad_norm": 0.6431350111961365,
      "learning_rate": 9.729086208503174e-05,
      "loss": 11.8509,
      "step": 30
    },
    {
      "epoch": 0.005662617590647547,
      "grad_norm": 0.6059749126434326,
      "learning_rate": 9.701596950580806e-05,
      "loss": 11.8625,
      "step": 31
    },
    {
      "epoch": 0.005845282674216823,
      "grad_norm": 0.4988415837287903,
      "learning_rate": 9.672822322997305e-05,
      "loss": 11.8419,
      "step": 32
    },
    {
      "epoch": 0.006027947757786099,
      "grad_norm": 0.5706687569618225,
      "learning_rate": 9.642770192448536e-05,
      "loss": 11.8465,
      "step": 33
    },
    {
      "epoch": 0.006210612841355375,
      "grad_norm": 0.5235163569450378,
      "learning_rate": 9.611448774886924e-05,
      "loss": 11.8487,
      "step": 34
    },
    {
      "epoch": 0.006393277924924651,
      "grad_norm": 0.5092278718948364,
      "learning_rate": 9.578866633275288e-05,
      "loss": 11.8554,
      "step": 35
    },
    {
      "epoch": 0.006575943008493927,
      "grad_norm": 0.45471179485321045,
      "learning_rate": 9.545032675245813e-05,
      "loss": 11.8492,
      "step": 36
    },
    {
      "epoch": 0.006758608092063202,
      "grad_norm": 0.4756167232990265,
      "learning_rate": 9.509956150664796e-05,
      "loss": 11.8481,
      "step": 37
    },
    {
      "epoch": 0.006941273175632478,
      "grad_norm": 0.4648807942867279,
      "learning_rate": 9.473646649103818e-05,
      "loss": 11.8485,
      "step": 38
    },
    {
      "epoch": 0.007123938259201754,
      "grad_norm": 0.37208059430122375,
      "learning_rate": 9.43611409721806e-05,
      "loss": 11.852,
      "step": 39
    },
    {
      "epoch": 0.0073066033427710295,
      "grad_norm": 0.304571270942688,
      "learning_rate": 9.397368756032445e-05,
      "loss": 11.8274,
      "step": 40
    },
    {
      "epoch": 0.007489268426340305,
      "grad_norm": 0.4037236273288727,
      "learning_rate": 9.357421218136386e-05,
      "loss": 11.8363,
      "step": 41
    },
    {
      "epoch": 0.00767193350990958,
      "grad_norm": 0.3837885856628418,
      "learning_rate": 9.316282404787871e-05,
      "loss": 11.8331,
      "step": 42
    },
    {
      "epoch": 0.007854598593478857,
      "grad_norm": 0.3640628159046173,
      "learning_rate": 9.273963562927695e-05,
      "loss": 11.8436,
      "step": 43
    },
    {
      "epoch": 0.008037263677048132,
      "grad_norm": 0.4586373269557953,
      "learning_rate": 9.230476262104677e-05,
      "loss": 11.8269,
      "step": 44
    },
    {
      "epoch": 0.008219928760617407,
      "grad_norm": 0.4907529354095459,
      "learning_rate": 9.185832391312644e-05,
      "loss": 11.8429,
      "step": 45
    },
    {
      "epoch": 0.008402593844186684,
      "grad_norm": 0.45087334513664246,
      "learning_rate": 9.140044155740101e-05,
      "loss": 11.8005,
      "step": 46
    },
    {
      "epoch": 0.008585258927755959,
      "grad_norm": 0.4851132929325104,
      "learning_rate": 9.093124073433463e-05,
      "loss": 11.8317,
      "step": 47
    },
    {
      "epoch": 0.008767924011325236,
      "grad_norm": 0.5941483378410339,
      "learning_rate": 9.045084971874738e-05,
      "loss": 11.8192,
      "step": 48
    },
    {
      "epoch": 0.00895058909489451,
      "grad_norm": 0.6480814814567566,
      "learning_rate": 8.995939984474624e-05,
      "loss": 11.7957,
      "step": 49
    },
    {
      "epoch": 0.009133254178463786,
      "grad_norm": 0.625806450843811,
      "learning_rate": 8.945702546981969e-05,
      "loss": 11.7945,
      "step": 50
    },
    {
      "epoch": 0.009133254178463786,
      "eval_loss": 11.844329833984375,
      "eval_runtime": 89.2908,
      "eval_samples_per_second": 103.258,
      "eval_steps_per_second": 25.815,
      "step": 50
    },
    {
      "epoch": 0.009315919262033063,
      "grad_norm": 0.4215286076068878,
      "learning_rate": 8.894386393810563e-05,
      "loss": 11.872,
      "step": 51
    },
    {
      "epoch": 0.009498584345602338,
      "grad_norm": 0.38242530822753906,
      "learning_rate": 8.842005554284296e-05,
      "loss": 11.8773,
      "step": 52
    },
    {
      "epoch": 0.009681249429171614,
      "grad_norm": 0.3612035810947418,
      "learning_rate": 8.788574348801675e-05,
      "loss": 11.8815,
      "step": 53
    },
    {
      "epoch": 0.00986391451274089,
      "grad_norm": 0.35544878244400024,
      "learning_rate": 8.73410738492077e-05,
      "loss": 11.8677,
      "step": 54
    },
    {
      "epoch": 0.010046579596310166,
      "grad_norm": 0.3199334740638733,
      "learning_rate": 8.678619553365659e-05,
      "loss": 11.8693,
      "step": 55
    },
    {
      "epoch": 0.010229244679879441,
      "grad_norm": 0.35227280855178833,
      "learning_rate": 8.622126023955446e-05,
      "loss": 11.8762,
      "step": 56
    },
    {
      "epoch": 0.010411909763448716,
      "grad_norm": 0.3259381055831909,
      "learning_rate": 8.564642241456986e-05,
      "loss": 11.8732,
      "step": 57
    },
    {
      "epoch": 0.010594574847017993,
      "grad_norm": 0.30155816674232483,
      "learning_rate": 8.506183921362443e-05,
      "loss": 11.8711,
      "step": 58
    },
    {
      "epoch": 0.010777239930587268,
      "grad_norm": 0.27700304985046387,
      "learning_rate": 8.44676704559283e-05,
      "loss": 11.855,
      "step": 59
    },
    {
      "epoch": 0.010959905014156545,
      "grad_norm": 0.2976490259170532,
      "learning_rate": 8.386407858128706e-05,
      "loss": 11.8626,
      "step": 60
    },
    {
      "epoch": 0.01114257009772582,
      "grad_norm": 0.31656286120414734,
      "learning_rate": 8.32512286056924e-05,
      "loss": 11.8676,
      "step": 61
    },
    {
      "epoch": 0.011325235181295095,
      "grad_norm": 0.29914695024490356,
      "learning_rate": 8.262928807620843e-05,
      "loss": 11.857,
      "step": 62
    },
    {
      "epoch": 0.011507900264864371,
      "grad_norm": 0.25016099214553833,
      "learning_rate": 8.199842702516583e-05,
      "loss": 11.8615,
      "step": 63
    },
    {
      "epoch": 0.011690565348433647,
      "grad_norm": 0.26599329710006714,
      "learning_rate": 8.135881792367686e-05,
      "loss": 11.8535,
      "step": 64
    },
    {
      "epoch": 0.011873230432002923,
      "grad_norm": 0.18718163669109344,
      "learning_rate": 8.07106356344834e-05,
      "loss": 11.8586,
      "step": 65
    },
    {
      "epoch": 0.012055895515572198,
      "grad_norm": 0.18403424322605133,
      "learning_rate": 8.005405736415126e-05,
      "loss": 11.8622,
      "step": 66
    },
    {
      "epoch": 0.012238560599141473,
      "grad_norm": 0.24058836698532104,
      "learning_rate": 7.938926261462366e-05,
      "loss": 11.8538,
      "step": 67
    },
    {
      "epoch": 0.01242122568271075,
      "grad_norm": 0.25762394070625305,
      "learning_rate": 7.871643313414718e-05,
      "loss": 11.8555,
      "step": 68
    },
    {
      "epoch": 0.012603890766280025,
      "grad_norm": 0.2264672964811325,
      "learning_rate": 7.803575286758364e-05,
      "loss": 11.861,
      "step": 69
    },
    {
      "epoch": 0.012786555849849302,
      "grad_norm": 0.20667871832847595,
      "learning_rate": 7.734740790612136e-05,
      "loss": 11.8506,
      "step": 70
    },
    {
      "epoch": 0.012969220933418577,
      "grad_norm": 0.19381491839885712,
      "learning_rate": 7.66515864363997e-05,
      "loss": 11.8489,
      "step": 71
    },
    {
      "epoch": 0.013151886016987854,
      "grad_norm": 0.17694498598575592,
      "learning_rate": 7.594847868906076e-05,
      "loss": 11.8581,
      "step": 72
    },
    {
      "epoch": 0.013334551100557129,
      "grad_norm": 0.20530350506305695,
      "learning_rate": 7.52382768867422e-05,
      "loss": 11.8504,
      "step": 73
    },
    {
      "epoch": 0.013517216184126404,
      "grad_norm": 0.26066431403160095,
      "learning_rate": 7.452117519152542e-05,
      "loss": 11.8361,
      "step": 74
    },
    {
      "epoch": 0.01369988126769568,
      "grad_norm": 0.20842689275741577,
      "learning_rate": 7.379736965185368e-05,
      "loss": 11.8456,
      "step": 75
    },
    {
      "epoch": 0.013882546351264955,
      "grad_norm": 0.23757736384868622,
      "learning_rate": 7.30670581489344e-05,
      "loss": 11.8476,
      "step": 76
    },
    {
      "epoch": 0.014065211434834232,
      "grad_norm": 0.2254098653793335,
      "learning_rate": 7.233044034264034e-05,
      "loss": 11.8304,
      "step": 77
    },
    {
      "epoch": 0.014247876518403507,
      "grad_norm": 0.2505176365375519,
      "learning_rate": 7.158771761692464e-05,
      "loss": 11.8264,
      "step": 78
    },
    {
      "epoch": 0.014430541601972782,
      "grad_norm": 0.2441072165966034,
      "learning_rate": 7.083909302476453e-05,
      "loss": 11.8482,
      "step": 79
    },
    {
      "epoch": 0.014613206685542059,
      "grad_norm": 0.24809595942497253,
      "learning_rate": 7.008477123264848e-05,
      "loss": 11.817,
      "step": 80
    },
    {
      "epoch": 0.014795871769111334,
      "grad_norm": 0.20436261594295502,
      "learning_rate": 6.932495846462261e-05,
      "loss": 11.8415,
      "step": 81
    },
    {
      "epoch": 0.01497853685268061,
      "grad_norm": 0.27289894223213196,
      "learning_rate": 6.855986244591104e-05,
      "loss": 11.8446,
      "step": 82
    },
    {
      "epoch": 0.015161201936249886,
      "grad_norm": 0.24686995148658752,
      "learning_rate": 6.778969234612584e-05,
      "loss": 11.83,
      "step": 83
    },
    {
      "epoch": 0.01534386701981916,
      "grad_norm": 0.2734980285167694,
      "learning_rate": 6.701465872208216e-05,
      "loss": 11.8289,
      "step": 84
    },
    {
      "epoch": 0.015526532103388438,
      "grad_norm": 0.2750653326511383,
      "learning_rate": 6.623497346023418e-05,
      "loss": 11.8305,
      "step": 85
    },
    {
      "epoch": 0.015709197186957714,
      "grad_norm": 0.3146180510520935,
      "learning_rate": 6.545084971874738e-05,
      "loss": 11.8163,
      "step": 86
    },
    {
      "epoch": 0.01589186227052699,
      "grad_norm": 0.24329495429992676,
      "learning_rate": 6.466250186922325e-05,
      "loss": 11.8271,
      "step": 87
    },
    {
      "epoch": 0.016074527354096264,
      "grad_norm": 0.3243590295314789,
      "learning_rate": 6.387014543809223e-05,
      "loss": 11.8306,
      "step": 88
    },
    {
      "epoch": 0.01625719243766554,
      "grad_norm": 0.25212496519088745,
      "learning_rate": 6.307399704769099e-05,
      "loss": 11.8358,
      "step": 89
    },
    {
      "epoch": 0.016439857521234814,
      "grad_norm": 0.30638131499290466,
      "learning_rate": 6.227427435703997e-05,
      "loss": 11.808,
      "step": 90
    },
    {
      "epoch": 0.016622522604804093,
      "grad_norm": 0.3348468542098999,
      "learning_rate": 6.147119600233758e-05,
      "loss": 11.8296,
      "step": 91
    },
    {
      "epoch": 0.016805187688373368,
      "grad_norm": 0.3098547160625458,
      "learning_rate": 6.066498153718735e-05,
      "loss": 11.8406,
      "step": 92
    },
    {
      "epoch": 0.016987852771942643,
      "grad_norm": 0.30615565180778503,
      "learning_rate": 5.985585137257401e-05,
      "loss": 11.8226,
      "step": 93
    },
    {
      "epoch": 0.017170517855511918,
      "grad_norm": 0.35668668150901794,
      "learning_rate": 5.90440267166055e-05,
      "loss": 11.8253,
      "step": 94
    },
    {
      "epoch": 0.017353182939081193,
      "grad_norm": 0.3204077184200287,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 11.8299,
      "step": 95
    },
    {
      "epoch": 0.01753584802265047,
      "grad_norm": 0.3196731209754944,
      "learning_rate": 5.74131823855921e-05,
      "loss": 11.8266,
      "step": 96
    },
    {
      "epoch": 0.017718513106219747,
      "grad_norm": 0.37118634581565857,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 11.8122,
      "step": 97
    },
    {
      "epoch": 0.01790117818978902,
      "grad_norm": 0.47526612877845764,
      "learning_rate": 5.577423184847932e-05,
      "loss": 11.8155,
      "step": 98
    },
    {
      "epoch": 0.018083843273358297,
      "grad_norm": 0.5606823563575745,
      "learning_rate": 5.495227651252315e-05,
      "loss": 11.7897,
      "step": 99
    },
    {
      "epoch": 0.01826650835692757,
      "grad_norm": 0.6295754313468933,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 11.7881,
      "step": 100
    },
    {
      "epoch": 0.01826650835692757,
      "eval_loss": 11.834319114685059,
      "eval_runtime": 89.0583,
      "eval_samples_per_second": 103.528,
      "eval_steps_per_second": 25.882,
      "step": 100
    },
    {
      "epoch": 0.01844917344049685,
      "grad_norm": 0.3653700053691864,
      "learning_rate": 5.330452921628497e-05,
      "loss": 11.8686,
      "step": 101
    },
    {
      "epoch": 0.018631838524066125,
      "grad_norm": 0.3573233485221863,
      "learning_rate": 5.247918773366112e-05,
      "loss": 11.8757,
      "step": 102
    },
    {
      "epoch": 0.0188145036076354,
      "grad_norm": 0.2834525406360626,
      "learning_rate": 5.165316846586541e-05,
      "loss": 11.866,
      "step": 103
    },
    {
      "epoch": 0.018997168691204675,
      "grad_norm": 0.33954983949661255,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 11.8628,
      "step": 104
    },
    {
      "epoch": 0.019179833774773954,
      "grad_norm": 0.3091084361076355,
      "learning_rate": 5e-05,
      "loss": 11.8581,
      "step": 105
    },
    {
      "epoch": 0.01936249885834323,
      "grad_norm": 0.2740994393825531,
      "learning_rate": 4.917330276168208e-05,
      "loss": 11.8606,
      "step": 106
    },
    {
      "epoch": 0.019545163941912504,
      "grad_norm": 0.31202438473701477,
      "learning_rate": 4.834683153413459e-05,
      "loss": 11.8506,
      "step": 107
    },
    {
      "epoch": 0.01972782902548178,
      "grad_norm": 0.2627750039100647,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 11.8582,
      "step": 108
    },
    {
      "epoch": 0.019910494109051054,
      "grad_norm": 0.329013466835022,
      "learning_rate": 4.669547078371504e-05,
      "loss": 11.8586,
      "step": 109
    },
    {
      "epoch": 0.020093159192620332,
      "grad_norm": 0.22419841587543488,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 11.8602,
      "step": 110
    },
    {
      "epoch": 0.020275824276189607,
      "grad_norm": 0.2340073436498642,
      "learning_rate": 4.504772348747687e-05,
      "loss": 11.8525,
      "step": 111
    },
    {
      "epoch": 0.020458489359758882,
      "grad_norm": 0.31201285123825073,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 11.8318,
      "step": 112
    },
    {
      "epoch": 0.020641154443328157,
      "grad_norm": 0.2925834655761719,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 11.8479,
      "step": 113
    },
    {
      "epoch": 0.020823819526897432,
      "grad_norm": 0.2644873857498169,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 11.847,
      "step": 114
    },
    {
      "epoch": 0.02100648461046671,
      "grad_norm": 0.2717932462692261,
      "learning_rate": 4.17702704859633e-05,
      "loss": 11.8337,
      "step": 115
    },
    {
      "epoch": 0.021189149694035986,
      "grad_norm": 0.21851548552513123,
      "learning_rate": 4.095597328339452e-05,
      "loss": 11.8367,
      "step": 116
    },
    {
      "epoch": 0.02137181477760526,
      "grad_norm": 0.2520385682582855,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 11.8478,
      "step": 117
    },
    {
      "epoch": 0.021554479861174536,
      "grad_norm": 0.1869116574525833,
      "learning_rate": 3.933501846281267e-05,
      "loss": 11.8404,
      "step": 118
    },
    {
      "epoch": 0.02173714494474381,
      "grad_norm": 0.23700135946273804,
      "learning_rate": 3.852880399766243e-05,
      "loss": 11.8413,
      "step": 119
    },
    {
      "epoch": 0.02191981002831309,
      "grad_norm": 0.22343429923057556,
      "learning_rate": 3.772572564296005e-05,
      "loss": 11.8336,
      "step": 120
    },
    {
      "epoch": 0.022102475111882364,
      "grad_norm": 0.2725762724876404,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 11.8398,
      "step": 121
    },
    {
      "epoch": 0.02228514019545164,
      "grad_norm": 0.2449876368045807,
      "learning_rate": 3.612985456190778e-05,
      "loss": 11.8404,
      "step": 122
    },
    {
      "epoch": 0.022467805279020914,
      "grad_norm": 0.29339998960494995,
      "learning_rate": 3.533749813077677e-05,
      "loss": 11.8229,
      "step": 123
    },
    {
      "epoch": 0.02265047036259019,
      "grad_norm": 0.2512277364730835,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 11.8453,
      "step": 124
    },
    {
      "epoch": 0.022833135446159468,
      "grad_norm": 0.2491714209318161,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 11.8227,
      "step": 125
    },
    {
      "epoch": 0.023015800529728743,
      "grad_norm": 0.26697126030921936,
      "learning_rate": 3.298534127791785e-05,
      "loss": 11.8164,
      "step": 126
    },
    {
      "epoch": 0.023198465613298018,
      "grad_norm": 0.2489868402481079,
      "learning_rate": 3.221030765387417e-05,
      "loss": 11.8189,
      "step": 127
    },
    {
      "epoch": 0.023381130696867293,
      "grad_norm": 0.25719183683395386,
      "learning_rate": 3.144013755408895e-05,
      "loss": 11.8267,
      "step": 128
    },
    {
      "epoch": 0.023563795780436568,
      "grad_norm": 0.2649552524089813,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 11.8061,
      "step": 129
    },
    {
      "epoch": 0.023746460864005847,
      "grad_norm": 0.23857741057872772,
      "learning_rate": 2.991522876735154e-05,
      "loss": 11.8377,
      "step": 130
    },
    {
      "epoch": 0.02392912594757512,
      "grad_norm": 0.2775125205516815,
      "learning_rate": 2.916090697523549e-05,
      "loss": 11.8143,
      "step": 131
    },
    {
      "epoch": 0.024111791031144397,
      "grad_norm": 0.30382922291755676,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 11.8277,
      "step": 132
    },
    {
      "epoch": 0.02429445611471367,
      "grad_norm": 0.32927384972572327,
      "learning_rate": 2.766955965735968e-05,
      "loss": 11.8087,
      "step": 133
    },
    {
      "epoch": 0.024477121198282947,
      "grad_norm": 0.2699863016605377,
      "learning_rate": 2.693294185106562e-05,
      "loss": 11.8317,
      "step": 134
    },
    {
      "epoch": 0.024659786281852225,
      "grad_norm": 0.30144500732421875,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 11.8123,
      "step": 135
    },
    {
      "epoch": 0.0248424513654215,
      "grad_norm": 0.2767479419708252,
      "learning_rate": 2.547882480847461e-05,
      "loss": 11.8192,
      "step": 136
    },
    {
      "epoch": 0.025025116448990775,
      "grad_norm": 0.29739877581596375,
      "learning_rate": 2.476172311325783e-05,
      "loss": 11.8268,
      "step": 137
    },
    {
      "epoch": 0.02520778153256005,
      "grad_norm": 0.2872367799282074,
      "learning_rate": 2.405152131093926e-05,
      "loss": 11.8314,
      "step": 138
    },
    {
      "epoch": 0.025390446616129325,
      "grad_norm": 0.3332676589488983,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 11.8087,
      "step": 139
    },
    {
      "epoch": 0.025573111699698604,
      "grad_norm": 0.32409340143203735,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 11.8007,
      "step": 140
    },
    {
      "epoch": 0.02575577678326788,
      "grad_norm": 0.2905734181404114,
      "learning_rate": 2.196424713241637e-05,
      "loss": 11.8206,
      "step": 141
    },
    {
      "epoch": 0.025938441866837154,
      "grad_norm": 0.38151589035987854,
      "learning_rate": 2.128356686585282e-05,
      "loss": 11.8187,
      "step": 142
    },
    {
      "epoch": 0.02612110695040643,
      "grad_norm": 0.31705883145332336,
      "learning_rate": 2.061073738537635e-05,
      "loss": 11.8001,
      "step": 143
    },
    {
      "epoch": 0.026303772033975707,
      "grad_norm": 0.3771505355834961,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 11.8197,
      "step": 144
    },
    {
      "epoch": 0.026486437117544982,
      "grad_norm": 0.32985883951187134,
      "learning_rate": 1.928936436551661e-05,
      "loss": 11.8025,
      "step": 145
    },
    {
      "epoch": 0.026669102201114257,
      "grad_norm": 0.4287067949771881,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 11.819,
      "step": 146
    },
    {
      "epoch": 0.026851767284683532,
      "grad_norm": 0.43624189496040344,
      "learning_rate": 1.800157297483417e-05,
      "loss": 11.7785,
      "step": 147
    },
    {
      "epoch": 0.027034432368252807,
      "grad_norm": 0.5138468146324158,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 11.8001,
      "step": 148
    },
    {
      "epoch": 0.027217097451822086,
      "grad_norm": 0.5513505935668945,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 11.8046,
      "step": 149
    },
    {
      "epoch": 0.02739976253539136,
      "grad_norm": 0.6666286587715149,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 11.792,
      "step": 150
    },
    {
      "epoch": 0.02739976253539136,
      "eval_loss": 11.825672149658203,
      "eval_runtime": 89.2187,
      "eval_samples_per_second": 103.342,
      "eval_steps_per_second": 25.835,
      "step": 150
    },
    {
      "epoch": 0.027582427618960636,
      "grad_norm": 0.3495938181877136,
      "learning_rate": 1.553232954407171e-05,
      "loss": 11.8697,
      "step": 151
    },
    {
      "epoch": 0.02776509270252991,
      "grad_norm": 0.3430638015270233,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 11.8565,
      "step": 152
    },
    {
      "epoch": 0.027947757786099186,
      "grad_norm": 0.34175825119018555,
      "learning_rate": 1.435357758543015e-05,
      "loss": 11.8635,
      "step": 153
    },
    {
      "epoch": 0.028130422869668464,
      "grad_norm": 0.35935842990875244,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 11.8654,
      "step": 154
    },
    {
      "epoch": 0.02831308795323774,
      "grad_norm": 0.3737039864063263,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 11.8585,
      "step": 155
    },
    {
      "epoch": 0.028495753036807014,
      "grad_norm": 0.29347941279411316,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 11.8373,
      "step": 156
    },
    {
      "epoch": 0.02867841812037629,
      "grad_norm": 0.29615211486816406,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 11.8576,
      "step": 157
    },
    {
      "epoch": 0.028861083203945564,
      "grad_norm": 0.32246923446655273,
      "learning_rate": 1.157994445715706e-05,
      "loss": 11.8538,
      "step": 158
    },
    {
      "epoch": 0.029043748287514843,
      "grad_norm": 0.345444917678833,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 11.8528,
      "step": 159
    },
    {
      "epoch": 0.029226413371084118,
      "grad_norm": 0.24580144882202148,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 11.8444,
      "step": 160
    },
    {
      "epoch": 0.029409078454653393,
      "grad_norm": 0.2688215970993042,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 11.8565,
      "step": 161
    },
    {
      "epoch": 0.029591743538222668,
      "grad_norm": 0.29417479038238525,
      "learning_rate": 9.549150281252633e-06,
      "loss": 11.8401,
      "step": 162
    },
    {
      "epoch": 0.029774408621791943,
      "grad_norm": 0.23299115896224976,
      "learning_rate": 9.068759265665384e-06,
      "loss": 11.848,
      "step": 163
    },
    {
      "epoch": 0.02995707370536122,
      "grad_norm": 0.2752237617969513,
      "learning_rate": 8.599558442598998e-06,
      "loss": 11.8445,
      "step": 164
    },
    {
      "epoch": 0.030139738788930497,
      "grad_norm": 0.2819494307041168,
      "learning_rate": 8.141676086873572e-06,
      "loss": 11.8446,
      "step": 165
    },
    {
      "epoch": 0.03032240387249977,
      "grad_norm": 0.26293766498565674,
      "learning_rate": 7.695237378953223e-06,
      "loss": 11.8396,
      "step": 166
    },
    {
      "epoch": 0.030505068956069047,
      "grad_norm": 0.1969415843486786,
      "learning_rate": 7.260364370723044e-06,
      "loss": 11.8386,
      "step": 167
    },
    {
      "epoch": 0.03068773403963832,
      "grad_norm": 0.2682812511920929,
      "learning_rate": 6.837175952121306e-06,
      "loss": 11.8322,
      "step": 168
    },
    {
      "epoch": 0.0308703991232076,
      "grad_norm": 0.2711208760738373,
      "learning_rate": 6.425787818636131e-06,
      "loss": 11.8396,
      "step": 169
    },
    {
      "epoch": 0.031053064206776875,
      "grad_norm": 0.4540082812309265,
      "learning_rate": 6.026312439675552e-06,
      "loss": 11.8263,
      "step": 170
    },
    {
      "epoch": 0.03123572929034615,
      "grad_norm": 0.30682945251464844,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 11.8191,
      "step": 171
    },
    {
      "epoch": 0.03141839437391543,
      "grad_norm": 0.2404305785894394,
      "learning_rate": 5.263533508961827e-06,
      "loss": 11.8502,
      "step": 172
    },
    {
      "epoch": 0.031601059457484704,
      "grad_norm": 0.29850706458091736,
      "learning_rate": 4.900438493352055e-06,
      "loss": 11.8245,
      "step": 173
    },
    {
      "epoch": 0.03178372454105398,
      "grad_norm": 0.2587142288684845,
      "learning_rate": 4.549673247541875e-06,
      "loss": 11.8296,
      "step": 174
    },
    {
      "epoch": 0.031966389624623254,
      "grad_norm": 0.28899532556533813,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 11.8397,
      "step": 175
    },
    {
      "epoch": 0.03214905470819253,
      "grad_norm": 0.24390612542629242,
      "learning_rate": 3.885512251130763e-06,
      "loss": 11.8154,
      "step": 176
    },
    {
      "epoch": 0.032331719791761804,
      "grad_norm": 0.24571648240089417,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 11.8296,
      "step": 177
    },
    {
      "epoch": 0.03251438487533108,
      "grad_norm": 0.2516442537307739,
      "learning_rate": 3.271776770026963e-06,
      "loss": 11.8169,
      "step": 178
    },
    {
      "epoch": 0.032697049958900354,
      "grad_norm": 0.23058979213237762,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 11.8165,
      "step": 179
    },
    {
      "epoch": 0.03287971504246963,
      "grad_norm": 0.2522067129611969,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 11.8079,
      "step": 180
    },
    {
      "epoch": 0.03306238012603891,
      "grad_norm": 0.20576074719429016,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 11.8267,
      "step": 181
    },
    {
      "epoch": 0.033245045209608186,
      "grad_norm": 0.2478885054588318,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 11.8202,
      "step": 182
    },
    {
      "epoch": 0.03342771029317746,
      "grad_norm": 0.3452244997024536,
      "learning_rate": 1.962316193157593e-06,
      "loss": 11.7945,
      "step": 183
    },
    {
      "epoch": 0.033610375376746736,
      "grad_norm": 0.26303645968437195,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 11.8056,
      "step": 184
    },
    {
      "epoch": 0.03379304046031601,
      "grad_norm": 0.3305163085460663,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 11.8004,
      "step": 185
    },
    {
      "epoch": 0.033975705543885286,
      "grad_norm": 0.24932773411273956,
      "learning_rate": 1.333670137599713e-06,
      "loss": 11.8157,
      "step": 186
    },
    {
      "epoch": 0.03415837062745456,
      "grad_norm": 0.3025909662246704,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 11.8113,
      "step": 187
    },
    {
      "epoch": 0.034341035711023836,
      "grad_norm": 0.27995580434799194,
      "learning_rate": 9.810017062595322e-07,
      "loss": 11.8078,
      "step": 188
    },
    {
      "epoch": 0.03452370079459311,
      "grad_norm": 0.3074042499065399,
      "learning_rate": 8.247462563808817e-07,
      "loss": 11.8229,
      "step": 189
    },
    {
      "epoch": 0.034706365878162386,
      "grad_norm": 0.25891295075416565,
      "learning_rate": 6.819348298638839e-07,
      "loss": 11.7957,
      "step": 190
    },
    {
      "epoch": 0.03488903096173167,
      "grad_norm": 0.3224402666091919,
      "learning_rate": 5.526064699265753e-07,
      "loss": 11.8218,
      "step": 191
    },
    {
      "epoch": 0.03507169604530094,
      "grad_norm": 0.31382831931114197,
      "learning_rate": 4.367965336512403e-07,
      "loss": 11.8085,
      "step": 192
    },
    {
      "epoch": 0.03525436112887022,
      "grad_norm": 0.4858558177947998,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 11.7745,
      "step": 193
    },
    {
      "epoch": 0.03543702621243949,
      "grad_norm": 0.38745400309562683,
      "learning_rate": 2.458548727494292e-07,
      "loss": 11.8126,
      "step": 194
    },
    {
      "epoch": 0.03561969129600877,
      "grad_norm": 0.4174445867538452,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 11.8239,
      "step": 195
    },
    {
      "epoch": 0.03580235637957804,
      "grad_norm": 0.43451353907585144,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 11.8275,
      "step": 196
    },
    {
      "epoch": 0.03598502146314732,
      "grad_norm": 0.4543963372707367,
      "learning_rate": 6.150154258476315e-08,
      "loss": 11.8013,
      "step": 197
    },
    {
      "epoch": 0.03616768654671659,
      "grad_norm": 0.4442960023880005,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 11.7934,
      "step": 198
    },
    {
      "epoch": 0.03635035163028587,
      "grad_norm": 0.5629929900169373,
      "learning_rate": 6.834750376549792e-09,
      "loss": 11.7847,
      "step": 199
    },
    {
      "epoch": 0.03653301671385514,
      "grad_norm": 0.7875304222106934,
      "learning_rate": 0.0,
      "loss": 11.7894,
      "step": 200
    },
    {
      "epoch": 0.03653301671385514,
      "eval_loss": 11.824332237243652,
      "eval_runtime": 89.3025,
      "eval_samples_per_second": 103.245,
      "eval_steps_per_second": 25.811,
      "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": 220358246400000.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}