{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9990403071017274,
  "eval_steps": 40.0,
  "global_step": 347,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.1999938524606463e-05,
      "loss": 2.6336,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1999754099685592e-05,
      "loss": 2.287,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1999446729016585e-05,
      "loss": 1.9992,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1999016418898025e-05,
      "loss": 1.9326,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1998463178147731e-05,
      "loss": 1.773,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1997787018102607e-05,
      "loss": 1.6502,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1996987952618384e-05,
      "loss": 1.5299,
      "step": 7
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1996065998069355e-05,
      "loss": 1.5532,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.1995021173348024e-05,
      "loss": 1.4142,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.1993853499864725e-05,
      "loss": 1.5215,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.1992563001547192e-05,
      "loss": 1.4892,
      "step": 11
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.1991149704840053e-05,
      "loss": 1.5194,
      "step": 12
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.1989613638704296e-05,
      "loss": 1.6177,
      "step": 13
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.1987954834616683e-05,
      "loss": 1.468,
      "step": 14
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.1986173326569089e-05,
      "loss": 1.4379,
      "step": 15
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.1984269151067816e-05,
      "loss": 1.4556,
      "step": 16
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.1982242347132847e-05,
      "loss": 1.4937,
      "step": 17
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.1980092956297035e-05,
      "loss": 1.581,
      "step": 18
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.1977821022605263e-05,
      "loss": 1.4928,
      "step": 19
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.1975426592613539e-05,
      "loss": 1.4856,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.1972909715388035e-05,
      "loss": 1.5337,
      "step": 21
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.1970270442504094e-05,
      "loss": 1.4193,
      "step": 22
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.196750882804516e-05,
      "loss": 1.5631,
      "step": 23
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.196462492860168e-05,
      "loss": 1.4632,
      "step": 24
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.1961618803269935e-05,
      "loss": 1.5262,
      "step": 25
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.195849051365084e-05,
      "loss": 1.4775,
      "step": 26
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.1955240123848675e-05,
      "loss": 1.2899,
      "step": 27
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.1951867700469767e-05,
      "loss": 1.5403,
      "step": 28
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.194837331262114e-05,
      "loss": 1.4858,
      "step": 29
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.1944757031909076e-05,
      "loss": 1.4674,
      "step": 30
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.1941018932437674e-05,
      "loss": 1.4887,
      "step": 31
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.193715909080731e-05,
      "loss": 1.5058,
      "step": 32
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.1933177586113079e-05,
      "loss": 1.3734,
      "step": 33
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.192907449994317e-05,
      "loss": 1.4409,
      "step": 34
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.1924849916377196e-05,
      "loss": 1.3174,
      "step": 35
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.1920503921984469e-05,
      "loss": 1.4628,
      "step": 36
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.1916036605822227e-05,
      "loss": 1.3974,
      "step": 37
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.191144805943381e-05,
      "loss": 1.5108,
      "step": 38
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.1906738376846786e-05,
      "loss": 1.4599,
      "step": 39
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.1901907654571013e-05,
      "loss": 1.5997,
      "step": 40
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.1896955991596679e-05,
      "loss": 1.3063,
      "step": 41
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.189188348939226e-05,
      "loss": 1.4545,
      "step": 42
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.1886690251902445e-05,
      "loss": 1.3852,
      "step": 43
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.188137638554601e-05,
      "loss": 1.4043,
      "step": 44
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.1875941999213624e-05,
      "loss": 1.3884,
      "step": 45
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.187038720426564e-05,
      "loss": 1.4439,
      "step": 46
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1864712114529786e-05,
      "loss": 1.4599,
      "step": 47
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1858916846298859e-05,
      "loss": 1.4984,
      "step": 48
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1853001518328323e-05,
      "loss": 1.3452,
      "step": 49
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1846966251833882e-05,
      "loss": 1.3936,
      "step": 50
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.1840811170488998e-05,
      "loss": 1.4235,
      "step": 51
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.1834536400422352e-05,
      "loss": 1.4595,
      "step": 52
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.1828142070215266e-05,
      "loss": 1.4447,
      "step": 53
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.1821628310899061e-05,
      "loss": 1.5234,
      "step": 54
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.1814995255952373e-05,
      "loss": 1.4176,
      "step": 55
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.1808243041298429e-05,
      "loss": 1.4023,
      "step": 56
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.1801371805302244e-05,
      "loss": 1.4635,
      "step": 57
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.1794381688767794e-05,
      "loss": 1.4292,
      "step": 58
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.1787272834935138e-05,
      "loss": 1.4894,
      "step": 59
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.1780045389477471e-05,
      "loss": 1.2991,
      "step": 60
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.1772699500498142e-05,
      "loss": 1.4466,
      "step": 61
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.1765235318527625e-05,
      "loss": 1.41,
      "step": 62
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.1757652996520427e-05,
      "loss": 1.3542,
      "step": 63
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.174995268985196e-05,
      "loss": 1.4283,
      "step": 64
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.1742134556315348e-05,
      "loss": 1.4417,
      "step": 65
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.1734198756118204e-05,
      "loss": 1.4124,
      "step": 66
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.1726145451879344e-05,
      "loss": 1.4461,
      "step": 67
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1717974808625447e-05,
      "loss": 1.3768,
      "step": 68
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1709686993787686e-05,
      "loss": 1.4599,
      "step": 69
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1701282177198285e-05,
      "loss": 1.5219,
      "step": 70
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1692760531087049e-05,
      "loss": 1.4588,
      "step": 71
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.1684122230077822e-05,
      "loss": 1.3674,
      "step": 72
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.1675367451184928e-05,
      "loss": 1.422,
      "step": 73
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.1666496373809524e-05,
      "loss": 1.471,
      "step": 74
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.1657509179735934e-05,
      "loss": 1.5206,
      "step": 75
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.164840605312792e-05,
      "loss": 1.4817,
      "step": 76
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.1639187180524918e-05,
      "loss": 1.4353,
      "step": 77
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.1629852750838196e-05,
      "loss": 1.4303,
      "step": 78
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.1620402955347004e-05,
      "loss": 1.4781,
      "step": 79
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.1610837987694637e-05,
      "loss": 1.4428,
      "step": 80
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.1601158043884482e-05,
      "loss": 1.4312,
      "step": 81
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.159136332227599e-05,
      "loss": 1.4551,
      "step": 82
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.1581454023580616e-05,
      "loss": 1.4503,
      "step": 83
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.1571430350857704e-05,
      "loss": 1.4528,
      "step": 84
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.156129250951033e-05,
      "loss": 1.3285,
      "step": 85
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.1551040707281093e-05,
      "loss": 1.4585,
      "step": 86
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.1540675154247845e-05,
      "loss": 1.304,
      "step": 87
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.153019606281941e-05,
      "loss": 1.4546,
      "step": 88
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.1519603647731208e-05,
      "loss": 1.3858,
      "step": 89
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.1508898126040866e-05,
      "loss": 1.3672,
      "step": 90
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.1498079717123774e-05,
      "loss": 1.3506,
      "step": 91
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.1487148642668576e-05,
      "loss": 1.4802,
      "step": 92
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.1476105126672646e-05,
      "loss": 1.41,
      "step": 93
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.1464949395437477e-05,
      "loss": 1.4322,
      "step": 94
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.145368167756406e-05,
      "loss": 1.394,
      "step": 95
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.144230220394819e-05,
      "loss": 1.42,
      "step": 96
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.1430811207775741e-05,
      "loss": 1.3121,
      "step": 97
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.1419208924517886e-05,
      "loss": 1.3474,
      "step": 98
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.1407495591926263e-05,
      "loss": 1.2316,
      "step": 99
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.1395671450028118e-05,
      "loss": 1.4253,
      "step": 100
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.138373674112138e-05,
      "loss": 1.4087,
      "step": 101
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.1371691709769685e-05,
      "loss": 1.3573,
      "step": 102
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.1359536602797386e-05,
      "loss": 1.4106,
      "step": 103
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.1347271669284475e-05,
      "loss": 1.3924,
      "step": 104
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.133489716056149e-05,
      "loss": 1.4377,
      "step": 105
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.1322413330204362e-05,
      "loss": 1.4068,
      "step": 106
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.1309820434029221e-05,
      "loss": 1.3437,
      "step": 107
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.1297118730087148e-05,
      "loss": 1.4283,
      "step": 108
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.1284308478658893e-05,
      "loss": 1.4738,
      "step": 109
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.127138994224954e-05,
      "loss": 1.4097,
      "step": 110
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.1258363385583122e-05,
      "loss": 1.3924,
      "step": 111
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.1245229075597209e-05,
      "loss": 1.4808,
      "step": 112
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.1231987281437423e-05,
      "loss": 1.3838,
      "step": 113
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.1218638274451936e-05,
      "loss": 1.4327,
      "step": 114
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.1205182328185897e-05,
      "loss": 1.3547,
      "step": 115
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.1191619718375839e-05,
      "loss": 1.3291,
      "step": 116
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.117795072294402e-05,
      "loss": 1.304,
      "step": 117
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.116417562199273e-05,
      "loss": 1.3836,
      "step": 118
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.1150294697798553e-05,
      "loss": 1.4498,
      "step": 119
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.1136308234806587e-05,
      "loss": 1.5027,
      "step": 120
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.1122216519624596e-05,
      "loss": 1.418,
      "step": 121
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.1108019841017165e-05,
      "loss": 1.3934,
      "step": 122
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.1093718489899758e-05,
      "loss": 1.3492,
      "step": 123
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.1079312759332772e-05,
      "loss": 1.3846,
      "step": 124
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.106480294451553e-05,
      "loss": 1.402,
      "step": 125
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.1050189342780217e-05,
      "loss": 1.4242,
      "step": 126
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.103547225358581e-05,
      "loss": 1.4199,
      "step": 127
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.1020651978511924e-05,
      "loss": 1.4253,
      "step": 128
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.1005728821252641e-05,
      "loss": 1.4173,
      "step": 129
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.0990703087610283e-05,
      "loss": 1.4704,
      "step": 130
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.0975575085489143e-05,
      "loss": 1.4215,
      "step": 131
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.0960345124889188e-05,
      "loss": 1.4109,
      "step": 132
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.0945013517899686e-05,
      "loss": 1.4212,
      "step": 133
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.092958057869283e-05,
      "loss": 1.393,
      "step": 134
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.0914046623517292e-05,
      "loss": 1.3462,
      "step": 135
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.089841197069174e-05,
      "loss": 1.4998,
      "step": 136
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.0882676940598322e-05,
      "loss": 1.4839,
      "step": 137
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.0866841855676088e-05,
      "loss": 1.4445,
      "step": 138
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.08509070404144e-05,
      "loss": 1.419,
      "step": 139
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.0834872821346274e-05,
      "loss": 1.3788,
      "step": 140
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.0818739527041681e-05,
      "loss": 1.4787,
      "step": 141
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.0802507488100828e-05,
      "loss": 1.3911,
      "step": 142
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.0786177037147378e-05,
      "loss": 1.343,
      "step": 143
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.0769748508821627e-05,
      "loss": 1.4525,
      "step": 144
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.075322223977366e-05,
      "loss": 1.4751,
      "step": 145
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.0736598568656436e-05,
      "loss": 1.3912,
      "step": 146
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.071987783611887e-05,
      "loss": 1.3316,
      "step": 147
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.0703060384798828e-05,
      "loss": 1.465,
      "step": 148
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.0686146559316125e-05,
      "loss": 1.3809,
      "step": 149
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.066913670626545e-05,
      "loss": 1.4012,
      "step": 150
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.0652031174209279e-05,
      "loss": 1.5069,
      "step": 151
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.0634830313670714e-05,
      "loss": 1.3887,
      "step": 152
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.061753447712631e-05,
      "loss": 1.3444,
      "step": 153
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.0600144018998857e-05,
      "loss": 1.5074,
      "step": 154
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.0582659295650102e-05,
      "loss": 1.426,
      "step": 155
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.0565080665373467e-05,
      "loss": 1.3974,
      "step": 156
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.0547408488386689e-05,
      "loss": 1.507,
      "step": 157
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.0529643126824442e-05,
      "loss": 1.2848,
      "step": 158
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0511784944730928e-05,
      "loss": 1.3705,
      "step": 159
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.04938343080524e-05,
      "loss": 1.372,
      "step": 160
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0475791584629683e-05,
      "loss": 1.3737,
      "step": 161
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0457657144190614e-05,
      "loss": 1.4055,
      "step": 162
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0439431358342477e-05,
      "loss": 1.4183,
      "step": 163
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0421114600564397e-05,
      "loss": 1.3938,
      "step": 164
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0402707246199669e-05,
      "loss": 1.4279,
      "step": 165
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0384209672448079e-05,
      "loss": 1.3658,
      "step": 166
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0365622258358166e-05,
      "loss": 1.4269,
      "step": 167
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0346945384819466e-05,
      "loss": 1.3599,
      "step": 168
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0328179434554692e-05,
      "loss": 1.3039,
      "step": 169
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0309324792111906e-05,
      "loss": 1.4862,
      "step": 170
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0290381843856627e-05,
      "loss": 1.4285,
      "step": 171
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0271350977963922e-05,
      "loss": 1.3973,
      "step": 172
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0252232584410449e-05,
      "loss": 1.2789,
      "step": 173
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0233027054966462e-05,
      "loss": 1.3758,
      "step": 174
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0213734783187791e-05,
      "loss": 1.343,
      "step": 175
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0194356164407765e-05,
      "loss": 1.4214,
      "step": 176
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0174891595729128e-05,
      "loss": 1.3872,
      "step": 177
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.015534147601588e-05,
      "loss": 1.3549,
      "step": 178
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.0135706205885127e-05,
      "loss": 1.4698,
      "step": 179
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.0115986187698855e-05,
      "loss": 1.4067,
      "step": 180
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.0096181825555686e-05,
      "loss": 1.4373,
      "step": 181
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.007629352528261e-05,
      "loss": 1.3448,
      "step": 182
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.0056321694426653e-05,
      "loss": 1.474,
      "step": 183
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.0036266742246536e-05,
      "loss": 1.3803,
      "step": 184
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.0016129079704286e-05,
      "loss": 1.337,
      "step": 185
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.995909119456813e-06,
      "loss": 1.3436,
      "step": 186
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.975607275847452e-06,
      "loss": 1.2478,
      "step": 187
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.95522396489748e-06,
      "loss": 1.3558,
      "step": 188
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.934759604297585e-06,
      "loss": 1.332,
      "step": 189
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.91421461339931e-06,
      "loss": 1.3528,
      "step": 190
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.893589413206449e-06,
      "loss": 1.4392,
      "step": 191
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.87288442636644e-06,
      "loss": 1.4533,
      "step": 192
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.852100077161686e-06,
      "loss": 1.3014,
      "step": 193
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.831236791500866e-06,
      "loss": 1.4457,
      "step": 194
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.810294996910218e-06,
      "loss": 1.3272,
      "step": 195
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.789275122524762e-06,
      "loss": 1.4885,
      "step": 196
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.76817759907951e-06,
      "loss": 1.3802,
      "step": 197
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.747002858900654e-06,
      "loss": 1.3995,
      "step": 198
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.725751335896684e-06,
      "loss": 1.3889,
      "step": 199
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.704423465549514e-06,
      "loss": 1.404,
      "step": 200
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.683019684905555e-06,
      "loss": 1.4303,
      "step": 201
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.66154043256675e-06,
      "loss": 1.4106,
      "step": 202
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.639986148681595e-06,
      "loss": 1.3556,
      "step": 203
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.618357274936122e-06,
      "loss": 1.3358,
      "step": 204
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.596654254544837e-06,
      "loss": 1.3653,
      "step": 205
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.574877532241646e-06,
      "loss": 1.3943,
      "step": 206
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.553027554270741e-06,
      "loss": 1.2431,
      "step": 207
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.531104768377454e-06,
      "loss": 1.4285,
      "step": 208
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.509109623799082e-06,
      "loss": 1.4397,
      "step": 209
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.48704257125568e-06,
      "loss": 1.3748,
      "step": 210
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.464904062940827e-06,
      "loss": 1.4576,
      "step": 211
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.442694552512365e-06,
      "loss": 1.3955,
      "step": 212
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.420414495083084e-06,
      "loss": 1.5999,
      "step": 213
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.398064347211423e-06,
      "loss": 1.3066,
      "step": 214
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.37564456689209e-06,
      "loss": 1.4238,
      "step": 215
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.353155613546693e-06,
      "loss": 1.3093,
      "step": 216
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.330597948014321e-06,
      "loss": 1.3458,
      "step": 217
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.30797203254209e-06,
      "loss": 1.2949,
      "step": 218
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.28527833077569e-06,
      "loss": 1.3479,
      "step": 219
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.26251730774987e-06,
      "loss": 1.4428,
      "step": 220
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.23968942987891e-06,
      "loss": 1.4494,
      "step": 221
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.216795164947067e-06,
      "loss": 1.513,
      "step": 222
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.193834982098994e-06,
      "loss": 1.3999,
      "step": 223
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.170809351830119e-06,
      "loss": 1.4124,
      "step": 224
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.147718745976995e-06,
      "loss": 1.4184,
      "step": 225
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.124563637707654e-06,
      "loss": 1.4625,
      "step": 226
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.101344501511893e-06,
      "loss": 1.3812,
      "step": 227
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.078061813191557e-06,
      "loss": 1.3352,
      "step": 228
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.054716049850788e-06,
      "loss": 1.373,
      "step": 229
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.03130768988625e-06,
      "loss": 1.4279,
      "step": 230
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.007837212977322e-06,
      "loss": 1.3979,
      "step": 231
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.984305100076274e-06,
      "loss": 1.3694,
      "step": 232
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.960711833398402e-06,
      "loss": 1.4064,
      "step": 233
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.937057896412163e-06,
      "loss": 1.4104,
      "step": 234
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.913343773829249e-06,
      "loss": 1.3844,
      "step": 235
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.889569951594666e-06,
      "loss": 1.3653,
      "step": 236
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.865736916876774e-06,
      "loss": 1.427,
      "step": 237
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.841845158057302e-06,
      "loss": 1.3435,
      "step": 238
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.817895164721344e-06,
      "loss": 1.4427,
      "step": 239
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.793887427647316e-06,
      "loss": 1.3507,
      "step": 240
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.76982243879692e-06,
      "loss": 1.424,
      "step": 241
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.745700691305039e-06,
      "loss": 1.4637,
      "step": 242
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.721522679469645e-06,
      "loss": 1.3721,
      "step": 243
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.697288898741672e-06,
      "loss": 1.3589,
      "step": 244
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.672999845714856e-06,
      "loss": 1.4479,
      "step": 245
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.648656018115557e-06,
      "loss": 1.4105,
      "step": 246
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.624257914792573e-06,
      "loss": 1.3511,
      "step": 247
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.599806035706903e-06,
      "loss": 1.3963,
      "step": 248
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.575300881921513e-06,
      "loss": 1.3543,
      "step": 249
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.550742955591057e-06,
      "loss": 1.4107,
      "step": 250
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.526132759951598e-06,
      "loss": 1.399,
      "step": 251
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.501470799310294e-06,
      "loss": 1.4452,
      "step": 252
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.476757579035049e-06,
      "loss": 1.4045,
      "step": 253
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.451993605544183e-06,
      "loss": 1.4063,
      "step": 254
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.427179386296034e-06,
      "loss": 1.3813,
      "step": 255
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.402315429778566e-06,
      "loss": 1.4301,
      "step": 256
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.377402245498948e-06,
      "loss": 1.361,
      "step": 257
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.352440343973117e-06,
      "loss": 1.4488,
      "step": 258
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.327430236715313e-06,
      "loss": 1.4259,
      "step": 259
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.3023724362276e-06,
      "loss": 1.3858,
      "step": 260
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.277267455989356e-06,
      "loss": 1.429,
      "step": 261
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.252115810446766e-06,
      "loss": 1.4905,
      "step": 262
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.22691801500226e-06,
      "loss": 1.4224,
      "step": 263
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.201674586003973e-06,
      "loss": 1.4441,
      "step": 264
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.176386040735146e-06,
      "loss": 1.3486,
      "step": 265
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.15105289740354e-06,
      "loss": 1.3677,
      "step": 266
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.125675675130805e-06,
      "loss": 1.4015,
      "step": 267
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.100254893941847e-06,
      "loss": 1.434,
      "step": 268
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.07479107475418e-06,
      "loss": 1.4285,
      "step": 269
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.049284739367235e-06,
      "loss": 1.3749,
      "step": 270
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.023736410451683e-06,
      "loss": 1.3624,
      "step": 271
    },
    {
      "epoch": 0.78,
      "learning_rate": 7.998146611538716e-06,
      "loss": 1.4756,
      "step": 272
    },
    {
      "epoch": 0.79,
      "learning_rate": 7.972515867009318e-06,
      "loss": 1.4431,
      "step": 273
    },
    {
      "epoch": 0.79,
      "learning_rate": 7.946844702083526e-06,
      "loss": 1.376,
      "step": 274
    },
    {
      "epoch": 0.79,
      "learning_rate": 7.921133642809661e-06,
      "loss": 1.3855,
      "step": 275
    },
    {
      "epoch": 0.79,
      "learning_rate": 7.895383216053555e-06,
      "loss": 1.3976,
      "step": 276
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.869593949487746e-06,
      "loss": 1.3488,
      "step": 277
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.843766371580668e-06,
      "loss": 1.2903,
      "step": 278
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.817901011585828e-06,
      "loss": 1.3089,
      "step": 279
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.791998399530957e-06,
      "loss": 1.3641,
      "step": 280
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.766059066207142e-06,
      "loss": 1.3597,
      "step": 281
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.740083543157959e-06,
      "loss": 1.3307,
      "step": 282
    },
    {
      "epoch": 0.81,
      "learning_rate": 7.714072362668576e-06,
      "loss": 1.3195,
      "step": 283
    },
    {
      "epoch": 0.82,
      "learning_rate": 7.68802605775484e-06,
      "loss": 1.4422,
      "step": 284
    },
    {
      "epoch": 0.82,
      "learning_rate": 7.661945162152376e-06,
      "loss": 1.3965,
      "step": 285
    },
    {
      "epoch": 0.82,
      "learning_rate": 7.635830210305614e-06,
      "loss": 1.4662,
      "step": 286
    },
    {
      "epoch": 0.83,
      "learning_rate": 7.609681737356874e-06,
      "loss": 1.3645,
      "step": 287
    },
    {
      "epoch": 0.83,
      "learning_rate": 7.5835002791353755e-06,
      "loss": 1.3993,
      "step": 288
    },
    {
      "epoch": 0.83,
      "learning_rate": 7.557286372146269e-06,
      "loss": 1.3555,
      "step": 289
    },
    {
      "epoch": 0.83,
      "learning_rate": 7.531040553559637e-06,
      "loss": 1.4398,
      "step": 290
    },
    {
      "epoch": 0.84,
      "learning_rate": 7.5047633611994865e-06,
      "loss": 1.3965,
      "step": 291
    },
    {
      "epoch": 0.84,
      "learning_rate": 7.478455333532731e-06,
      "loss": 1.3462,
      "step": 292
    },
    {
      "epoch": 0.84,
      "learning_rate": 7.45211700965816e-06,
      "loss": 1.3702,
      "step": 293
    },
    {
      "epoch": 0.85,
      "learning_rate": 7.425748929295378e-06,
      "loss": 1.3787,
      "step": 294
    },
    {
      "epoch": 0.85,
      "learning_rate": 7.399351632773757e-06,
      "loss": 1.3006,
      "step": 295
    },
    {
      "epoch": 0.85,
      "learning_rate": 7.372925661021363e-06,
      "loss": 1.2651,
      "step": 296
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.346471555553865e-06,
      "loss": 1.3216,
      "step": 297
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.319989858463447e-06,
      "loss": 1.4034,
      "step": 298
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.2934811124076895e-06,
      "loss": 1.2659,
      "step": 299
    },
    {
      "epoch": 0.86,
      "learning_rate": 7.26694586059846e-06,
      "loss": 1.3787,
      "step": 300
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.240384646790773e-06,
      "loss": 1.3123,
      "step": 301
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.213798015271654e-06,
      "loss": 1.4925,
      "step": 302
    },
    {
      "epoch": 0.87,
      "learning_rate": 7.18718651084898e-06,
      "loss": 1.3573,
      "step": 303
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.160550678840318e-06,
      "loss": 1.4239,
      "step": 304
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.13389106506176e-06,
      "loss": 1.3931,
      "step": 305
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.107208215816714e-06,
      "loss": 1.2298,
      "step": 306
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.0805026778847374e-06,
      "loss": 1.3927,
      "step": 307
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.053774998510315e-06,
      "loss": 1.3206,
      "step": 308
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.027025725391647e-06,
      "loss": 1.4185,
      "step": 309
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.000255406669432e-06,
      "loss": 1.3403,
      "step": 310
    },
    {
      "epoch": 0.9,
      "learning_rate": 6.973464590915631e-06,
      "loss": 1.3842,
      "step": 311
    },
    {
      "epoch": 0.9,
      "learning_rate": 6.9466538271222195e-06,
      "loss": 1.437,
      "step": 312
    },
    {
      "epoch": 0.9,
      "learning_rate": 6.9198236646899554e-06,
      "loss": 1.3241,
      "step": 313
    },
    {
      "epoch": 0.9,
      "learning_rate": 6.8929746534170995e-06,
      "loss": 1.4092,
      "step": 314
    },
    {
      "epoch": 0.91,
      "learning_rate": 6.866107343488164e-06,
      "loss": 1.3778,
      "step": 315
    },
    {
      "epoch": 0.91,
      "learning_rate": 6.839222285462635e-06,
      "loss": 1.3312,
      "step": 316
    },
    {
      "epoch": 0.91,
      "learning_rate": 6.812320030263681e-06,
      "loss": 1.3811,
      "step": 317
    },
    {
      "epoch": 0.92,
      "learning_rate": 6.7854011291668854e-06,
      "loss": 1.3,
      "step": 318
    },
    {
      "epoch": 0.92,
      "learning_rate": 6.75846613378892e-06,
      "loss": 1.2676,
      "step": 319
    },
    {
      "epoch": 0.92,
      "learning_rate": 6.7315155960762695e-06,
      "loss": 1.3647,
      "step": 320
    },
    {
      "epoch": 0.92,
      "learning_rate": 6.704550068293903e-06,
      "loss": 1.323,
      "step": 321
    },
    {
      "epoch": 0.93,
      "learning_rate": 6.677570103013964e-06,
      "loss": 1.3604,
      "step": 322
    },
    {
      "epoch": 0.93,
      "learning_rate": 6.650576253104449e-06,
      "loss": 1.3911,
      "step": 323
    },
    {
      "epoch": 0.93,
      "learning_rate": 6.623569071717872e-06,
      "loss": 1.3921,
      "step": 324
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.596549112279932e-06,
      "loss": 1.376,
      "step": 325
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.56951692847818e-06,
      "loss": 1.4246,
      "step": 326
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.54247307425066e-06,
      "loss": 1.2893,
      "step": 327
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.5154181037745655e-06,
      "loss": 1.3598,
      "step": 328
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.488352571454883e-06,
      "loss": 1.5025,
      "step": 329
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.461277031913027e-06,
      "loss": 1.4079,
      "step": 330
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.434192039975484e-06,
      "loss": 1.4089,
      "step": 331
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.407098150662432e-06,
      "loss": 1.3471,
      "step": 332
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.379995919176371e-06,
      "loss": 1.398,
      "step": 333
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.352885900890754e-06,
      "loss": 1.2381,
      "step": 334
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.3257686513385925e-06,
      "loss": 1.3302,
      "step": 335
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.298644726201083e-06,
      "loss": 1.3804,
      "step": 336
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.271514681296216e-06,
      "loss": 1.3909,
      "step": 337
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.2443790725673895e-06,
      "loss": 1.4171,
      "step": 338
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.217238456072008e-06,
      "loss": 1.4425,
      "step": 339
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.190093387970102e-06,
      "loss": 1.3923,
      "step": 340
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.1629444245129175e-06,
      "loss": 1.361,
      "step": 341
    },
    {
      "epoch": 0.98,
      "learning_rate": 6.135792122031526e-06,
      "loss": 1.3116,
      "step": 342
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.10863703692542e-06,
      "loss": 1.3277,
      "step": 343
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.081479725651116e-06,
      "loss": 1.4442,
      "step": 344
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.054320744710744e-06,
      "loss": 1.4354,
      "step": 345
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.027160650640654e-06,
      "loss": 1.3192,
      "step": 346
    },
    {
      "epoch": 1.0,
      "learning_rate": 6e-06,
      "loss": 1.359,
      "step": 347
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.359147548675537,
      "eval_runtime": 99.243,
      "eval_samples_per_second": 9.31,
      "eval_steps_per_second": 1.169,
      "step": 347
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 694,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100.0,
  "total_flos": 1.9566296450924544e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}