{
  "best_metric": 0.05885966122150421,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.5267175572519083,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007633587786259542,
      "grad_norm": 2.390812873840332,
      "learning_rate": 1e-05,
      "loss": 2.8009,
      "step": 1
    },
    {
      "epoch": 0.007633587786259542,
      "eval_loss": 2.893526077270508,
      "eval_runtime": 32.5559,
      "eval_samples_per_second": 6.788,
      "eval_steps_per_second": 1.72,
      "step": 1
    },
    {
      "epoch": 0.015267175572519083,
      "grad_norm": 2.3168396949768066,
      "learning_rate": 2e-05,
      "loss": 2.8226,
      "step": 2
    },
    {
      "epoch": 0.022900763358778626,
      "grad_norm": 2.1638612747192383,
      "learning_rate": 3e-05,
      "loss": 2.8413,
      "step": 3
    },
    {
      "epoch": 0.030534351145038167,
      "grad_norm": 1.8229483366012573,
      "learning_rate": 4e-05,
      "loss": 2.7832,
      "step": 4
    },
    {
      "epoch": 0.03816793893129771,
      "grad_norm": 1.1173290014266968,
      "learning_rate": 5e-05,
      "loss": 2.7285,
      "step": 5
    },
    {
      "epoch": 0.04580152671755725,
      "grad_norm": 1.2543779611587524,
      "learning_rate": 6e-05,
      "loss": 2.6482,
      "step": 6
    },
    {
      "epoch": 0.05343511450381679,
      "grad_norm": 1.554586410522461,
      "learning_rate": 7e-05,
      "loss": 2.7165,
      "step": 7
    },
    {
      "epoch": 0.061068702290076333,
      "grad_norm": 0.8699727654457092,
      "learning_rate": 8e-05,
      "loss": 2.5403,
      "step": 8
    },
    {
      "epoch": 0.06870229007633588,
      "grad_norm": 0.5670621991157532,
      "learning_rate": 9e-05,
      "loss": 2.5952,
      "step": 9
    },
    {
      "epoch": 0.07633587786259542,
      "grad_norm": 0.7762729525566101,
      "learning_rate": 0.0001,
      "loss": 2.5244,
      "step": 10
    },
    {
      "epoch": 0.08396946564885496,
      "grad_norm": 0.6237712502479553,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.5451,
      "step": 11
    },
    {
      "epoch": 0.0916030534351145,
      "grad_norm": 0.4845149517059326,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.505,
      "step": 12
    },
    {
      "epoch": 0.09923664122137404,
      "grad_norm": 0.6282029151916504,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.5013,
      "step": 13
    },
    {
      "epoch": 0.10687022900763359,
      "grad_norm": 0.5899912118911743,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.4252,
      "step": 14
    },
    {
      "epoch": 0.11450381679389313,
      "grad_norm": 0.4619622528553009,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.4597,
      "step": 15
    },
    {
      "epoch": 0.12213740458015267,
      "grad_norm": 0.37971362471580505,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.4744,
      "step": 16
    },
    {
      "epoch": 0.1297709923664122,
      "grad_norm": 0.39914175868034363,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.423,
      "step": 17
    },
    {
      "epoch": 0.13740458015267176,
      "grad_norm": 0.3970184028148651,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.4461,
      "step": 18
    },
    {
      "epoch": 0.1450381679389313,
      "grad_norm": 0.4032096564769745,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.373,
      "step": 19
    },
    {
      "epoch": 0.15267175572519084,
      "grad_norm": 0.40481388568878174,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.395,
      "step": 20
    },
    {
      "epoch": 0.16030534351145037,
      "grad_norm": 0.3778122663497925,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.2901,
      "step": 21
    },
    {
      "epoch": 0.16793893129770993,
      "grad_norm": 0.42610296607017517,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.3959,
      "step": 22
    },
    {
      "epoch": 0.17557251908396945,
      "grad_norm": 0.4685070216655731,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.2855,
      "step": 23
    },
    {
      "epoch": 0.183206106870229,
      "grad_norm": 0.3950805366039276,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.3527,
      "step": 24
    },
    {
      "epoch": 0.19083969465648856,
      "grad_norm": 0.3592756688594818,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.4257,
      "step": 25
    },
    {
      "epoch": 0.1984732824427481,
      "grad_norm": 0.3844394385814667,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.3053,
      "step": 26
    },
    {
      "epoch": 0.20610687022900764,
      "grad_norm": 0.38410964608192444,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.2435,
      "step": 27
    },
    {
      "epoch": 0.21374045801526717,
      "grad_norm": 0.375165730714798,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.3184,
      "step": 28
    },
    {
      "epoch": 0.22137404580152673,
      "grad_norm": 0.36574894189834595,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.2301,
      "step": 29
    },
    {
      "epoch": 0.22900763358778625,
      "grad_norm": 0.39953410625457764,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.2138,
      "step": 30
    },
    {
      "epoch": 0.2366412213740458,
      "grad_norm": 0.3939075767993927,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.2083,
      "step": 31
    },
    {
      "epoch": 0.24427480916030533,
      "grad_norm": 0.40324828028678894,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.2283,
      "step": 32
    },
    {
      "epoch": 0.25190839694656486,
      "grad_norm": 0.5142034888267517,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.1121,
      "step": 33
    },
    {
      "epoch": 0.2595419847328244,
      "grad_norm": 0.4672495722770691,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.1553,
      "step": 34
    },
    {
      "epoch": 0.26717557251908397,
      "grad_norm": 0.45975813269615173,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.0679,
      "step": 35
    },
    {
      "epoch": 0.2748091603053435,
      "grad_norm": 0.4338775873184204,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.162,
      "step": 36
    },
    {
      "epoch": 0.2824427480916031,
      "grad_norm": 0.48687753081321716,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.0547,
      "step": 37
    },
    {
      "epoch": 0.2900763358778626,
      "grad_norm": 0.4537368714809418,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.1289,
      "step": 38
    },
    {
      "epoch": 0.29770992366412213,
      "grad_norm": 0.5074473023414612,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.9662,
      "step": 39
    },
    {
      "epoch": 0.3053435114503817,
      "grad_norm": 0.5017813444137573,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.9752,
      "step": 40
    },
    {
      "epoch": 0.31297709923664124,
      "grad_norm": 0.7643575668334961,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.9955,
      "step": 41
    },
    {
      "epoch": 0.32061068702290074,
      "grad_norm": 0.6210652589797974,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.9334,
      "step": 42
    },
    {
      "epoch": 0.3282442748091603,
      "grad_norm": 0.5529518127441406,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.9448,
      "step": 43
    },
    {
      "epoch": 0.33587786259541985,
      "grad_norm": 0.6030047535896301,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.94,
      "step": 44
    },
    {
      "epoch": 0.3435114503816794,
      "grad_norm": 0.6213891506195068,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.7824,
      "step": 45
    },
    {
      "epoch": 0.3511450381679389,
      "grad_norm": 0.6524145603179932,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.903,
      "step": 46
    },
    {
      "epoch": 0.35877862595419846,
      "grad_norm": 0.6205057501792908,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.8396,
      "step": 47
    },
    {
      "epoch": 0.366412213740458,
      "grad_norm": 0.6533191204071045,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.7249,
      "step": 48
    },
    {
      "epoch": 0.37404580152671757,
      "grad_norm": 0.7670183777809143,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.6989,
      "step": 49
    },
    {
      "epoch": 0.3816793893129771,
      "grad_norm": 0.8481380939483643,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.7715,
      "step": 50
    },
    {
      "epoch": 0.3816793893129771,
      "eval_loss": 1.7358461618423462,
      "eval_runtime": 33.1013,
      "eval_samples_per_second": 6.676,
      "eval_steps_per_second": 1.692,
      "step": 50
    },
    {
      "epoch": 0.3893129770992366,
      "grad_norm": 0.7989552617073059,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.6829,
      "step": 51
    },
    {
      "epoch": 0.3969465648854962,
      "grad_norm": 0.7253351807594299,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.6634,
      "step": 52
    },
    {
      "epoch": 0.40458015267175573,
      "grad_norm": 1.0601885318756104,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.6526,
      "step": 53
    },
    {
      "epoch": 0.4122137404580153,
      "grad_norm": 0.9982365369796753,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.5821,
      "step": 54
    },
    {
      "epoch": 0.4198473282442748,
      "grad_norm": 0.7914581894874573,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.7089,
      "step": 55
    },
    {
      "epoch": 0.42748091603053434,
      "grad_norm": 0.8627219200134277,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.7088,
      "step": 56
    },
    {
      "epoch": 0.4351145038167939,
      "grad_norm": 0.9394534826278687,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4807,
      "step": 57
    },
    {
      "epoch": 0.44274809160305345,
      "grad_norm": 1.6982685327529907,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.5971,
      "step": 58
    },
    {
      "epoch": 0.45038167938931295,
      "grad_norm": 1.324668049812317,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.5522,
      "step": 59
    },
    {
      "epoch": 0.4580152671755725,
      "grad_norm": 0.9600746035575867,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.4259,
      "step": 60
    },
    {
      "epoch": 0.46564885496183206,
      "grad_norm": 1.0156992673873901,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.595,
      "step": 61
    },
    {
      "epoch": 0.4732824427480916,
      "grad_norm": 0.8910471796989441,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.3821,
      "step": 62
    },
    {
      "epoch": 0.48091603053435117,
      "grad_norm": 0.9320412874221802,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.5287,
      "step": 63
    },
    {
      "epoch": 0.48854961832061067,
      "grad_norm": 1.2646182775497437,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4282,
      "step": 64
    },
    {
      "epoch": 0.4961832061068702,
      "grad_norm": 1.2024239301681519,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.4227,
      "step": 65
    },
    {
      "epoch": 0.5038167938931297,
      "grad_norm": 0.9754292368888855,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.4681,
      "step": 66
    },
    {
      "epoch": 0.5114503816793893,
      "grad_norm": 1.1136093139648438,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.2851,
      "step": 67
    },
    {
      "epoch": 0.5190839694656488,
      "grad_norm": 1.1642425060272217,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.319,
      "step": 68
    },
    {
      "epoch": 0.5267175572519084,
      "grad_norm": 1.0689488649368286,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.421,
      "step": 69
    },
    {
      "epoch": 0.5343511450381679,
      "grad_norm": 1.157355785369873,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.3123,
      "step": 70
    },
    {
      "epoch": 0.5419847328244275,
      "grad_norm": 0.9805869460105896,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.2074,
      "step": 71
    },
    {
      "epoch": 0.549618320610687,
      "grad_norm": 1.0093194246292114,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.2151,
      "step": 72
    },
    {
      "epoch": 0.5572519083969466,
      "grad_norm": 1.0736011266708374,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.2563,
      "step": 73
    },
    {
      "epoch": 0.5648854961832062,
      "grad_norm": 0.9279552102088928,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.1685,
      "step": 74
    },
    {
      "epoch": 0.5725190839694656,
      "grad_norm": 1.3419692516326904,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.0466,
      "step": 75
    },
    {
      "epoch": 0.5801526717557252,
      "grad_norm": 1.0275050401687622,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.0661,
      "step": 76
    },
    {
      "epoch": 0.5877862595419847,
      "grad_norm": 1.1091060638427734,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.0072,
      "step": 77
    },
    {
      "epoch": 0.5954198473282443,
      "grad_norm": 1.1305744647979736,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.9354,
      "step": 78
    },
    {
      "epoch": 0.6030534351145038,
      "grad_norm": 1.0726759433746338,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.9506,
      "step": 79
    },
    {
      "epoch": 0.6106870229007634,
      "grad_norm": 1.1804280281066895,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.9821,
      "step": 80
    },
    {
      "epoch": 0.6183206106870229,
      "grad_norm": 1.4476608037948608,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.8601,
      "step": 81
    },
    {
      "epoch": 0.6259541984732825,
      "grad_norm": 1.053503394126892,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.7595,
      "step": 82
    },
    {
      "epoch": 0.6335877862595419,
      "grad_norm": 1.1157360076904297,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.7621,
      "step": 83
    },
    {
      "epoch": 0.6412213740458015,
      "grad_norm": 1.0095500946044922,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.8939,
      "step": 84
    },
    {
      "epoch": 0.648854961832061,
      "grad_norm": 1.344562292098999,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.7774,
      "step": 85
    },
    {
      "epoch": 0.6564885496183206,
      "grad_norm": 1.4751650094985962,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.885,
      "step": 86
    },
    {
      "epoch": 0.6641221374045801,
      "grad_norm": 1.7424293756484985,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.855,
      "step": 87
    },
    {
      "epoch": 0.6717557251908397,
      "grad_norm": 1.1279538869857788,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.9253,
      "step": 88
    },
    {
      "epoch": 0.6793893129770993,
      "grad_norm": 1.1632224321365356,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.6673,
      "step": 89
    },
    {
      "epoch": 0.6870229007633588,
      "grad_norm": 1.5618454217910767,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.7616,
      "step": 90
    },
    {
      "epoch": 0.6946564885496184,
      "grad_norm": 1.3283543586730957,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.8159,
      "step": 91
    },
    {
      "epoch": 0.7022900763358778,
      "grad_norm": 1.1570829153060913,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.7926,
      "step": 92
    },
    {
      "epoch": 0.7099236641221374,
      "grad_norm": 1.1225333213806152,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.6645,
      "step": 93
    },
    {
      "epoch": 0.7175572519083969,
      "grad_norm": 1.1877424716949463,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.661,
      "step": 94
    },
    {
      "epoch": 0.7251908396946565,
      "grad_norm": 1.2467960119247437,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.6066,
      "step": 95
    },
    {
      "epoch": 0.732824427480916,
      "grad_norm": 1.3837032318115234,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.5809,
      "step": 96
    },
    {
      "epoch": 0.7404580152671756,
      "grad_norm": 1.2893717288970947,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.569,
      "step": 97
    },
    {
      "epoch": 0.7480916030534351,
      "grad_norm": 1.2071267366409302,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.6268,
      "step": 98
    },
    {
      "epoch": 0.7557251908396947,
      "grad_norm": 1.2456225156784058,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.7181,
      "step": 99
    },
    {
      "epoch": 0.7633587786259542,
      "grad_norm": 1.1326571702957153,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.6443,
      "step": 100
    },
    {
      "epoch": 0.7633587786259542,
      "eval_loss": 0.5189197659492493,
      "eval_runtime": 33.1324,
      "eval_samples_per_second": 6.67,
      "eval_steps_per_second": 1.69,
      "step": 100
    },
    {
      "epoch": 0.7709923664122137,
      "grad_norm": 1.122446060180664,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.4971,
      "step": 101
    },
    {
      "epoch": 0.7786259541984732,
      "grad_norm": 1.1619453430175781,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.6129,
      "step": 102
    },
    {
      "epoch": 0.7862595419847328,
      "grad_norm": 1.257523536682129,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.6548,
      "step": 103
    },
    {
      "epoch": 0.7938931297709924,
      "grad_norm": 1.1001484394073486,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.5744,
      "step": 104
    },
    {
      "epoch": 0.8015267175572519,
      "grad_norm": 1.1093965768814087,
      "learning_rate": 5e-05,
      "loss": 0.4444,
      "step": 105
    },
    {
      "epoch": 0.8091603053435115,
      "grad_norm": 1.1482518911361694,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.4133,
      "step": 106
    },
    {
      "epoch": 0.816793893129771,
      "grad_norm": 1.0960545539855957,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.4786,
      "step": 107
    },
    {
      "epoch": 0.8244274809160306,
      "grad_norm": 1.010697841644287,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.4315,
      "step": 108
    },
    {
      "epoch": 0.8320610687022901,
      "grad_norm": 1.3616607189178467,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4141,
      "step": 109
    },
    {
      "epoch": 0.8396946564885496,
      "grad_norm": 1.2480610609054565,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3624,
      "step": 110
    },
    {
      "epoch": 0.8473282442748091,
      "grad_norm": 1.0892648696899414,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.3761,
      "step": 111
    },
    {
      "epoch": 0.8549618320610687,
      "grad_norm": 1.0077061653137207,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.3275,
      "step": 112
    },
    {
      "epoch": 0.8625954198473282,
      "grad_norm": 0.9549785256385803,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3581,
      "step": 113
    },
    {
      "epoch": 0.8702290076335878,
      "grad_norm": 1.1935012340545654,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.2861,
      "step": 114
    },
    {
      "epoch": 0.8778625954198473,
      "grad_norm": 1.007349967956543,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.3204,
      "step": 115
    },
    {
      "epoch": 0.8854961832061069,
      "grad_norm": 1.0531667470932007,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.321,
      "step": 116
    },
    {
      "epoch": 0.8931297709923665,
      "grad_norm": 1.0745865106582642,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.2648,
      "step": 117
    },
    {
      "epoch": 0.9007633587786259,
      "grad_norm": 1.1531168222427368,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.2684,
      "step": 118
    },
    {
      "epoch": 0.9083969465648855,
      "grad_norm": 1.0445278882980347,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.2358,
      "step": 119
    },
    {
      "epoch": 0.916030534351145,
      "grad_norm": 0.8978492617607117,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.2525,
      "step": 120
    },
    {
      "epoch": 0.9236641221374046,
      "grad_norm": 0.8278452754020691,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.2797,
      "step": 121
    },
    {
      "epoch": 0.9312977099236641,
      "grad_norm": 0.9945726990699768,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.1907,
      "step": 122
    },
    {
      "epoch": 0.9389312977099237,
      "grad_norm": 0.9082306623458862,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3027,
      "step": 123
    },
    {
      "epoch": 0.9465648854961832,
      "grad_norm": 1.0918529033660889,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.2531,
      "step": 124
    },
    {
      "epoch": 0.9541984732824428,
      "grad_norm": 1.3911166191101074,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.2406,
      "step": 125
    },
    {
      "epoch": 0.9618320610687023,
      "grad_norm": 1.3826948404312134,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.316,
      "step": 126
    },
    {
      "epoch": 0.9694656488549618,
      "grad_norm": 0.9968087673187256,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.2686,
      "step": 127
    },
    {
      "epoch": 0.9770992366412213,
      "grad_norm": 0.9983251690864563,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.24,
      "step": 128
    },
    {
      "epoch": 0.9847328244274809,
      "grad_norm": 0.9860944151878357,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.2883,
      "step": 129
    },
    {
      "epoch": 0.9923664122137404,
      "grad_norm": 0.9234929084777832,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.1924,
      "step": 130
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0712145566940308,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.1951,
      "step": 131
    },
    {
      "epoch": 1.0076335877862594,
      "grad_norm": 1.0491852760314941,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.1464,
      "step": 132
    },
    {
      "epoch": 1.015267175572519,
      "grad_norm": 0.8747281432151794,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.1253,
      "step": 133
    },
    {
      "epoch": 1.0229007633587786,
      "grad_norm": 0.8298887014389038,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.1224,
      "step": 134
    },
    {
      "epoch": 1.0305343511450382,
      "grad_norm": 0.8321114182472229,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.1389,
      "step": 135
    },
    {
      "epoch": 1.0381679389312977,
      "grad_norm": 0.8393105268478394,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.1887,
      "step": 136
    },
    {
      "epoch": 1.0458015267175573,
      "grad_norm": 0.737535297870636,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.1076,
      "step": 137
    },
    {
      "epoch": 1.0534351145038168,
      "grad_norm": 0.7223812341690063,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.1472,
      "step": 138
    },
    {
      "epoch": 1.0610687022900764,
      "grad_norm": 0.7303343415260315,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.1005,
      "step": 139
    },
    {
      "epoch": 1.0687022900763359,
      "grad_norm": 0.5926880836486816,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.0694,
      "step": 140
    },
    {
      "epoch": 1.0763358778625953,
      "grad_norm": 0.7348659634590149,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.0871,
      "step": 141
    },
    {
      "epoch": 1.083969465648855,
      "grad_norm": 0.5914371609687805,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.0751,
      "step": 142
    },
    {
      "epoch": 1.0916030534351144,
      "grad_norm": 0.7125929594039917,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0741,
      "step": 143
    },
    {
      "epoch": 1.099236641221374,
      "grad_norm": 0.7729383707046509,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0879,
      "step": 144
    },
    {
      "epoch": 1.1068702290076335,
      "grad_norm": 0.5619477033615112,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.0583,
      "step": 145
    },
    {
      "epoch": 1.1145038167938932,
      "grad_norm": 0.542046070098877,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.0525,
      "step": 146
    },
    {
      "epoch": 1.1221374045801527,
      "grad_norm": 0.4766978919506073,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.0606,
      "step": 147
    },
    {
      "epoch": 1.1297709923664123,
      "grad_norm": 0.4976109266281128,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.0487,
      "step": 148
    },
    {
      "epoch": 1.1374045801526718,
      "grad_norm": 0.6034334897994995,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.0536,
      "step": 149
    },
    {
      "epoch": 1.1450381679389312,
      "grad_norm": 0.4450475871562958,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.0533,
      "step": 150
    },
    {
      "epoch": 1.1450381679389312,
      "eval_loss": 0.09603867679834366,
      "eval_runtime": 33.0939,
      "eval_samples_per_second": 6.678,
      "eval_steps_per_second": 1.692,
      "step": 150
    },
    {
      "epoch": 1.1526717557251909,
      "grad_norm": 0.5484526753425598,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.0534,
      "step": 151
    },
    {
      "epoch": 1.1603053435114503,
      "grad_norm": 0.5983319282531738,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.0793,
      "step": 152
    },
    {
      "epoch": 1.16793893129771,
      "grad_norm": 0.46181467175483704,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.0636,
      "step": 153
    },
    {
      "epoch": 1.1755725190839694,
      "grad_norm": 0.7171333432197571,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.0701,
      "step": 154
    },
    {
      "epoch": 1.183206106870229,
      "grad_norm": 0.664388120174408,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.0489,
      "step": 155
    },
    {
      "epoch": 1.1908396946564885,
      "grad_norm": 0.46471717953681946,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.0556,
      "step": 156
    },
    {
      "epoch": 1.1984732824427482,
      "grad_norm": 0.5029077529907227,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.064,
      "step": 157
    },
    {
      "epoch": 1.2061068702290076,
      "grad_norm": 0.617490828037262,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.0905,
      "step": 158
    },
    {
      "epoch": 1.213740458015267,
      "grad_norm": 0.7010232210159302,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.1256,
      "step": 159
    },
    {
      "epoch": 1.2213740458015268,
      "grad_norm": 0.5796628594398499,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.0815,
      "step": 160
    },
    {
      "epoch": 1.2290076335877862,
      "grad_norm": 0.44820165634155273,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.0694,
      "step": 161
    },
    {
      "epoch": 1.2366412213740459,
      "grad_norm": 0.5325570702552795,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0471,
      "step": 162
    },
    {
      "epoch": 1.2442748091603053,
      "grad_norm": 0.3855190575122833,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.0596,
      "step": 163
    },
    {
      "epoch": 1.2519083969465647,
      "grad_norm": 0.7213780283927917,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.0767,
      "step": 164
    },
    {
      "epoch": 1.2595419847328244,
      "grad_norm": 0.4919697344303131,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.072,
      "step": 165
    },
    {
      "epoch": 1.267175572519084,
      "grad_norm": 0.6034940481185913,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.0624,
      "step": 166
    },
    {
      "epoch": 1.2748091603053435,
      "grad_norm": 0.3831624984741211,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.044,
      "step": 167
    },
    {
      "epoch": 1.282442748091603,
      "grad_norm": 0.384133517742157,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.0402,
      "step": 168
    },
    {
      "epoch": 1.2900763358778626,
      "grad_norm": 0.44408854842185974,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.0623,
      "step": 169
    },
    {
      "epoch": 1.297709923664122,
      "grad_norm": 0.4559786915779114,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.0453,
      "step": 170
    },
    {
      "epoch": 1.3053435114503817,
      "grad_norm": 0.3425256013870239,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.0424,
      "step": 171
    },
    {
      "epoch": 1.3129770992366412,
      "grad_norm": 0.4652338922023773,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.0596,
      "step": 172
    },
    {
      "epoch": 1.3206106870229006,
      "grad_norm": 0.46287375688552856,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.0467,
      "step": 173
    },
    {
      "epoch": 1.3282442748091603,
      "grad_norm": 0.4973945617675781,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.0522,
      "step": 174
    },
    {
      "epoch": 1.33587786259542,
      "grad_norm": 0.33528539538383484,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.0395,
      "step": 175
    },
    {
      "epoch": 1.3435114503816794,
      "grad_norm": 0.3701744079589844,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.0412,
      "step": 176
    },
    {
      "epoch": 1.3511450381679388,
      "grad_norm": 0.6020646691322327,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.101,
      "step": 177
    },
    {
      "epoch": 1.3587786259541985,
      "grad_norm": 0.29945698380470276,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.0322,
      "step": 178
    },
    {
      "epoch": 1.366412213740458,
      "grad_norm": 0.3263508081436157,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0355,
      "step": 179
    },
    {
      "epoch": 1.3740458015267176,
      "grad_norm": 0.3674665689468384,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.0428,
      "step": 180
    },
    {
      "epoch": 1.381679389312977,
      "grad_norm": 0.3940519690513611,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.0656,
      "step": 181
    },
    {
      "epoch": 1.3893129770992365,
      "grad_norm": 0.32840511202812195,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.0402,
      "step": 182
    },
    {
      "epoch": 1.3969465648854962,
      "grad_norm": 0.40500274300575256,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.0523,
      "step": 183
    },
    {
      "epoch": 1.4045801526717558,
      "grad_norm": 0.3779938817024231,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.0482,
      "step": 184
    },
    {
      "epoch": 1.4122137404580153,
      "grad_norm": 0.2691035270690918,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.0352,
      "step": 185
    },
    {
      "epoch": 1.4198473282442747,
      "grad_norm": 0.2667372226715088,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0347,
      "step": 186
    },
    {
      "epoch": 1.4274809160305344,
      "grad_norm": 0.30084162950515747,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0453,
      "step": 187
    },
    {
      "epoch": 1.4351145038167938,
      "grad_norm": 0.3017407953739166,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.0438,
      "step": 188
    },
    {
      "epoch": 1.4427480916030535,
      "grad_norm": 0.4236651659011841,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.075,
      "step": 189
    },
    {
      "epoch": 1.450381679389313,
      "grad_norm": 0.3632592558860779,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0562,
      "step": 190
    },
    {
      "epoch": 1.4580152671755724,
      "grad_norm": 0.32428884506225586,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.0416,
      "step": 191
    },
    {
      "epoch": 1.465648854961832,
      "grad_norm": 0.38446319103240967,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0366,
      "step": 192
    },
    {
      "epoch": 1.4732824427480917,
      "grad_norm": 0.3266568183898926,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.0457,
      "step": 193
    },
    {
      "epoch": 1.4809160305343512,
      "grad_norm": 0.3107873797416687,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0398,
      "step": 194
    },
    {
      "epoch": 1.4885496183206106,
      "grad_norm": 0.37324926257133484,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.0671,
      "step": 195
    },
    {
      "epoch": 1.4961832061068703,
      "grad_norm": 0.42532414197921753,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.042,
      "step": 196
    },
    {
      "epoch": 1.5038167938931297,
      "grad_norm": 0.47520244121551514,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.1062,
      "step": 197
    },
    {
      "epoch": 1.5114503816793894,
      "grad_norm": 0.3452816903591156,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.0437,
      "step": 198
    },
    {
      "epoch": 1.5190839694656488,
      "grad_norm": 0.3584132492542267,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0509,
      "step": 199
    },
    {
      "epoch": 1.5267175572519083,
      "grad_norm": 0.43450045585632324,
      "learning_rate": 0.0,
      "loss": 0.0599,
      "step": 200
    },
    {
      "epoch": 1.5267175572519083,
      "eval_loss": 0.05885966122150421,
      "eval_runtime": 33.1395,
      "eval_samples_per_second": 6.669,
      "eval_steps_per_second": 1.69,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 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": 6.03028303103656e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}