{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0004614674665435,
  "eval_steps": 500,
  "global_step": 271,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 6.060606060606061e-07,
      "loss": 1.6011,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.2121212121212122e-06,
      "loss": 1.3923,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 1.5238,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4242424242424244e-06,
      "loss": 1.4057,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 1.4638,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 1.4316,
      "step": 6
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.242424242424243e-06,
      "loss": 1.3427,
      "step": 7
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.848484848484849e-06,
      "loss": 1.2477,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 1.2386,
      "step": 9
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.060606060606061e-06,
      "loss": 1.2007,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.1452,
      "step": 11
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.272727272727273e-06,
      "loss": 1.1837,
      "step": 12
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.87878787878788e-06,
      "loss": 1.1542,
      "step": 13
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.484848484848486e-06,
      "loss": 1.2,
      "step": 14
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.090909090909091e-06,
      "loss": 1.2151,
      "step": 15
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.696969696969698e-06,
      "loss": 1.202,
      "step": 16
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0303030303030304e-05,
      "loss": 1.209,
      "step": 17
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 1.1633,
      "step": 18
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.1515151515151517e-05,
      "loss": 1.1233,
      "step": 19
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 1.1684,
      "step": 20
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 1.1677,
      "step": 21
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.1706,
      "step": 22
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3939393939393942e-05,
      "loss": 1.1212,
      "step": 23
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4545454545454546e-05,
      "loss": 1.2346,
      "step": 24
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 1.205,
      "step": 25
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.575757575757576e-05,
      "loss": 1.1821,
      "step": 26
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 1.18,
      "step": 27
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6969696969696972e-05,
      "loss": 1.2078,
      "step": 28
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7575757575757576e-05,
      "loss": 1.1999,
      "step": 29
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 1.1746,
      "step": 30
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.8787878787878792e-05,
      "loss": 1.1802,
      "step": 31
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9393939393939395e-05,
      "loss": 1.1883,
      "step": 32
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 1.1985,
      "step": 33
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9999954983050696e-05,
      "loss": 1.0979,
      "step": 34
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.999981993260809e-05,
      "loss": 1.2325,
      "step": 35
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9999594849888083e-05,
      "loss": 1.2247,
      "step": 36
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.99992797369172e-05,
      "loss": 1.1616,
      "step": 37
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9998874596532512e-05,
      "loss": 1.2291,
      "step": 38
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.999837943238166e-05,
      "loss": 1.1251,
      "step": 39
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.99977942489228e-05,
      "loss": 1.12,
      "step": 40
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.999711905142457e-05,
      "loss": 1.1749,
      "step": 41
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9996353845966033e-05,
      "loss": 1.2086,
      "step": 42
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9995498639436634e-05,
      "loss": 1.2592,
      "step": 43
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9994553439536125e-05,
      "loss": 1.2809,
      "step": 44
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9993518254774517e-05,
      "loss": 1.1214,
      "step": 45
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9992393094471976e-05,
      "loss": 1.1859,
      "step": 46
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9991177968758764e-05,
      "loss": 1.1512,
      "step": 47
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.998987288857513e-05,
      "loss": 1.1474,
      "step": 48
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9988477865671217e-05,
      "loss": 1.2545,
      "step": 49
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.998699291260696e-05,
      "loss": 1.2222,
      "step": 50
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9985418042751975e-05,
      "loss": 1.2038,
      "step": 51
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9983753270285423e-05,
      "loss": 1.142,
      "step": 52
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.998199861019591e-05,
      "loss": 1.1784,
      "step": 53
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.998015407828131e-05,
      "loss": 1.2366,
      "step": 54
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9978219691148676e-05,
      "loss": 1.1631,
      "step": 55
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9976195466214042e-05,
      "loss": 1.1555,
      "step": 56
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9974081421702296e-05,
      "loss": 1.2391,
      "step": 57
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9971877576647005e-05,
      "loss": 1.1841,
      "step": 58
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9969583950890245e-05,
      "loss": 1.2531,
      "step": 59
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9967200565082426e-05,
      "loss": 1.2242,
      "step": 60
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9964727440682097e-05,
      "loss": 1.2037,
      "step": 61
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9962164599955762e-05,
      "loss": 1.1442,
      "step": 62
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9959512065977673e-05,
      "loss": 1.2214,
      "step": 63
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.995676986262963e-05,
      "loss": 1.2022,
      "step": 64
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9953938014600757e-05,
      "loss": 1.1414,
      "step": 65
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9951016547387286e-05,
      "loss": 1.1876,
      "step": 66
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.994800548729233e-05,
      "loss": 1.1716,
      "step": 67
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9944904861425626e-05,
      "loss": 1.2206,
      "step": 68
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9941714697703333e-05,
      "loss": 1.1521,
      "step": 69
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9938435024847723e-05,
      "loss": 1.2028,
      "step": 70
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9935065872386977e-05,
      "loss": 1.1917,
      "step": 71
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.993160727065489e-05,
      "loss": 1.2168,
      "step": 72
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.99280592507906e-05,
      "loss": 1.2537,
      "step": 73
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.992442184473831e-05,
      "loss": 1.2286,
      "step": 74
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9920695085247012e-05,
      "loss": 1.1492,
      "step": 75
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9916879005870164e-05,
      "loss": 1.2348,
      "step": 76
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9912973640965423e-05,
      "loss": 1.1437,
      "step": 77
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9908979025694312e-05,
      "loss": 1.2257,
      "step": 78
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.990489519602191e-05,
      "loss": 1.1692,
      "step": 79
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9900722188716526e-05,
      "loss": 1.2699,
      "step": 80
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.989646004134937e-05,
      "loss": 1.2475,
      "step": 81
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.989210879229422e-05,
      "loss": 1.261,
      "step": 82
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9887668480727066e-05,
      "loss": 1.1786,
      "step": 83
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9883139146625763e-05,
      "loss": 1.2237,
      "step": 84
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9878520830769675e-05,
      "loss": 1.2214,
      "step": 85
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9873813574739293e-05,
      "loss": 1.3091,
      "step": 86
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9869017420915888e-05,
      "loss": 1.2442,
      "step": 87
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9864132412481094e-05,
      "loss": 1.2017,
      "step": 88
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9859158593416554e-05,
      "loss": 1.1557,
      "step": 89
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9854096008503495e-05,
      "loss": 1.194,
      "step": 90
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9848944703322345e-05,
      "loss": 1.1097,
      "step": 91
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9843704724252308e-05,
      "loss": 1.1112,
      "step": 92
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9838376118470965e-05,
      "loss": 1.1634,
      "step": 93
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.983295893395383e-05,
      "loss": 1.0942,
      "step": 94
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9827453219473925e-05,
      "loss": 1.1493,
      "step": 95
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9821859024601345e-05,
      "loss": 1.2171,
      "step": 96
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9816176399702806e-05,
      "loss": 1.1748,
      "step": 97
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.98104053959412e-05,
      "loss": 1.2336,
      "step": 98
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9804546065275116e-05,
      "loss": 1.1497,
      "step": 99
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9798598460458394e-05,
      "loss": 1.1544,
      "step": 100
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.979256263503965e-05,
      "loss": 1.2676,
      "step": 101
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.978643864336176e-05,
      "loss": 1.1786,
      "step": 102
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9780226540561413e-05,
      "loss": 1.2087,
      "step": 103
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9773926382568592e-05,
      "loss": 1.1794,
      "step": 104
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9767538226106078e-05,
      "loss": 1.1964,
      "step": 105
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9761062128688932e-05,
      "loss": 1.2106,
      "step": 106
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9754498148623985e-05,
      "loss": 1.2571,
      "step": 107
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9747846345009306e-05,
      "loss": 1.2233,
      "step": 108
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.974110677773368e-05,
      "loss": 1.2183,
      "step": 109
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9734279507476057e-05,
      "loss": 1.2022,
      "step": 110
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9727364595705012e-05,
      "loss": 1.1858,
      "step": 111
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9720362104678193e-05,
      "loss": 1.1514,
      "step": 112
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9713272097441755e-05,
      "loss": 1.2193,
      "step": 113
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9706094637829797e-05,
      "loss": 1.1742,
      "step": 114
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9698829790463792e-05,
      "loss": 1.2172,
      "step": 115
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9691477620751985e-05,
      "loss": 1.1813,
      "step": 116
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9684038194888827e-05,
      "loss": 1.1926,
      "step": 117
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9676511579854375e-05,
      "loss": 1.2264,
      "step": 118
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9668897843413676e-05,
      "loss": 1.1865,
      "step": 119
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9661197054116165e-05,
      "loss": 1.1506,
      "step": 120
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9653409281295053e-05,
      "loss": 1.2092,
      "step": 121
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9645534595066697e-05,
      "loss": 1.232,
      "step": 122
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.963757306632996e-05,
      "loss": 1.1755,
      "step": 123
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9629524766765593e-05,
      "loss": 1.1795,
      "step": 124
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.962138976883558e-05,
      "loss": 1.2398,
      "step": 125
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9613168145782468e-05,
      "loss": 1.2437,
      "step": 126
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9604859971628743e-05,
      "loss": 1.1782,
      "step": 127
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9596465321176136e-05,
      "loss": 1.1971,
      "step": 128
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.958798427000495e-05,
      "loss": 1.1401,
      "step": 129
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9579416894473407e-05,
      "loss": 1.1967,
      "step": 130
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.957076327171692e-05,
      "loss": 1.188,
      "step": 131
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.956202347964743e-05,
      "loss": 1.1636,
      "step": 132
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.955319759695269e-05,
      "loss": 1.1595,
      "step": 133
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9544285703095565e-05,
      "loss": 1.193,
      "step": 134
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9535287878313315e-05,
      "loss": 1.1201,
      "step": 135
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.952620420361686e-05,
      "loss": 1.2006,
      "step": 136
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9517034760790064e-05,
      "loss": 1.1913,
      "step": 137
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9507779632388997e-05,
      "loss": 1.23,
      "step": 138
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9498438901741186e-05,
      "loss": 1.2601,
      "step": 139
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9489012652944874e-05,
      "loss": 1.1785,
      "step": 140
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.947950097086825e-05,
      "loss": 1.1947,
      "step": 141
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.94699039411487e-05,
      "loss": 1.1873,
      "step": 142
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9460221650192016e-05,
      "loss": 1.2313,
      "step": 143
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.945045418517165e-05,
      "loss": 1.166,
      "step": 144
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.9440601634027892e-05,
      "loss": 1.2167,
      "step": 145
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.94306640854671e-05,
      "loss": 1.2201,
      "step": 146
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.9420641628960897e-05,
      "loss": 1.1424,
      "step": 147
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.9410534354745367e-05,
      "loss": 1.265,
      "step": 148
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.9400342353820244e-05,
      "loss": 1.1723,
      "step": 149
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.9390065717948084e-05,
      "loss": 1.153,
      "step": 150
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9379704539653443e-05,
      "loss": 1.2422,
      "step": 151
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9369258912222052e-05,
      "loss": 1.2091,
      "step": 152
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9358728929699966e-05,
      "loss": 1.1898,
      "step": 153
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.9348114686892722e-05,
      "loss": 1.1857,
      "step": 154
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.9337416279364486e-05,
      "loss": 1.1432,
      "step": 155
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.9326633803437197e-05,
      "loss": 1.2366,
      "step": 156
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.931576735618968e-05,
      "loss": 1.2082,
      "step": 157
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.9304817035456804e-05,
      "loss": 1.1841,
      "step": 158
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.929378293982857e-05,
      "loss": 1.203,
      "step": 159
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.928266516864925e-05,
      "loss": 1.187,
      "step": 160
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.9271463822016465e-05,
      "loss": 1.2004,
      "step": 161
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.926017900078031e-05,
      "loss": 1.1816,
      "step": 162
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.924881080654243e-05,
      "loss": 1.1433,
      "step": 163
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.9237359341655108e-05,
      "loss": 1.2313,
      "step": 164
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.922582470922034e-05,
      "loss": 1.2453,
      "step": 165
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.9214207013088935e-05,
      "loss": 1.1982,
      "step": 166
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.920250635785953e-05,
      "loss": 1.2433,
      "step": 167
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.9190722848877683e-05,
      "loss": 1.1923,
      "step": 168
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.9178856592234927e-05,
      "loss": 1.1939,
      "step": 169
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.916690769476779e-05,
      "loss": 1.1539,
      "step": 170
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.9154876264056863e-05,
      "loss": 1.1694,
      "step": 171
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.9142762408425797e-05,
      "loss": 1.1914,
      "step": 172
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.9130566236940363e-05,
      "loss": 1.2548,
      "step": 173
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.911828785940745e-05,
      "loss": 1.1776,
      "step": 174
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.910592738637407e-05,
      "loss": 1.1697,
      "step": 175
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.9093484929126383e-05,
      "loss": 1.2309,
      "step": 176
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.908096059968869e-05,
      "loss": 1.1831,
      "step": 177
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.9068354510822402e-05,
      "loss": 1.228,
      "step": 178
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.905566677602506e-05,
      "loss": 1.1951,
      "step": 179
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.904289750952928e-05,
      "loss": 1.205,
      "step": 180
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.9030046826301746e-05,
      "loss": 1.2015,
      "step": 181
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.9017114842042174e-05,
      "loss": 1.2009,
      "step": 182
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.900410167318226e-05,
      "loss": 1.1674,
      "step": 183
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8991007436884633e-05,
      "loss": 1.2286,
      "step": 184
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.897783225104181e-05,
      "loss": 1.171,
      "step": 185
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8964576234275123e-05,
      "loss": 1.1635,
      "step": 186
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8951239505933663e-05,
      "loss": 1.1771,
      "step": 187
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.893782218609319e-05,
      "loss": 1.1829,
      "step": 188
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8924324395555066e-05,
      "loss": 1.1819,
      "step": 189
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8910746255845168e-05,
      "loss": 1.1657,
      "step": 190
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.8897087889212772e-05,
      "loss": 1.1807,
      "step": 191
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.8883349418629487e-05,
      "loss": 1.1763,
      "step": 192
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.886953096778811e-05,
      "loss": 1.1322,
      "step": 193
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.885563266110155e-05,
      "loss": 1.1607,
      "step": 194
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.8841654623701673e-05,
      "loss": 1.2171,
      "step": 195
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.8827596981438202e-05,
      "loss": 1.1621,
      "step": 196
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.8813459860877575e-05,
      "loss": 1.1894,
      "step": 197
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.8799243389301796e-05,
      "loss": 1.1554,
      "step": 198
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.87849476947073e-05,
      "loss": 1.152,
      "step": 199
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.8770572905803806e-05,
      "loss": 1.1941,
      "step": 200
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.8756119152013134e-05,
      "loss": 1.2741,
      "step": 201
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.8741586563468064e-05,
      "loss": 1.164,
      "step": 202
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.8726975271011163e-05,
      "loss": 1.1811,
      "step": 203
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.8712285406193585e-05,
      "loss": 1.2335,
      "step": 204
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.869751710127392e-05,
      "loss": 1.1534,
      "step": 205
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.868267048921697e-05,
      "loss": 1.1557,
      "step": 206
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.866774570369257e-05,
      "loss": 1.152,
      "step": 207
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.8652742879074384e-05,
      "loss": 1.1982,
      "step": 208
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.8637662150438695e-05,
      "loss": 1.1531,
      "step": 209
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.8622503653563173e-05,
      "loss": 1.2339,
      "step": 210
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.8607267524925684e-05,
      "loss": 1.1669,
      "step": 211
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.8591953901703028e-05,
      "loss": 1.251,
      "step": 212
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.8576562921769727e-05,
      "loss": 1.2002,
      "step": 213
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.8561094723696776e-05,
      "loss": 1.171,
      "step": 214
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.8545549446750392e-05,
      "loss": 1.2238,
      "step": 215
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.8529927230890757e-05,
      "loss": 1.2063,
      "step": 216
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.8514228216770784e-05,
      "loss": 1.1634,
      "step": 217
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.8498452545734808e-05,
      "loss": 1.2448,
      "step": 218
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8482600359817344e-05,
      "loss": 1.2604,
      "step": 219
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8466671801741812e-05,
      "loss": 1.1813,
      "step": 220
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.845066701491922e-05,
      "loss": 1.1993,
      "step": 221
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.843458614344691e-05,
      "loss": 1.1974,
      "step": 222
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.841842933210723e-05,
      "loss": 1.1981,
      "step": 223
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.840219672636626e-05,
      "loss": 1.1443,
      "step": 224
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.8385888472372474e-05,
      "loss": 1.1971,
      "step": 225
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.836950471695544e-05,
      "loss": 1.1828,
      "step": 226
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.8353045607624494e-05,
      "loss": 1.2389,
      "step": 227
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.833651129256742e-05,
      "loss": 1.1886,
      "step": 228
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.8319901920649096e-05,
      "loss": 1.1794,
      "step": 229
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.8303217641410174e-05,
      "loss": 1.1842,
      "step": 230
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.828645860506573e-05,
      "loss": 1.2395,
      "step": 231
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.8269624962503895e-05,
      "loss": 1.189,
      "step": 232
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.825271686528452e-05,
      "loss": 1.2021,
      "step": 233
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.8235734465637794e-05,
      "loss": 1.1385,
      "step": 234
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.8218677916462882e-05,
      "loss": 1.1857,
      "step": 235
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.8201547371326553e-05,
      "loss": 1.1745,
      "step": 236
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.8184342984461766e-05,
      "loss": 1.2081,
      "step": 237
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.816706491076634e-05,
      "loss": 1.1183,
      "step": 238
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.8149713305801505e-05,
      "loss": 1.2019,
      "step": 239
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.8132288325790518e-05,
      "loss": 1.1882,
      "step": 240
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.8114790127617274e-05,
      "loss": 1.2139,
      "step": 241
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.809721886882487e-05,
      "loss": 1.1907,
      "step": 242
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.8079574707614202e-05,
      "loss": 1.1665,
      "step": 243
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.806185780284253e-05,
      "loss": 1.2126,
      "step": 244
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.8044068314022057e-05,
      "loss": 1.2309,
      "step": 245
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.802620640131848e-05,
      "loss": 1.2309,
      "step": 246
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.800827222554957e-05,
      "loss": 1.2466,
      "step": 247
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.79902659481837e-05,
      "loss": 1.2281,
      "step": 248
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.797218773133841e-05,
      "loss": 1.208,
      "step": 249
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.7954037737778927e-05,
      "loss": 1.132,
      "step": 250
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.7935816130916724e-05,
      "loss": 1.138,
      "step": 251
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.7917523074808024e-05,
      "loss": 1.2041,
      "step": 252
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.789915873415235e-05,
      "loss": 1.2736,
      "step": 253
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7880723274291023e-05,
      "loss": 1.2191,
      "step": 254
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.786221686120567e-05,
      "loss": 1.1143,
      "step": 255
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.7843639661516743e-05,
      "loss": 1.2028,
      "step": 256
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.7824991842482014e-05,
      "loss": 1.1834,
      "step": 257
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.7806273571995066e-05,
      "loss": 1.1844,
      "step": 258
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.7787485018583792e-05,
      "loss": 1.1251,
      "step": 259
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.7768626351408856e-05,
      "loss": 1.1508,
      "step": 260
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.7749697740262197e-05,
      "loss": 1.184,
      "step": 261
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.7730699355565478e-05,
      "loss": 1.2147,
      "step": 262
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.7711631368368564e-05,
      "loss": 1.1857,
      "step": 263
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.769249395034797e-05,
      "loss": 1.2376,
      "step": 264
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.7673287273805342e-05,
      "loss": 1.1883,
      "step": 265
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.7654011511665875e-05,
      "loss": 1.213,
      "step": 266
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.7634666837476765e-05,
      "loss": 1.1331,
      "step": 267
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.761525342540566e-05,
      "loss": 1.201,
      "step": 268
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.7595771450239075e-05,
      "loss": 1.1785,
      "step": 269
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.757622108738083e-05,
      "loss": 1.1081,
      "step": 270
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.7556602512850466e-05,
      "loss": 1.1684,
      "step": 271
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1080,
  "num_train_epochs": 4,
  "save_steps": 271,
  "total_flos": 2.0804857960196997e+18,
  "trial_name": null,
  "trial_params": null
}