{
  "best_metric": 0.8087199926376343,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 2.1052631578947367,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.010526315789473684,
      "grad_norm": 9.168891906738281,
      "learning_rate": 1e-05,
      "loss": 2.4705,
      "step": 1
    },
    {
      "epoch": 0.010526315789473684,
      "eval_loss": 2.706315279006958,
      "eval_runtime": 10.6508,
      "eval_samples_per_second": 15.022,
      "eval_steps_per_second": 3.756,
      "step": 1
    },
    {
      "epoch": 0.021052631578947368,
      "grad_norm": 9.934759140014648,
      "learning_rate": 2e-05,
      "loss": 2.5412,
      "step": 2
    },
    {
      "epoch": 0.031578947368421054,
      "grad_norm": 9.2969388961792,
      "learning_rate": 3e-05,
      "loss": 2.5141,
      "step": 3
    },
    {
      "epoch": 0.042105263157894736,
      "grad_norm": 6.368276596069336,
      "learning_rate": 4e-05,
      "loss": 2.2398,
      "step": 4
    },
    {
      "epoch": 0.05263157894736842,
      "grad_norm": 4.614349365234375,
      "learning_rate": 5e-05,
      "loss": 1.9648,
      "step": 5
    },
    {
      "epoch": 0.06315789473684211,
      "grad_norm": 3.133300542831421,
      "learning_rate": 6e-05,
      "loss": 1.6843,
      "step": 6
    },
    {
      "epoch": 0.07368421052631578,
      "grad_norm": 2.8085145950317383,
      "learning_rate": 7e-05,
      "loss": 1.6361,
      "step": 7
    },
    {
      "epoch": 0.08421052631578947,
      "grad_norm": 2.5651118755340576,
      "learning_rate": 8e-05,
      "loss": 1.5673,
      "step": 8
    },
    {
      "epoch": 0.09473684210526316,
      "grad_norm": 2.077059745788574,
      "learning_rate": 9e-05,
      "loss": 1.492,
      "step": 9
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 1.6968876123428345,
      "learning_rate": 0.0001,
      "loss": 1.4898,
      "step": 10
    },
    {
      "epoch": 0.11578947368421053,
      "grad_norm": 1.5873697996139526,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.3308,
      "step": 11
    },
    {
      "epoch": 0.12631578947368421,
      "grad_norm": 1.692470908164978,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.379,
      "step": 12
    },
    {
      "epoch": 0.1368421052631579,
      "grad_norm": 1.6148205995559692,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.3958,
      "step": 13
    },
    {
      "epoch": 0.14736842105263157,
      "grad_norm": 1.6712144613265991,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.3219,
      "step": 14
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 1.5100483894348145,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4039,
      "step": 15
    },
    {
      "epoch": 0.16842105263157894,
      "grad_norm": 1.7261104583740234,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.325,
      "step": 16
    },
    {
      "epoch": 0.17894736842105263,
      "grad_norm": 1.8913586139678955,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.2242,
      "step": 17
    },
    {
      "epoch": 0.18947368421052632,
      "grad_norm": 1.5054707527160645,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.1904,
      "step": 18
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4594191312789917,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.2123,
      "step": 19
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 1.667661190032959,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3184,
      "step": 20
    },
    {
      "epoch": 0.22105263157894736,
      "grad_norm": 1.8904058933258057,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.408,
      "step": 21
    },
    {
      "epoch": 0.23157894736842105,
      "grad_norm": 1.6303354501724243,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.514,
      "step": 22
    },
    {
      "epoch": 0.24210526315789474,
      "grad_norm": 2.281747817993164,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.5531,
      "step": 23
    },
    {
      "epoch": 0.25263157894736843,
      "grad_norm": 1.6432545185089111,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.2644,
      "step": 24
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 1.327402114868164,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.1711,
      "step": 25
    },
    {
      "epoch": 0.2736842105263158,
      "grad_norm": 1.334049940109253,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.2629,
      "step": 26
    },
    {
      "epoch": 0.28421052631578947,
      "grad_norm": 1.2756692171096802,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.2028,
      "step": 27
    },
    {
      "epoch": 0.29473684210526313,
      "grad_norm": 1.2860331535339355,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.1545,
      "step": 28
    },
    {
      "epoch": 0.30526315789473685,
      "grad_norm": 1.3939191102981567,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.2031,
      "step": 29
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 1.2090693712234497,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.1945,
      "step": 30
    },
    {
      "epoch": 0.3263157894736842,
      "grad_norm": 1.0800600051879883,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.1995,
      "step": 31
    },
    {
      "epoch": 0.3368421052631579,
      "grad_norm": 1.0718035697937012,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.1467,
      "step": 32
    },
    {
      "epoch": 0.3473684210526316,
      "grad_norm": 1.1007486581802368,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.1223,
      "step": 33
    },
    {
      "epoch": 0.35789473684210527,
      "grad_norm": 1.071140170097351,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.0291,
      "step": 34
    },
    {
      "epoch": 0.3684210526315789,
      "grad_norm": 1.0698237419128418,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.0913,
      "step": 35
    },
    {
      "epoch": 0.37894736842105264,
      "grad_norm": 1.2563166618347168,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.1284,
      "step": 36
    },
    {
      "epoch": 0.3894736842105263,
      "grad_norm": 1.2085785865783691,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.1161,
      "step": 37
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1608299016952515,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.0755,
      "step": 38
    },
    {
      "epoch": 0.4105263157894737,
      "grad_norm": 1.1647933721542358,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.0498,
      "step": 39
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 1.1612097024917603,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.1051,
      "step": 40
    },
    {
      "epoch": 0.43157894736842106,
      "grad_norm": 1.060537338256836,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.9606,
      "step": 41
    },
    {
      "epoch": 0.4421052631578947,
      "grad_norm": 1.2803679704666138,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.0484,
      "step": 42
    },
    {
      "epoch": 0.45263157894736844,
      "grad_norm": 1.1957224607467651,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.0931,
      "step": 43
    },
    {
      "epoch": 0.4631578947368421,
      "grad_norm": 1.2331677675247192,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.3474,
      "step": 44
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 1.3564037084579468,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.3815,
      "step": 45
    },
    {
      "epoch": 0.4842105263157895,
      "grad_norm": 1.447937250137329,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5345,
      "step": 46
    },
    {
      "epoch": 0.49473684210526314,
      "grad_norm": 1.0807641744613647,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.171,
      "step": 47
    },
    {
      "epoch": 0.5052631578947369,
      "grad_norm": 1.1544193029403687,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.1188,
      "step": 48
    },
    {
      "epoch": 0.5157894736842106,
      "grad_norm": 1.007704257965088,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.2125,
      "step": 49
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 1.086031198501587,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.1356,
      "step": 50
    },
    {
      "epoch": 0.5263157894736842,
      "eval_loss": 1.0524232387542725,
      "eval_runtime": 10.9787,
      "eval_samples_per_second": 14.574,
      "eval_steps_per_second": 3.643,
      "step": 50
    },
    {
      "epoch": 0.5368421052631579,
      "grad_norm": 1.0077000856399536,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.0592,
      "step": 51
    },
    {
      "epoch": 0.5473684210526316,
      "grad_norm": 1.1932862997055054,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.1516,
      "step": 52
    },
    {
      "epoch": 0.5578947368421052,
      "grad_norm": 1.0632336139678955,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.0657,
      "step": 53
    },
    {
      "epoch": 0.5684210526315789,
      "grad_norm": 1.1923375129699707,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.0276,
      "step": 54
    },
    {
      "epoch": 0.5789473684210527,
      "grad_norm": 1.1214863061904907,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.0396,
      "step": 55
    },
    {
      "epoch": 0.5894736842105263,
      "grad_norm": 1.1271682977676392,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.104,
      "step": 56
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.062990665435791,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.1821,
      "step": 57
    },
    {
      "epoch": 0.6105263157894737,
      "grad_norm": 1.065353274345398,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.9104,
      "step": 58
    },
    {
      "epoch": 0.6210526315789474,
      "grad_norm": 1.1170157194137573,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.0309,
      "step": 59
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 1.0927435159683228,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.9275,
      "step": 60
    },
    {
      "epoch": 0.6421052631578947,
      "grad_norm": 1.1789442300796509,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.1096,
      "step": 61
    },
    {
      "epoch": 0.6526315789473685,
      "grad_norm": 1.178418517112732,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.0645,
      "step": 62
    },
    {
      "epoch": 0.6631578947368421,
      "grad_norm": 1.2233365774154663,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.9945,
      "step": 63
    },
    {
      "epoch": 0.6736842105263158,
      "grad_norm": 1.1698808670043945,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.9676,
      "step": 64
    },
    {
      "epoch": 0.6842105263157895,
      "grad_norm": 1.2741317749023438,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.8927,
      "step": 65
    },
    {
      "epoch": 0.6947368421052632,
      "grad_norm": 1.2696579694747925,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.9449,
      "step": 66
    },
    {
      "epoch": 0.7052631578947368,
      "grad_norm": 1.28653085231781,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.1598,
      "step": 67
    },
    {
      "epoch": 0.7157894736842105,
      "grad_norm": 1.190481424331665,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.2999,
      "step": 68
    },
    {
      "epoch": 0.7263157894736842,
      "grad_norm": 1.3982832431793213,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4338,
      "step": 69
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 1.0171552896499634,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.021,
      "step": 70
    },
    {
      "epoch": 0.7473684210526316,
      "grad_norm": 0.9902419447898865,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.0426,
      "step": 71
    },
    {
      "epoch": 0.7578947368421053,
      "grad_norm": 1.038367509841919,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.1029,
      "step": 72
    },
    {
      "epoch": 0.7684210526315789,
      "grad_norm": 0.9751963019371033,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.007,
      "step": 73
    },
    {
      "epoch": 0.7789473684210526,
      "grad_norm": 1.0030914545059204,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.075,
      "step": 74
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 1.0930050611495972,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.0726,
      "step": 75
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9872201085090637,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.8942,
      "step": 76
    },
    {
      "epoch": 0.8105263157894737,
      "grad_norm": 1.0911402702331543,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.0035,
      "step": 77
    },
    {
      "epoch": 0.8210526315789474,
      "grad_norm": 1.248184084892273,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.0278,
      "step": 78
    },
    {
      "epoch": 0.8315789473684211,
      "grad_norm": 1.140582799911499,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.9931,
      "step": 79
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 1.2266650199890137,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.9708,
      "step": 80
    },
    {
      "epoch": 0.8526315789473684,
      "grad_norm": 1.173133134841919,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.049,
      "step": 81
    },
    {
      "epoch": 0.8631578947368421,
      "grad_norm": 1.241225242614746,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.0773,
      "step": 82
    },
    {
      "epoch": 0.8736842105263158,
      "grad_norm": 1.057950735092163,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.881,
      "step": 83
    },
    {
      "epoch": 0.8842105263157894,
      "grad_norm": 1.2249699831008911,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.9547,
      "step": 84
    },
    {
      "epoch": 0.8947368421052632,
      "grad_norm": 1.3183869123458862,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.9861,
      "step": 85
    },
    {
      "epoch": 0.9052631578947369,
      "grad_norm": 1.1633167266845703,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.9224,
      "step": 86
    },
    {
      "epoch": 0.9157894736842105,
      "grad_norm": 1.0952543020248413,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.8516,
      "step": 87
    },
    {
      "epoch": 0.9263157894736842,
      "grad_norm": 1.2492117881774902,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.8927,
      "step": 88
    },
    {
      "epoch": 0.9368421052631579,
      "grad_norm": 1.2381356954574585,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.8933,
      "step": 89
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 1.4425264596939087,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.0052,
      "step": 90
    },
    {
      "epoch": 0.9578947368421052,
      "grad_norm": 1.2662031650543213,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.2001,
      "step": 91
    },
    {
      "epoch": 0.968421052631579,
      "grad_norm": 1.380860447883606,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2722,
      "step": 92
    },
    {
      "epoch": 0.9789473684210527,
      "grad_norm": 0.9866234660148621,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.0599,
      "step": 93
    },
    {
      "epoch": 0.9894736842105263,
      "grad_norm": 1.072219967842102,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.9319,
      "step": 94
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2425395250320435,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.0947,
      "step": 95
    },
    {
      "epoch": 1.0105263157894737,
      "grad_norm": 0.8957297801971436,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.0187,
      "step": 96
    },
    {
      "epoch": 1.0210526315789474,
      "grad_norm": 0.9413051605224609,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.9436,
      "step": 97
    },
    {
      "epoch": 1.0315789473684212,
      "grad_norm": 0.9626156687736511,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.8942,
      "step": 98
    },
    {
      "epoch": 1.0421052631578946,
      "grad_norm": 0.9372084736824036,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.8256,
      "step": 99
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 1.0069046020507812,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.8654,
      "step": 100
    },
    {
      "epoch": 1.0526315789473684,
      "eval_loss": 0.9086695909500122,
      "eval_runtime": 10.9842,
      "eval_samples_per_second": 14.566,
      "eval_steps_per_second": 3.642,
      "step": 100
    },
    {
      "epoch": 1.063157894736842,
      "grad_norm": 0.9769554734230042,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.8353,
      "step": 101
    },
    {
      "epoch": 1.0736842105263158,
      "grad_norm": 1.072082757949829,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.7628,
      "step": 102
    },
    {
      "epoch": 1.0842105263157895,
      "grad_norm": 1.131913661956787,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.8054,
      "step": 103
    },
    {
      "epoch": 1.0947368421052632,
      "grad_norm": 1.0401241779327393,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.7554,
      "step": 104
    },
    {
      "epoch": 1.1052631578947367,
      "grad_norm": 1.1907769441604614,
      "learning_rate": 5e-05,
      "loss": 0.8031,
      "step": 105
    },
    {
      "epoch": 1.1157894736842104,
      "grad_norm": 1.197701334953308,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.7472,
      "step": 106
    },
    {
      "epoch": 1.1263157894736842,
      "grad_norm": 1.144927978515625,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.6614,
      "step": 107
    },
    {
      "epoch": 1.1368421052631579,
      "grad_norm": 1.166655421257019,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.6584,
      "step": 108
    },
    {
      "epoch": 1.1473684210526316,
      "grad_norm": 1.2122156620025635,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.7671,
      "step": 109
    },
    {
      "epoch": 1.1578947368421053,
      "grad_norm": 1.233668565750122,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.7544,
      "step": 110
    },
    {
      "epoch": 1.168421052631579,
      "grad_norm": 1.2453961372375488,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.653,
      "step": 111
    },
    {
      "epoch": 1.1789473684210527,
      "grad_norm": 1.2108608484268188,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.6219,
      "step": 112
    },
    {
      "epoch": 1.1894736842105262,
      "grad_norm": 1.2988357543945312,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.6398,
      "step": 113
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4753323793411255,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.7204,
      "step": 114
    },
    {
      "epoch": 1.2105263157894737,
      "grad_norm": 1.4893972873687744,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.7253,
      "step": 115
    },
    {
      "epoch": 1.2210526315789474,
      "grad_norm": 1.373826026916504,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.9182,
      "step": 116
    },
    {
      "epoch": 1.231578947368421,
      "grad_norm": 1.3359689712524414,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.9016,
      "step": 117
    },
    {
      "epoch": 1.2421052631578948,
      "grad_norm": 1.5541832447052002,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.984,
      "step": 118
    },
    {
      "epoch": 1.2526315789473683,
      "grad_norm": 1.2436407804489136,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.8583,
      "step": 119
    },
    {
      "epoch": 1.263157894736842,
      "grad_norm": 1.091597318649292,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.739,
      "step": 120
    },
    {
      "epoch": 1.2736842105263158,
      "grad_norm": 1.0907912254333496,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.8073,
      "step": 121
    },
    {
      "epoch": 1.2842105263157895,
      "grad_norm": 1.0711778402328491,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.8872,
      "step": 122
    },
    {
      "epoch": 1.2947368421052632,
      "grad_norm": 1.075634479522705,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.753,
      "step": 123
    },
    {
      "epoch": 1.305263157894737,
      "grad_norm": 1.1155037879943848,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.809,
      "step": 124
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 1.160321831703186,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.7665,
      "step": 125
    },
    {
      "epoch": 1.3263157894736843,
      "grad_norm": 1.0835896730422974,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.6832,
      "step": 126
    },
    {
      "epoch": 1.3368421052631578,
      "grad_norm": 1.1323840618133545,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.7475,
      "step": 127
    },
    {
      "epoch": 1.3473684210526315,
      "grad_norm": 1.1929233074188232,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.7142,
      "step": 128
    },
    {
      "epoch": 1.3578947368421053,
      "grad_norm": 1.2793809175491333,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.7361,
      "step": 129
    },
    {
      "epoch": 1.368421052631579,
      "grad_norm": 1.162257432937622,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.6343,
      "step": 130
    },
    {
      "epoch": 1.3789473684210527,
      "grad_norm": 1.317219614982605,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.6337,
      "step": 131
    },
    {
      "epoch": 1.3894736842105262,
      "grad_norm": 1.2479079961776733,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.6729,
      "step": 132
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3856415748596191,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.7528,
      "step": 133
    },
    {
      "epoch": 1.4105263157894736,
      "grad_norm": 1.3722341060638428,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.6511,
      "step": 134
    },
    {
      "epoch": 1.4210526315789473,
      "grad_norm": 1.3193578720092773,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.6161,
      "step": 135
    },
    {
      "epoch": 1.431578947368421,
      "grad_norm": 1.2954341173171997,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.6445,
      "step": 136
    },
    {
      "epoch": 1.4421052631578948,
      "grad_norm": 1.290369987487793,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.5826,
      "step": 137
    },
    {
      "epoch": 1.4526315789473685,
      "grad_norm": 1.4394952058792114,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.6496,
      "step": 138
    },
    {
      "epoch": 1.4631578947368422,
      "grad_norm": 1.522017240524292,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.7063,
      "step": 139
    },
    {
      "epoch": 1.4736842105263157,
      "grad_norm": 1.3147685527801514,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.9626,
      "step": 140
    },
    {
      "epoch": 1.4842105263157894,
      "grad_norm": 1.5802985429763794,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.9413,
      "step": 141
    },
    {
      "epoch": 1.4947368421052631,
      "grad_norm": 1.2526700496673584,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.8346,
      "step": 142
    },
    {
      "epoch": 1.5052631578947369,
      "grad_norm": 1.258183240890503,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.7935,
      "step": 143
    },
    {
      "epoch": 1.5157894736842106,
      "grad_norm": 1.3632400035858154,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.7967,
      "step": 144
    },
    {
      "epoch": 1.526315789473684,
      "grad_norm": 1.2614063024520874,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.7836,
      "step": 145
    },
    {
      "epoch": 1.5368421052631578,
      "grad_norm": 1.3294516801834106,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.7997,
      "step": 146
    },
    {
      "epoch": 1.5473684210526315,
      "grad_norm": 1.3192791938781738,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.7401,
      "step": 147
    },
    {
      "epoch": 1.5578947368421052,
      "grad_norm": 1.3414055109024048,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.6897,
      "step": 148
    },
    {
      "epoch": 1.568421052631579,
      "grad_norm": 1.2699148654937744,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.6897,
      "step": 149
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 1.2533363103866577,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.7959,
      "step": 150
    },
    {
      "epoch": 1.5789473684210527,
      "eval_loss": 0.8304319381713867,
      "eval_runtime": 10.9904,
      "eval_samples_per_second": 14.558,
      "eval_steps_per_second": 3.64,
      "step": 150
    },
    {
      "epoch": 1.5894736842105264,
      "grad_norm": 1.2978832721710205,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.628,
      "step": 151
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3389592170715332,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.6815,
      "step": 152
    },
    {
      "epoch": 1.6105263157894738,
      "grad_norm": 1.352181315422058,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.6532,
      "step": 153
    },
    {
      "epoch": 1.6210526315789475,
      "grad_norm": 1.3468338251113892,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.7073,
      "step": 154
    },
    {
      "epoch": 1.631578947368421,
      "grad_norm": 1.2590681314468384,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.6532,
      "step": 155
    },
    {
      "epoch": 1.6421052631578947,
      "grad_norm": 1.3821194171905518,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.5732,
      "step": 156
    },
    {
      "epoch": 1.6526315789473685,
      "grad_norm": 1.5178122520446777,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.7807,
      "step": 157
    },
    {
      "epoch": 1.663157894736842,
      "grad_norm": 1.2092030048370361,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.6053,
      "step": 158
    },
    {
      "epoch": 1.6736842105263157,
      "grad_norm": 1.2986096143722534,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.5737,
      "step": 159
    },
    {
      "epoch": 1.6842105263157894,
      "grad_norm": 1.4210915565490723,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.571,
      "step": 160
    },
    {
      "epoch": 1.694736842105263,
      "grad_norm": 1.2712725400924683,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.6034,
      "step": 161
    },
    {
      "epoch": 1.7052631578947368,
      "grad_norm": 1.4907017946243286,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.8371,
      "step": 162
    },
    {
      "epoch": 1.7157894736842105,
      "grad_norm": 1.4650241136550903,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.7888,
      "step": 163
    },
    {
      "epoch": 1.7263157894736842,
      "grad_norm": 1.5433107614517212,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.9085,
      "step": 164
    },
    {
      "epoch": 1.736842105263158,
      "grad_norm": 1.238660454750061,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.7676,
      "step": 165
    },
    {
      "epoch": 1.7473684210526317,
      "grad_norm": 1.2105224132537842,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.7797,
      "step": 166
    },
    {
      "epoch": 1.7578947368421054,
      "grad_norm": 1.3489967584609985,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.818,
      "step": 167
    },
    {
      "epoch": 1.768421052631579,
      "grad_norm": 1.4526557922363281,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.7486,
      "step": 168
    },
    {
      "epoch": 1.7789473684210526,
      "grad_norm": 1.2024424076080322,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.6745,
      "step": 169
    },
    {
      "epoch": 1.7894736842105263,
      "grad_norm": 1.1779471635818481,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.7127,
      "step": 170
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3835088014602661,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.694,
      "step": 171
    },
    {
      "epoch": 1.8105263157894735,
      "grad_norm": 1.216235637664795,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.6415,
      "step": 172
    },
    {
      "epoch": 1.8210526315789473,
      "grad_norm": 1.2513995170593262,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.6892,
      "step": 173
    },
    {
      "epoch": 1.831578947368421,
      "grad_norm": 1.3052314519882202,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.7033,
      "step": 174
    },
    {
      "epoch": 1.8421052631578947,
      "grad_norm": 1.3266228437423706,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.6935,
      "step": 175
    },
    {
      "epoch": 1.8526315789473684,
      "grad_norm": 1.259132981300354,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.6738,
      "step": 176
    },
    {
      "epoch": 1.8631578947368421,
      "grad_norm": 1.4048694372177124,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.6973,
      "step": 177
    },
    {
      "epoch": 1.8736842105263158,
      "grad_norm": 1.3221054077148438,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.6476,
      "step": 178
    },
    {
      "epoch": 1.8842105263157896,
      "grad_norm": 1.3181681632995605,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.6618,
      "step": 179
    },
    {
      "epoch": 1.8947368421052633,
      "grad_norm": 1.2113510370254517,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.6224,
      "step": 180
    },
    {
      "epoch": 1.905263157894737,
      "grad_norm": 1.3947473764419556,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.6623,
      "step": 181
    },
    {
      "epoch": 1.9157894736842105,
      "grad_norm": 1.2772618532180786,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.5244,
      "step": 182
    },
    {
      "epoch": 1.9263157894736842,
      "grad_norm": 1.487331748008728,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.597,
      "step": 183
    },
    {
      "epoch": 1.936842105263158,
      "grad_norm": 1.5507277250289917,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.6178,
      "step": 184
    },
    {
      "epoch": 1.9473684210526314,
      "grad_norm": 1.6640244722366333,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.8157,
      "step": 185
    },
    {
      "epoch": 1.9578947368421051,
      "grad_norm": 1.388296127319336,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.8824,
      "step": 186
    },
    {
      "epoch": 1.9684210526315788,
      "grad_norm": 1.5966399908065796,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0097,
      "step": 187
    },
    {
      "epoch": 1.9789473684210526,
      "grad_norm": 1.032527208328247,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.6459,
      "step": 188
    },
    {
      "epoch": 1.9894736842105263,
      "grad_norm": 1.1817500591278076,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.6843,
      "step": 189
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4044283628463745,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.6426,
      "step": 190
    },
    {
      "epoch": 2.0105263157894737,
      "grad_norm": 1.0268614292144775,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.7309,
      "step": 191
    },
    {
      "epoch": 2.0210526315789474,
      "grad_norm": 1.0670136213302612,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.6426,
      "step": 192
    },
    {
      "epoch": 2.031578947368421,
      "grad_norm": 1.0951060056686401,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.7445,
      "step": 193
    },
    {
      "epoch": 2.042105263157895,
      "grad_norm": 1.150851845741272,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.7063,
      "step": 194
    },
    {
      "epoch": 2.0526315789473686,
      "grad_norm": 1.100847601890564,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.6454,
      "step": 195
    },
    {
      "epoch": 2.0631578947368423,
      "grad_norm": 1.1162431240081787,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.6035,
      "step": 196
    },
    {
      "epoch": 2.0736842105263156,
      "grad_norm": 1.0202869176864624,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.5302,
      "step": 197
    },
    {
      "epoch": 2.0842105263157893,
      "grad_norm": 1.2192672491073608,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.6254,
      "step": 198
    },
    {
      "epoch": 2.094736842105263,
      "grad_norm": 1.218741536140442,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.6507,
      "step": 199
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 1.1481413841247559,
      "learning_rate": 0.0,
      "loss": 0.5537,
      "step": 200
    },
    {
      "epoch": 2.1052631578947367,
      "eval_loss": 0.8087199926376343,
      "eval_runtime": 10.9936,
      "eval_samples_per_second": 14.554,
      "eval_steps_per_second": 3.638,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 1.5529660911294874e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}