{
  "best_metric": 0.0012758667580783367,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3225806451612903,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016129032258064516,
      "grad_norm": 0.6049064993858337,
      "learning_rate": 1e-05,
      "loss": 0.9039,
      "step": 1
    },
    {
      "epoch": 0.0016129032258064516,
      "eval_loss": 0.7379059195518494,
      "eval_runtime": 106.5243,
      "eval_samples_per_second": 9.801,
      "eval_steps_per_second": 2.45,
      "step": 1
    },
    {
      "epoch": 0.0032258064516129032,
      "grad_norm": 0.7415255904197693,
      "learning_rate": 2e-05,
      "loss": 1.1299,
      "step": 2
    },
    {
      "epoch": 0.004838709677419355,
      "grad_norm": 0.8382638692855835,
      "learning_rate": 3e-05,
      "loss": 1.2151,
      "step": 3
    },
    {
      "epoch": 0.0064516129032258064,
      "grad_norm": 0.862156331539154,
      "learning_rate": 4e-05,
      "loss": 1.2969,
      "step": 4
    },
    {
      "epoch": 0.008064516129032258,
      "grad_norm": 0.9778095483779907,
      "learning_rate": 5e-05,
      "loss": 1.2646,
      "step": 5
    },
    {
      "epoch": 0.00967741935483871,
      "grad_norm": 0.9458996057510376,
      "learning_rate": 6e-05,
      "loss": 1.2015,
      "step": 6
    },
    {
      "epoch": 0.01129032258064516,
      "grad_norm": 0.981987476348877,
      "learning_rate": 7e-05,
      "loss": 1.0207,
      "step": 7
    },
    {
      "epoch": 0.012903225806451613,
      "grad_norm": 0.8597192764282227,
      "learning_rate": 8e-05,
      "loss": 0.7901,
      "step": 8
    },
    {
      "epoch": 0.014516129032258065,
      "grad_norm": 0.9466065764427185,
      "learning_rate": 9e-05,
      "loss": 0.7297,
      "step": 9
    },
    {
      "epoch": 0.016129032258064516,
      "grad_norm": 1.1080763339996338,
      "learning_rate": 0.0001,
      "loss": 0.4713,
      "step": 10
    },
    {
      "epoch": 0.017741935483870968,
      "grad_norm": 0.8401578068733215,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.2136,
      "step": 11
    },
    {
      "epoch": 0.01935483870967742,
      "grad_norm": 0.9712587594985962,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.2193,
      "step": 12
    },
    {
      "epoch": 0.020967741935483872,
      "grad_norm": 1.2566802501678467,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.2793,
      "step": 13
    },
    {
      "epoch": 0.02258064516129032,
      "grad_norm": 0.9939321875572205,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.1432,
      "step": 14
    },
    {
      "epoch": 0.024193548387096774,
      "grad_norm": 1.6880418062210083,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.1768,
      "step": 15
    },
    {
      "epoch": 0.025806451612903226,
      "grad_norm": 1.028950810432434,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.1185,
      "step": 16
    },
    {
      "epoch": 0.027419354838709678,
      "grad_norm": 0.7586027383804321,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.0659,
      "step": 17
    },
    {
      "epoch": 0.02903225806451613,
      "grad_norm": 0.9536590576171875,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.072,
      "step": 18
    },
    {
      "epoch": 0.03064516129032258,
      "grad_norm": 0.7935003638267517,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.0533,
      "step": 19
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 0.9564946293830872,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.0539,
      "step": 20
    },
    {
      "epoch": 0.03387096774193549,
      "grad_norm": 0.6796997785568237,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.0269,
      "step": 21
    },
    {
      "epoch": 0.035483870967741936,
      "grad_norm": 1.2710872888565063,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.1013,
      "step": 22
    },
    {
      "epoch": 0.037096774193548385,
      "grad_norm": 0.5349439382553101,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.0248,
      "step": 23
    },
    {
      "epoch": 0.03870967741935484,
      "grad_norm": 1.372443437576294,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.1846,
      "step": 24
    },
    {
      "epoch": 0.04032258064516129,
      "grad_norm": 0.7447899580001831,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.0277,
      "step": 25
    },
    {
      "epoch": 0.041935483870967745,
      "grad_norm": 0.2933860719203949,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.0103,
      "step": 26
    },
    {
      "epoch": 0.043548387096774194,
      "grad_norm": 1.2192440032958984,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.0694,
      "step": 27
    },
    {
      "epoch": 0.04516129032258064,
      "grad_norm": 1.3143314123153687,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.0395,
      "step": 28
    },
    {
      "epoch": 0.0467741935483871,
      "grad_norm": 0.8509743213653564,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.1224,
      "step": 29
    },
    {
      "epoch": 0.04838709677419355,
      "grad_norm": 0.9023564457893372,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.0558,
      "step": 30
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3841659128665924,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.0202,
      "step": 31
    },
    {
      "epoch": 0.05161290322580645,
      "grad_norm": 2.0831139087677,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.0486,
      "step": 32
    },
    {
      "epoch": 0.0532258064516129,
      "grad_norm": 0.3329290747642517,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.0155,
      "step": 33
    },
    {
      "epoch": 0.054838709677419356,
      "grad_norm": 0.3646816313266754,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.0219,
      "step": 34
    },
    {
      "epoch": 0.056451612903225805,
      "grad_norm": 0.5079269409179688,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.0392,
      "step": 35
    },
    {
      "epoch": 0.05806451612903226,
      "grad_norm": 1.1356701850891113,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.0314,
      "step": 36
    },
    {
      "epoch": 0.05967741935483871,
      "grad_norm": 0.7165648341178894,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.0622,
      "step": 37
    },
    {
      "epoch": 0.06129032258064516,
      "grad_norm": 0.9975515604019165,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.0386,
      "step": 38
    },
    {
      "epoch": 0.06290322580645161,
      "grad_norm": 0.5520864725112915,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.0421,
      "step": 39
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 0.6942296028137207,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.0219,
      "step": 40
    },
    {
      "epoch": 0.06612903225806452,
      "grad_norm": 1.3481652736663818,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.143,
      "step": 41
    },
    {
      "epoch": 0.06774193548387097,
      "grad_norm": 2.7365214824676514,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.0288,
      "step": 42
    },
    {
      "epoch": 0.06935483870967742,
      "grad_norm": 0.2986465394496918,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.0081,
      "step": 43
    },
    {
      "epoch": 0.07096774193548387,
      "grad_norm": 0.42408058047294617,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.013,
      "step": 44
    },
    {
      "epoch": 0.07258064516129033,
      "grad_norm": 1.1357313394546509,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.0186,
      "step": 45
    },
    {
      "epoch": 0.07419354838709677,
      "grad_norm": 0.9855334162712097,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.0275,
      "step": 46
    },
    {
      "epoch": 0.07580645161290323,
      "grad_norm": 0.14201870560646057,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.0036,
      "step": 47
    },
    {
      "epoch": 0.07741935483870968,
      "grad_norm": 1.7319813966751099,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.0958,
      "step": 48
    },
    {
      "epoch": 0.07903225806451612,
      "grad_norm": 1.869276523590088,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.0504,
      "step": 49
    },
    {
      "epoch": 0.08064516129032258,
      "grad_norm": 1.4869447946548462,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.0542,
      "step": 50
    },
    {
      "epoch": 0.08064516129032258,
      "eval_loss": 0.006671774201095104,
      "eval_runtime": 107.2911,
      "eval_samples_per_second": 9.731,
      "eval_steps_per_second": 2.433,
      "step": 50
    },
    {
      "epoch": 0.08225806451612903,
      "grad_norm": 0.2840280532836914,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.0103,
      "step": 51
    },
    {
      "epoch": 0.08387096774193549,
      "grad_norm": 0.15522587299346924,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.0158,
      "step": 52
    },
    {
      "epoch": 0.08548387096774193,
      "grad_norm": 0.3222741186618805,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.0205,
      "step": 53
    },
    {
      "epoch": 0.08709677419354839,
      "grad_norm": 0.33048781752586365,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.0087,
      "step": 54
    },
    {
      "epoch": 0.08870967741935484,
      "grad_norm": 0.2995803952217102,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.0403,
      "step": 55
    },
    {
      "epoch": 0.09032258064516129,
      "grad_norm": 0.08299575001001358,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.0036,
      "step": 56
    },
    {
      "epoch": 0.09193548387096774,
      "grad_norm": 0.32099059224128723,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.0155,
      "step": 57
    },
    {
      "epoch": 0.0935483870967742,
      "grad_norm": 0.34219861030578613,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.0486,
      "step": 58
    },
    {
      "epoch": 0.09516129032258064,
      "grad_norm": 0.26563310623168945,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.0367,
      "step": 59
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 0.21417857706546783,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.0111,
      "step": 60
    },
    {
      "epoch": 0.09838709677419355,
      "grad_norm": 0.2877480089664459,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.0081,
      "step": 61
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2827915549278259,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.0205,
      "step": 62
    },
    {
      "epoch": 0.10161290322580645,
      "grad_norm": 0.33711934089660645,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.0171,
      "step": 63
    },
    {
      "epoch": 0.1032258064516129,
      "grad_norm": 0.34038734436035156,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.0216,
      "step": 64
    },
    {
      "epoch": 0.10483870967741936,
      "grad_norm": 0.46213850378990173,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.0312,
      "step": 65
    },
    {
      "epoch": 0.1064516129032258,
      "grad_norm": 0.4424307346343994,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.0291,
      "step": 66
    },
    {
      "epoch": 0.10806451612903226,
      "grad_norm": 0.47928720712661743,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.0603,
      "step": 67
    },
    {
      "epoch": 0.10967741935483871,
      "grad_norm": 0.6958261132240295,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.0796,
      "step": 68
    },
    {
      "epoch": 0.11129032258064517,
      "grad_norm": 0.059392184019088745,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.0037,
      "step": 69
    },
    {
      "epoch": 0.11290322580645161,
      "grad_norm": 0.6522670388221741,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.0313,
      "step": 70
    },
    {
      "epoch": 0.11451612903225807,
      "grad_norm": 0.27442312240600586,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.0173,
      "step": 71
    },
    {
      "epoch": 0.11612903225806452,
      "grad_norm": 0.20664764940738678,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.0071,
      "step": 72
    },
    {
      "epoch": 0.11774193548387096,
      "grad_norm": 0.28357335925102234,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.0129,
      "step": 73
    },
    {
      "epoch": 0.11935483870967742,
      "grad_norm": 0.3043783903121948,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.0293,
      "step": 74
    },
    {
      "epoch": 0.12096774193548387,
      "grad_norm": 0.12785613536834717,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.0058,
      "step": 75
    },
    {
      "epoch": 0.12258064516129032,
      "grad_norm": 0.12152642756700516,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.0045,
      "step": 76
    },
    {
      "epoch": 0.12419354838709677,
      "grad_norm": 0.4066691994667053,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.0145,
      "step": 77
    },
    {
      "epoch": 0.12580645161290321,
      "grad_norm": 0.65605628490448,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.0223,
      "step": 78
    },
    {
      "epoch": 0.12741935483870967,
      "grad_norm": 0.5198997855186462,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.0116,
      "step": 79
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 0.6631190180778503,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.0243,
      "step": 80
    },
    {
      "epoch": 0.13064516129032258,
      "grad_norm": 0.729720950126648,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.0409,
      "step": 81
    },
    {
      "epoch": 0.13225806451612904,
      "grad_norm": 0.0241117961704731,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.0013,
      "step": 82
    },
    {
      "epoch": 0.1338709677419355,
      "grad_norm": 0.09967964142560959,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.0023,
      "step": 83
    },
    {
      "epoch": 0.13548387096774195,
      "grad_norm": 0.43611156940460205,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.0282,
      "step": 84
    },
    {
      "epoch": 0.13709677419354838,
      "grad_norm": 1.0305331945419312,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.0838,
      "step": 85
    },
    {
      "epoch": 0.13870967741935483,
      "grad_norm": 1.1009966135025024,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.0768,
      "step": 86
    },
    {
      "epoch": 0.1403225806451613,
      "grad_norm": 0.5247541069984436,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.0225,
      "step": 87
    },
    {
      "epoch": 0.14193548387096774,
      "grad_norm": 0.33542248606681824,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.0189,
      "step": 88
    },
    {
      "epoch": 0.1435483870967742,
      "grad_norm": 0.037109486758708954,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.0015,
      "step": 89
    },
    {
      "epoch": 0.14516129032258066,
      "grad_norm": 1.4759800434112549,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.0169,
      "step": 90
    },
    {
      "epoch": 0.14677419354838708,
      "grad_norm": 0.24253155291080475,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.0078,
      "step": 91
    },
    {
      "epoch": 0.14838709677419354,
      "grad_norm": 1.7516717910766602,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.0455,
      "step": 92
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1160598993301392,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.0299,
      "step": 93
    },
    {
      "epoch": 0.15161290322580645,
      "grad_norm": 0.13985562324523926,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.0035,
      "step": 94
    },
    {
      "epoch": 0.1532258064516129,
      "grad_norm": 0.12249292433261871,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.0041,
      "step": 95
    },
    {
      "epoch": 0.15483870967741936,
      "grad_norm": 0.24244475364685059,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.0022,
      "step": 96
    },
    {
      "epoch": 0.15645161290322582,
      "grad_norm": 0.06717562675476074,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.0027,
      "step": 97
    },
    {
      "epoch": 0.15806451612903225,
      "grad_norm": 1.463485836982727,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.0626,
      "step": 98
    },
    {
      "epoch": 0.1596774193548387,
      "grad_norm": 0.30929034948349,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.0062,
      "step": 99
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 1.0688246488571167,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.0313,
      "step": 100
    },
    {
      "epoch": 0.16129032258064516,
      "eval_loss": 0.0023250828962773085,
      "eval_runtime": 107.3887,
      "eval_samples_per_second": 9.722,
      "eval_steps_per_second": 2.43,
      "step": 100
    },
    {
      "epoch": 0.1629032258064516,
      "grad_norm": 0.2973756194114685,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.0151,
      "step": 101
    },
    {
      "epoch": 0.16451612903225807,
      "grad_norm": 0.23198020458221436,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.0082,
      "step": 102
    },
    {
      "epoch": 0.16612903225806452,
      "grad_norm": 0.22707688808441162,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.0071,
      "step": 103
    },
    {
      "epoch": 0.16774193548387098,
      "grad_norm": 0.1593625545501709,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.0158,
      "step": 104
    },
    {
      "epoch": 0.1693548387096774,
      "grad_norm": 0.17460094392299652,
      "learning_rate": 5e-05,
      "loss": 0.0057,
      "step": 105
    },
    {
      "epoch": 0.17096774193548386,
      "grad_norm": 0.18611103296279907,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.0036,
      "step": 106
    },
    {
      "epoch": 0.17258064516129032,
      "grad_norm": 0.22693701088428497,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.0074,
      "step": 107
    },
    {
      "epoch": 0.17419354838709677,
      "grad_norm": 0.06350814551115036,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.0019,
      "step": 108
    },
    {
      "epoch": 0.17580645161290323,
      "grad_norm": 0.17770349979400635,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.0087,
      "step": 109
    },
    {
      "epoch": 0.1774193548387097,
      "grad_norm": 0.30756187438964844,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.0195,
      "step": 110
    },
    {
      "epoch": 0.17903225806451614,
      "grad_norm": 0.2772008776664734,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.0212,
      "step": 111
    },
    {
      "epoch": 0.18064516129032257,
      "grad_norm": 0.24916262924671173,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.0133,
      "step": 112
    },
    {
      "epoch": 0.18225806451612903,
      "grad_norm": 0.2733086943626404,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.0181,
      "step": 113
    },
    {
      "epoch": 0.18387096774193548,
      "grad_norm": 0.41303563117980957,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.0327,
      "step": 114
    },
    {
      "epoch": 0.18548387096774194,
      "grad_norm": 0.24434718489646912,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.0225,
      "step": 115
    },
    {
      "epoch": 0.1870967741935484,
      "grad_norm": 0.1770666241645813,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.009,
      "step": 116
    },
    {
      "epoch": 0.18870967741935485,
      "grad_norm": 0.1502278596162796,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.0056,
      "step": 117
    },
    {
      "epoch": 0.19032258064516128,
      "grad_norm": 0.21350064873695374,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.0059,
      "step": 118
    },
    {
      "epoch": 0.19193548387096773,
      "grad_norm": 0.2722512483596802,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.0289,
      "step": 119
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 0.07606415450572968,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.0031,
      "step": 120
    },
    {
      "epoch": 0.19516129032258064,
      "grad_norm": 0.21624216437339783,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.0092,
      "step": 121
    },
    {
      "epoch": 0.1967741935483871,
      "grad_norm": 0.11716532707214355,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.0034,
      "step": 122
    },
    {
      "epoch": 0.19838709677419356,
      "grad_norm": 0.45593780279159546,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.026,
      "step": 123
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3051721453666687,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.0114,
      "step": 124
    },
    {
      "epoch": 0.20161290322580644,
      "grad_norm": 0.549540102481842,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.0093,
      "step": 125
    },
    {
      "epoch": 0.2032258064516129,
      "grad_norm": 0.11999449133872986,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.0033,
      "step": 126
    },
    {
      "epoch": 0.20483870967741935,
      "grad_norm": 0.1965821236371994,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.0054,
      "step": 127
    },
    {
      "epoch": 0.2064516129032258,
      "grad_norm": 0.10459789633750916,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.0036,
      "step": 128
    },
    {
      "epoch": 0.20806451612903226,
      "grad_norm": 0.07582688331604004,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.0028,
      "step": 129
    },
    {
      "epoch": 0.20967741935483872,
      "grad_norm": 0.32175418734550476,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.0099,
      "step": 130
    },
    {
      "epoch": 0.21129032258064517,
      "grad_norm": 0.3683537244796753,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.0193,
      "step": 131
    },
    {
      "epoch": 0.2129032258064516,
      "grad_norm": 0.08368095010519028,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.002,
      "step": 132
    },
    {
      "epoch": 0.21451612903225806,
      "grad_norm": 0.04766213521361351,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.0014,
      "step": 133
    },
    {
      "epoch": 0.2161290322580645,
      "grad_norm": 0.37475845217704773,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.0354,
      "step": 134
    },
    {
      "epoch": 0.21774193548387097,
      "grad_norm": 0.023638686165213585,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.0011,
      "step": 135
    },
    {
      "epoch": 0.21935483870967742,
      "grad_norm": 0.8545042276382446,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.0142,
      "step": 136
    },
    {
      "epoch": 0.22096774193548388,
      "grad_norm": 0.02345900982618332,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.001,
      "step": 137
    },
    {
      "epoch": 0.22258064516129034,
      "grad_norm": 0.05260935798287392,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.0017,
      "step": 138
    },
    {
      "epoch": 0.22419354838709676,
      "grad_norm": 0.3367604911327362,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0132,
      "step": 139
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 0.07563036680221558,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.0023,
      "step": 140
    },
    {
      "epoch": 0.22741935483870968,
      "grad_norm": 0.199149489402771,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.0039,
      "step": 141
    },
    {
      "epoch": 0.22903225806451613,
      "grad_norm": 0.05658286437392235,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.0018,
      "step": 142
    },
    {
      "epoch": 0.2306451612903226,
      "grad_norm": 0.11913429945707321,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0031,
      "step": 143
    },
    {
      "epoch": 0.23225806451612904,
      "grad_norm": 0.8042193651199341,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0482,
      "step": 144
    },
    {
      "epoch": 0.23387096774193547,
      "grad_norm": 0.02670295350253582,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.001,
      "step": 145
    },
    {
      "epoch": 0.23548387096774193,
      "grad_norm": 0.1480982005596161,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.0036,
      "step": 146
    },
    {
      "epoch": 0.23709677419354838,
      "grad_norm": 3.5535287857055664,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.1386,
      "step": 147
    },
    {
      "epoch": 0.23870967741935484,
      "grad_norm": 0.1271081119775772,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.0021,
      "step": 148
    },
    {
      "epoch": 0.2403225806451613,
      "grad_norm": 2.022695779800415,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.0577,
      "step": 149
    },
    {
      "epoch": 0.24193548387096775,
      "grad_norm": 3.139491319656372,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.0528,
      "step": 150
    },
    {
      "epoch": 0.24193548387096775,
      "eval_loss": 0.0014273119159042835,
      "eval_runtime": 107.3465,
      "eval_samples_per_second": 9.726,
      "eval_steps_per_second": 2.431,
      "step": 150
    },
    {
      "epoch": 0.2435483870967742,
      "grad_norm": 0.19909407198429108,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.0062,
      "step": 151
    },
    {
      "epoch": 0.24516129032258063,
      "grad_norm": 0.41273730993270874,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.0201,
      "step": 152
    },
    {
      "epoch": 0.2467741935483871,
      "grad_norm": 0.11431430280208588,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.0041,
      "step": 153
    },
    {
      "epoch": 0.24838709677419354,
      "grad_norm": 0.11579876393079758,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.0057,
      "step": 154
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.05266593024134636,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.0018,
      "step": 155
    },
    {
      "epoch": 0.25161290322580643,
      "grad_norm": 0.1429545283317566,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.0035,
      "step": 156
    },
    {
      "epoch": 0.2532258064516129,
      "grad_norm": 0.30217474699020386,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.0254,
      "step": 157
    },
    {
      "epoch": 0.25483870967741934,
      "grad_norm": 0.6804100871086121,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.0499,
      "step": 158
    },
    {
      "epoch": 0.2564516129032258,
      "grad_norm": 0.09390906244516373,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.0038,
      "step": 159
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 0.13369528949260712,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.0022,
      "step": 160
    },
    {
      "epoch": 0.25967741935483873,
      "grad_norm": 0.05726723000407219,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.0013,
      "step": 161
    },
    {
      "epoch": 0.26129032258064516,
      "grad_norm": 0.1694510132074356,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0047,
      "step": 162
    },
    {
      "epoch": 0.2629032258064516,
      "grad_norm": 0.10400049388408661,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.0021,
      "step": 163
    },
    {
      "epoch": 0.2645161290322581,
      "grad_norm": 0.13057760894298553,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.0052,
      "step": 164
    },
    {
      "epoch": 0.2661290322580645,
      "grad_norm": 0.1514163613319397,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.008,
      "step": 165
    },
    {
      "epoch": 0.267741935483871,
      "grad_norm": 0.023335611447691917,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.001,
      "step": 166
    },
    {
      "epoch": 0.2693548387096774,
      "grad_norm": 0.4413110613822937,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.0132,
      "step": 167
    },
    {
      "epoch": 0.2709677419354839,
      "grad_norm": 0.1287180483341217,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.0047,
      "step": 168
    },
    {
      "epoch": 0.2725806451612903,
      "grad_norm": 0.19662626087665558,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.0125,
      "step": 169
    },
    {
      "epoch": 0.27419354838709675,
      "grad_norm": 0.01794038526713848,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.0009,
      "step": 170
    },
    {
      "epoch": 0.27580645161290324,
      "grad_norm": 0.13840854167938232,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.007,
      "step": 171
    },
    {
      "epoch": 0.27741935483870966,
      "grad_norm": 0.053238630294799805,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.002,
      "step": 172
    },
    {
      "epoch": 0.27903225806451615,
      "grad_norm": 0.24817954003810883,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.0071,
      "step": 173
    },
    {
      "epoch": 0.2806451612903226,
      "grad_norm": 0.014834782108664513,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.0008,
      "step": 174
    },
    {
      "epoch": 0.28225806451612906,
      "grad_norm": 0.549480676651001,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.0184,
      "step": 175
    },
    {
      "epoch": 0.2838709677419355,
      "grad_norm": 0.0825066789984703,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.0028,
      "step": 176
    },
    {
      "epoch": 0.2854838709677419,
      "grad_norm": 0.6054806113243103,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.0113,
      "step": 177
    },
    {
      "epoch": 0.2870967741935484,
      "grad_norm": 0.3656103014945984,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.028,
      "step": 178
    },
    {
      "epoch": 0.2887096774193548,
      "grad_norm": 0.14775890111923218,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0038,
      "step": 179
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 0.5688266754150391,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.0363,
      "step": 180
    },
    {
      "epoch": 0.29193548387096774,
      "grad_norm": 0.07384208589792252,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.0035,
      "step": 181
    },
    {
      "epoch": 0.29354838709677417,
      "grad_norm": 0.5008048415184021,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.0153,
      "step": 182
    },
    {
      "epoch": 0.29516129032258065,
      "grad_norm": 0.2678474485874176,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.003,
      "step": 183
    },
    {
      "epoch": 0.2967741935483871,
      "grad_norm": 0.050162795931100845,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.0013,
      "step": 184
    },
    {
      "epoch": 0.29838709677419356,
      "grad_norm": 0.6895274519920349,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.0614,
      "step": 185
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.03697023168206215,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.001,
      "step": 186
    },
    {
      "epoch": 0.3016129032258065,
      "grad_norm": 0.1690882444381714,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0045,
      "step": 187
    },
    {
      "epoch": 0.3032258064516129,
      "grad_norm": 0.010449085384607315,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.0005,
      "step": 188
    },
    {
      "epoch": 0.30483870967741933,
      "grad_norm": 0.8781279921531677,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.0285,
      "step": 189
    },
    {
      "epoch": 0.3064516129032258,
      "grad_norm": 0.06028104946017265,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0017,
      "step": 190
    },
    {
      "epoch": 0.30806451612903224,
      "grad_norm": 1.045876145362854,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.0334,
      "step": 191
    },
    {
      "epoch": 0.3096774193548387,
      "grad_norm": 0.7879181504249573,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0104,
      "step": 192
    },
    {
      "epoch": 0.31129032258064515,
      "grad_norm": 0.011220087297260761,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.0006,
      "step": 193
    },
    {
      "epoch": 0.31290322580645163,
      "grad_norm": 0.06424606591463089,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0021,
      "step": 194
    },
    {
      "epoch": 0.31451612903225806,
      "grad_norm": 1.546553134918213,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.0734,
      "step": 195
    },
    {
      "epoch": 0.3161290322580645,
      "grad_norm": 0.46784070134162903,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.005,
      "step": 196
    },
    {
      "epoch": 0.317741935483871,
      "grad_norm": 0.3088896572589874,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0076,
      "step": 197
    },
    {
      "epoch": 0.3193548387096774,
      "grad_norm": 3.3314967155456543,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.0574,
      "step": 198
    },
    {
      "epoch": 0.3209677419354839,
      "grad_norm": 0.2016996443271637,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0066,
      "step": 199
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 0.16793261468410492,
      "learning_rate": 0.0,
      "loss": 0.004,
      "step": 200
    },
    {
      "epoch": 0.3225806451612903,
      "eval_loss": 0.0012758667580783367,
      "eval_runtime": 107.252,
      "eval_samples_per_second": 9.734,
      "eval_steps_per_second": 2.434,
      "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": 5.09187566075904e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}