{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 342,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.5454545454545457e-07,
      "loss": 0.5787,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.090909090909091e-07,
      "loss": 0.5734,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.3636363636363636e-06,
      "loss": 0.5844,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 0.5838,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 0.5903,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 0.5808,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.181818181818182e-06,
      "loss": 0.5783,
      "step": 7
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 0.5656,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.0909090909090915e-06,
      "loss": 0.5886,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.5859,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 5e-06,
      "loss": 0.5928,
      "step": 11
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.999887396857563e-06,
      "loss": 0.5914,
      "step": 12
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.999549597573825e-06,
      "loss": 0.5799,
      "step": 13
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.998986632578596e-06,
      "loss": 0.5848,
      "step": 14
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.998198552585177e-06,
      "loss": 0.5573,
      "step": 15
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.997185428585795e-06,
      "loss": 0.5755,
      "step": 16
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.995947351845207e-06,
      "loss": 0.6106,
      "step": 17
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.994484433892479e-06,
      "loss": 0.58,
      "step": 18
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9927968065109366e-06,
      "loss": 0.5987,
      "step": 19
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.990884621726298e-06,
      "loss": 0.586,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9887480517929746e-06,
      "loss": 0.5599,
      "step": 21
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.986387289178559e-06,
      "loss": 0.554,
      "step": 22
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.98380254654648e-06,
      "loss": 0.6069,
      "step": 23
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.980994056736854e-06,
      "loss": 0.5691,
      "step": 24
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.977962072745503e-06,
      "loss": 0.596,
      "step": 25
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.974706867701166e-06,
      "loss": 0.5851,
      "step": 26
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.971228734840897e-06,
      "loss": 0.5624,
      "step": 27
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.967527987483649e-06,
      "loss": 0.57,
      "step": 28
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.963604959002047e-06,
      "loss": 0.5583,
      "step": 29
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.95946000279236e-06,
      "loss": 0.5726,
      "step": 30
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9550934922426604e-06,
      "loss": 0.5351,
      "step": 31
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.950505820699199e-06,
      "loss": 0.5647,
      "step": 32
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.94569740143096e-06,
      "loss": 0.5378,
      "step": 33
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.940668667592439e-06,
      "loss": 0.5635,
      "step": 34
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.935420072184624e-06,
      "loss": 0.5524,
      "step": 35
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.929952088014182e-06,
      "loss": 0.5612,
      "step": 36
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.924265207650874e-06,
      "loss": 0.5716,
      "step": 37
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9183599433831785e-06,
      "loss": 0.5318,
      "step": 38
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9122368271721485e-06,
      "loss": 0.5796,
      "step": 39
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.905896410603485e-06,
      "loss": 0.5473,
      "step": 40
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.899339264837851e-06,
      "loss": 0.5349,
      "step": 41
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8925659805594214e-06,
      "loss": 0.5275,
      "step": 42
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.885577167922672e-06,
      "loss": 0.5447,
      "step": 43
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.878373456497416e-06,
      "loss": 0.5417,
      "step": 44
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8709554952120865e-06,
      "loss": 0.5434,
      "step": 45
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.863323952295286e-06,
      "loss": 0.5515,
      "step": 46
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.855479515215583e-06,
      "loss": 0.5487,
      "step": 47
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.847422890619594e-06,
      "loss": 0.5347,
      "step": 48
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.839154804268313e-06,
      "loss": 0.5329,
      "step": 49
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.830676000971747e-06,
      "loss": 0.5493,
      "step": 50
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.82198724452181e-06,
      "loss": 0.5371,
      "step": 51
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.813089317623527e-06,
      "loss": 0.5527,
      "step": 52
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.803983021824523e-06,
      "loss": 0.5309,
      "step": 53
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.794669177442815e-06,
      "loss": 0.5482,
      "step": 54
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.785148623492919e-06,
      "loss": 0.5365,
      "step": 55
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.77542221761027e-06,
      "loss": 0.5298,
      "step": 56
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.765490835973962e-06,
      "loss": 0.5429,
      "step": 57
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.755355373227818e-06,
      "loss": 0.5178,
      "step": 58
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.745016742399804e-06,
      "loss": 0.5352,
      "step": 59
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.734475874819776e-06,
      "loss": 0.6381,
      "step": 60
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.723733720035583e-06,
      "loss": 0.5051,
      "step": 61
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.712791245727534e-06,
      "loss": 0.5506,
      "step": 62
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.701649437621224e-06,
      "loss": 0.526,
      "step": 63
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.690309299398736e-06,
      "loss": 0.5269,
      "step": 64
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.678771852608232e-06,
      "loss": 0.5383,
      "step": 65
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.6670381365719215e-06,
      "loss": 0.5245,
      "step": 66
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.655109208292443e-06,
      "loss": 0.5227,
      "step": 67
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6429861423576454e-06,
      "loss": 0.5474,
      "step": 68
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.630670030843785e-06,
      "loss": 0.5241,
      "step": 69
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.618161983217148e-06,
      "loss": 0.5496,
      "step": 70
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.60546312623411e-06,
      "loss": 0.5027,
      "step": 71
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.592574603839632e-06,
      "loss": 0.5227,
      "step": 72
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.579497577064213e-06,
      "loss": 0.508,
      "step": 73
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.566233223919299e-06,
      "loss": 0.5501,
      "step": 74
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.552782739291168e-06,
      "loss": 0.5139,
      "step": 75
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.539147334833288e-06,
      "loss": 0.5273,
      "step": 76
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.525328238857173e-06,
      "loss": 0.4811,
      "step": 77
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.5113266962217276e-06,
      "loss": 0.5551,
      "step": 78
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.497143968221113e-06,
      "loss": 0.5462,
      "step": 79
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.48278133247112e-06,
      "loss": 0.5279,
      "step": 80
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.468240082794087e-06,
      "loss": 0.5103,
      "step": 81
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.453521529102337e-06,
      "loss": 0.5307,
      "step": 82
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.438626997280191e-06,
      "loss": 0.5336,
      "step": 83
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.42355782906452e-06,
      "loss": 0.5197,
      "step": 84
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.408315381923878e-06,
      "loss": 0.5348,
      "step": 85
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.392901028936224e-06,
      "loss": 0.5311,
      "step": 86
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.377316158665224e-06,
      "loss": 0.5326,
      "step": 87
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.3615621750351735e-06,
      "loss": 0.5498,
      "step": 88
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.345640497204522e-06,
      "loss": 0.5346,
      "step": 89
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.329552559438034e-06,
      "loss": 0.5194,
      "step": 90
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.313299810977589e-06,
      "loss": 0.4867,
      "step": 91
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.2968837159116275e-06,
      "loss": 0.5113,
      "step": 92
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.280305753043259e-06,
      "loss": 0.5331,
      "step": 93
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.263567415757058e-06,
      "loss": 0.5288,
      "step": 94
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.246670211884527e-06,
      "loss": 0.5039,
      "step": 95
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.229615663568266e-06,
      "loss": 0.4969,
      "step": 96
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.212405307124865e-06,
      "loss": 0.5385,
      "step": 97
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.195040692906497e-06,
      "loss": 0.4763,
      "step": 98
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.177523385161264e-06,
      "loss": 0.4884,
      "step": 99
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.159854961892287e-06,
      "loss": 0.5023,
      "step": 100
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.14203701471555e-06,
      "loss": 0.515,
      "step": 101
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.124071148716529e-06,
      "loss": 0.4655,
      "step": 102
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.105958982305598e-06,
      "loss": 0.5274,
      "step": 103
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.087702147072241e-06,
      "loss": 0.5055,
      "step": 104
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.069302287638071e-06,
      "loss": 0.5155,
      "step": 105
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.050761061508685e-06,
      "loss": 0.4972,
      "step": 106
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.032080138924341e-06,
      "loss": 0.5123,
      "step": 107
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.013261202709512e-06,
      "loss": 0.5226,
      "step": 108
    },
    {
      "epoch": 0.32,
      "learning_rate": 3.99430594812128e-06,
      "loss": 0.505,
      "step": 109
    },
    {
      "epoch": 0.32,
      "learning_rate": 3.97521608269663e-06,
      "loss": 0.474,
      "step": 110
    },
    {
      "epoch": 0.32,
      "learning_rate": 3.9559933260986315e-06,
      "loss": 0.4581,
      "step": 111
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.936639409961524e-06,
      "loss": 0.462,
      "step": 112
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.917156077734728e-06,
      "loss": 0.4895,
      "step": 113
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.89754508452579e-06,
      "loss": 0.5384,
      "step": 114
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.877808196942281e-06,
      "loss": 0.5052,
      "step": 115
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.857947192932649e-06,
      "loss": 0.4943,
      "step": 116
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.837963861626067e-06,
      "loss": 0.5023,
      "step": 117
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.817860003171257e-06,
      "loss": 0.5072,
      "step": 118
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.7976374285743257e-06,
      "loss": 0.4904,
      "step": 119
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.777297959535634e-06,
      "loss": 0.5094,
      "step": 120
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.7568434282856836e-06,
      "loss": 0.4843,
      "step": 121
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.7362756774200727e-06,
      "loss": 0.4817,
      "step": 122
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.7155965597335042e-06,
      "loss": 0.488,
      "step": 123
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.6948079380528866e-06,
      "loss": 0.5071,
      "step": 124
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.673911685069521e-06,
      "loss": 0.4791,
      "step": 125
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.65290968317041e-06,
      "loss": 0.4826,
      "step": 126
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.6318038242686822e-06,
      "loss": 0.4855,
      "step": 127
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.610596009633166e-06,
      "loss": 0.4842,
      "step": 128
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.5892881497171195e-06,
      "loss": 0.475,
      "step": 129
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.5678821639861305e-06,
      "loss": 0.5255,
      "step": 130
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.5463799807452076e-06,
      "loss": 0.5222,
      "step": 131
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.5247835369650718e-06,
      "loss": 0.4932,
      "step": 132
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.503094778107673e-06,
      "loss": 0.4955,
      "step": 133
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.4813156579509304e-06,
      "loss": 0.4913,
      "step": 134
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.4594481384127424e-06,
      "loss": 0.5343,
      "step": 135
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.437494189374242e-06,
      "loss": 0.4839,
      "step": 136
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.4154557885023493e-06,
      "loss": 0.493,
      "step": 137
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.393334921071618e-06,
      "loss": 0.5271,
      "step": 138
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.371133579785398e-06,
      "loss": 0.5181,
      "step": 139
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.348853764596324e-06,
      "loss": 0.5016,
      "step": 140
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.3264974825261597e-06,
      "loss": 0.4483,
      "step": 141
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.3040667474849967e-06,
      "loss": 0.4436,
      "step": 142
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.281563580089836e-06,
      "loss": 0.4697,
      "step": 143
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.25899000748257e-06,
      "loss": 0.4865,
      "step": 144
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.2363480631473666e-06,
      "loss": 0.4569,
      "step": 145
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.2136397867274926e-06,
      "loss": 0.4938,
      "step": 146
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.190867223841576e-06,
      "loss": 0.4755,
      "step": 147
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.1680324258993302e-06,
      "loss": 0.5011,
      "step": 148
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.1451374499167586e-06,
      "loss": 0.4882,
      "step": 149
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.122184358330857e-06,
      "loss": 0.5231,
      "step": 150
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.0991752188138157e-06,
      "loss": 0.4985,
      "step": 151
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.0761121040867666e-06,
      "loss": 0.4783,
      "step": 152
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.0529970917330655e-06,
      "loss": 0.4541,
      "step": 153
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.0298322640111333e-06,
      "loss": 0.5024,
      "step": 154
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.006619707666888e-06,
      "loss": 0.4872,
      "step": 155
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.983361513745759e-06,
      "loss": 0.47,
      "step": 156
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.9600597774043256e-06,
      "loss": 0.4915,
      "step": 157
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.936716597721576e-06,
      "loss": 0.4698,
      "step": 158
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.91333407750982e-06,
      "loss": 0.5199,
      "step": 159
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.889914323125261e-06,
      "loss": 0.4851,
      "step": 160
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.86645944427825e-06,
      "loss": 0.4561,
      "step": 161
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.842971553843238e-06,
      "loss": 0.5208,
      "step": 162
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.8194527676684413e-06,
      "loss": 0.4667,
      "step": 163
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.7959052043852463e-06,
      "loss": 0.4923,
      "step": 164
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.772330985217348e-06,
      "loss": 0.5243,
      "step": 165
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.7487322337896757e-06,
      "loss": 0.4648,
      "step": 166
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.725111075937083e-06,
      "loss": 0.4944,
      "step": 167
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.7014696395128526e-06,
      "loss": 0.5079,
      "step": 168
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.677810054197009e-06,
      "loss": 0.5054,
      "step": 169
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.6541344513044794e-06,
      "loss": 0.4352,
      "step": 170
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.6304449635930882e-06,
      "loss": 0.4617,
      "step": 171
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.606743725071446e-06,
      "loss": 0.4915,
      "step": 172
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.583032870806701e-06,
      "loss": 0.4621,
      "step": 173
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.559314536732212e-06,
      "loss": 0.4597,
      "step": 174
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.5355908594551416e-06,
      "loss": 0.4914,
      "step": 175
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.5118639760639775e-06,
      "loss": 0.4921,
      "step": 176
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.488136023936023e-06,
      "loss": 0.4818,
      "step": 177
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.4644091405448596e-06,
      "loss": 0.4938,
      "step": 178
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.4406854632677883e-06,
      "loss": 0.4512,
      "step": 179
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.4169671291933e-06,
      "loss": 0.4514,
      "step": 180
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.3932562749285544e-06,
      "loss": 0.4888,
      "step": 181
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.3695550364069113e-06,
      "loss": 0.457,
      "step": 182
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.3458655486955214e-06,
      "loss": 0.4688,
      "step": 183
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.3221899458029917e-06,
      "loss": 0.4735,
      "step": 184
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.298530360487149e-06,
      "loss": 0.4509,
      "step": 185
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.274888924062917e-06,
      "loss": 0.4718,
      "step": 186
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.2512677662103248e-06,
      "loss": 0.4964,
      "step": 187
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.227669014782653e-06,
      "loss": 0.4486,
      "step": 188
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.204094795614755e-06,
      "loss": 0.4668,
      "step": 189
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.180547232331559e-06,
      "loss": 0.4781,
      "step": 190
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.1570284461567633e-06,
      "loss": 0.467,
      "step": 191
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.1335405557217507e-06,
      "loss": 0.4766,
      "step": 192
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.1100856768747403e-06,
      "loss": 0.4822,
      "step": 193
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.086665922490181e-06,
      "loss": 0.4631,
      "step": 194
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.0632834022784247e-06,
      "loss": 0.463,
      "step": 195
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.0399402225956757e-06,
      "loss": 0.4595,
      "step": 196
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.016638486254242e-06,
      "loss": 0.4592,
      "step": 197
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.993380292333113e-06,
      "loss": 0.4354,
      "step": 198
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.9701677359888667e-06,
      "loss": 0.4716,
      "step": 199
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.947002908266935e-06,
      "loss": 0.461,
      "step": 200
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.923887895913234e-06,
      "loss": 0.4684,
      "step": 201
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.9008247811861858e-06,
      "loss": 0.4863,
      "step": 202
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8778156416691435e-06,
      "loss": 0.4768,
      "step": 203
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8548625500832412e-06,
      "loss": 0.4596,
      "step": 204
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8319675741006704e-06,
      "loss": 0.5025,
      "step": 205
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8091327761584247e-06,
      "loss": 0.4339,
      "step": 206
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.7863602132725082e-06,
      "loss": 0.4363,
      "step": 207
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.7636519368526338e-06,
      "loss": 0.4652,
      "step": 208
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.7410099925174306e-06,
      "loss": 0.4781,
      "step": 209
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.7184364199101649e-06,
      "loss": 0.4177,
      "step": 210
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.6959332525150046e-06,
      "loss": 0.4839,
      "step": 211
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.6735025174738407e-06,
      "loss": 0.4387,
      "step": 212
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.6511462354036765e-06,
      "loss": 0.4842,
      "step": 213
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.628866420214603e-06,
      "loss": 0.4816,
      "step": 214
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.6066650789283824e-06,
      "loss": 0.4427,
      "step": 215
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.584544211497652e-06,
      "loss": 0.4849,
      "step": 216
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.5625058106257578e-06,
      "loss": 0.4604,
      "step": 217
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.5405518615872578e-06,
      "loss": 0.4378,
      "step": 218
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.5186843420490698e-06,
      "loss": 0.4749,
      "step": 219
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.4969052218923285e-06,
      "loss": 0.4713,
      "step": 220
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.4752164630349287e-06,
      "loss": 0.4396,
      "step": 221
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.4536200192547933e-06,
      "loss": 0.462,
      "step": 222
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.4321178360138705e-06,
      "loss": 0.4682,
      "step": 223
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.4107118502828816e-06,
      "loss": 0.4543,
      "step": 224
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.3894039903668346e-06,
      "loss": 0.4499,
      "step": 225
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.3681961757313184e-06,
      "loss": 0.4721,
      "step": 226
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.3470903168295904e-06,
      "loss": 0.4341,
      "step": 227
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.3260883149304793e-06,
      "loss": 0.4412,
      "step": 228
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.3051920619471145e-06,
      "loss": 0.4669,
      "step": 229
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.2844034402664958e-06,
      "loss": 0.4285,
      "step": 230
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.2637243225799283e-06,
      "loss": 0.4959,
      "step": 231
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.243156571714317e-06,
      "loss": 0.5028,
      "step": 232
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.2227020404643671e-06,
      "loss": 0.4709,
      "step": 233
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.2023625714256743e-06,
      "loss": 0.4599,
      "step": 234
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.1821399968287437e-06,
      "loss": 0.4909,
      "step": 235
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.162036138373933e-06,
      "loss": 0.4473,
      "step": 236
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.142052807067352e-06,
      "loss": 0.4587,
      "step": 237
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.122191803057721e-06,
      "loss": 0.4851,
      "step": 238
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.1024549154742098e-06,
      "loss": 0.4672,
      "step": 239
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.0828439222652722e-06,
      "loss": 0.4369,
      "step": 240
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.0633605900384763e-06,
      "loss": 0.4724,
      "step": 241
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.044006673901369e-06,
      "loss": 0.4613,
      "step": 242
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.0247839173033707e-06,
      "loss": 0.443,
      "step": 243
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.0056940518787212e-06,
      "loss": 0.4355,
      "step": 244
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.867387972904885e-07,
      "loss": 0.4584,
      "step": 245
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.679198610756587e-07,
      "loss": 0.4517,
      "step": 246
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.492389384913159e-07,
      "loss": 0.4289,
      "step": 247
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.306977123619293e-07,
      "loss": 0.4602,
      "step": 248
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.122978529277601e-07,
      "loss": 0.4808,
      "step": 249
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.940410176944029e-07,
      "loss": 0.4159,
      "step": 250
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.759288512834721e-07,
      "loss": 0.4281,
      "step": 251
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.579629852844501e-07,
      "loss": 0.4232,
      "step": 252
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.401450381077134e-07,
      "loss": 0.473,
      "step": 253
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.224766148387369e-07,
      "loss": 0.433,
      "step": 254
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.049593070935044e-07,
      "loss": 0.4403,
      "step": 255
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.875946928751355e-07,
      "loss": 0.4502,
      "step": 256
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.70384336431734e-07,
      "loss": 0.4362,
      "step": 257
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.533297881154739e-07,
      "loss": 0.4599,
      "step": 258
    },
    {
      "epoch": 0.76,
      "learning_rate": 7.364325842429417e-07,
      "loss": 0.444,
      "step": 259
    },
    {
      "epoch": 0.76,
      "learning_rate": 7.196942469567417e-07,
      "loss": 0.4388,
      "step": 260
    },
    {
      "epoch": 0.76,
      "learning_rate": 7.031162840883735e-07,
      "loss": 0.4725,
      "step": 261
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.867001890224109e-07,
      "loss": 0.4848,
      "step": 262
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.70447440561966e-07,
      "loss": 0.4487,
      "step": 263
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.543595027954786e-07,
      "loss": 0.4603,
      "step": 264
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.38437824964827e-07,
      "loss": 0.459,
      "step": 265
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.226838413347766e-07,
      "loss": 0.4686,
      "step": 266
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.070989710637773e-07,
      "loss": 0.4551,
      "step": 267
    },
    {
      "epoch": 0.78,
      "learning_rate": 5.916846180761229e-07,
      "loss": 0.4531,
      "step": 268
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.764421709354812e-07,
      "loss": 0.4629,
      "step": 269
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.613730027198089e-07,
      "loss": 0.4439,
      "step": 270
    },
    {
      "epoch": 0.79,
      "learning_rate": 5.464784708976639e-07,
      "loss": 0.4753,
      "step": 271
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.317599172059149e-07,
      "loss": 0.4385,
      "step": 272
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.17218667528881e-07,
      "loss": 0.457,
      "step": 273
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.02856031778888e-07,
      "loss": 0.4289,
      "step": 274
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.886733037782728e-07,
      "loss": 0.4376,
      "step": 275
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.746717611428278e-07,
      "loss": 0.4672,
      "step": 276
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.6085266516671306e-07,
      "loss": 0.4424,
      "step": 277
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.4721726070883377e-07,
      "loss": 0.4512,
      "step": 278
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.337667760807018e-07,
      "loss": 0.4387,
      "step": 279
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.2050242293578817e-07,
      "loss": 0.4689,
      "step": 280
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.0742539616036887e-07,
      "loss": 0.4681,
      "step": 281
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.945368737658908e-07,
      "loss": 0.4694,
      "step": 282
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.818380167828528e-07,
      "loss": 0.4439,
      "step": 283
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6932996915621604e-07,
      "loss": 0.4998,
      "step": 284
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.570138576423551e-07,
      "loss": 0.4355,
      "step": 285
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.448907917075575e-07,
      "loss": 0.4304,
      "step": 286
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.3296186342807845e-07,
      "loss": 0.4262,
      "step": 287
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.2122814739176833e-07,
      "loss": 0.4455,
      "step": 288
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.096907006012639e-07,
      "loss": 0.3815,
      "step": 289
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.983505623787766e-07,
      "loss": 0.4566,
      "step": 290
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.872087542724669e-07,
      "loss": 0.485,
      "step": 291
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.762662799644178e-07,
      "loss": 0.4406,
      "step": 292
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.6552412518022483e-07,
      "loss": 0.4758,
      "step": 293
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.5498325760019646e-07,
      "loss": 0.476,
      "step": 294
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.4464462677218276e-07,
      "loss": 0.4912,
      "step": 295
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.3450916402603884e-07,
      "loss": 0.4512,
      "step": 296
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.245777823897305e-07,
      "loss": 0.4709,
      "step": 297
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.148513765070817e-07,
      "loss": 0.4696,
      "step": 298
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.0533082255718594e-07,
      "loss": 0.4632,
      "step": 299
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.960169781754781e-07,
      "loss": 0.4388,
      "step": 300
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.8691068237647296e-07,
      "loss": 0.4199,
      "step": 301
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.7801275547819047e-07,
      "loss": 0.4436,
      "step": 302
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6932399902825385e-07,
      "loss": 0.4694,
      "step": 303
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6084519573168728e-07,
      "loss": 0.4449,
      "step": 304
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.525771093804068e-07,
      "loss": 0.3888,
      "step": 305
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.44520484784417e-07,
      "loss": 0.4361,
      "step": 306
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.366760477047152e-07,
      "loss": 0.4424,
      "step": 307
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.290445047879138e-07,
      "loss": 0.4806,
      "step": 308
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.2162654350258403e-07,
      "loss": 0.4559,
      "step": 309
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.1442283207732746e-07,
      "loss": 0.3992,
      "step": 310
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.0743401944057902e-07,
      "loss": 0.437,
      "step": 311
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.006607351621497e-07,
      "loss": 0.471,
      "step": 312
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.410358939651587e-08,
      "loss": 0.4356,
      "step": 313
    },
    {
      "epoch": 0.92,
      "learning_rate": 8.776317282785163e-08,
      "loss": 0.4598,
      "step": 314
    },
    {
      "epoch": 0.92,
      "learning_rate": 8.164005661682155e-08,
      "loss": 0.4688,
      "step": 315
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.573479234912723e-08,
      "loss": 0.4137,
      "step": 316
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.004791198581939e-08,
      "loss": 0.4536,
      "step": 317
    },
    {
      "epoch": 0.93,
      "learning_rate": 6.457992781537692e-08,
      "loss": 0.4766,
      "step": 318
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.933133240756117e-08,
      "loss": 0.4133,
      "step": 319
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.430259856904041e-08,
      "loss": 0.4352,
      "step": 320
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.949417930080136e-08,
      "loss": 0.4586,
      "step": 321
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.490650775733962e-08,
      "loss": 0.4582,
      "step": 322
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.05399972076409e-08,
      "loss": 0.4722,
      "step": 323
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.6395040997952714e-08,
      "loss": 0.4568,
      "step": 324
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.2472012516350793e-08,
      "loss": 0.4583,
      "step": 325
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.8771265159102945e-08,
      "loss": 0.4655,
      "step": 326
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.529313229883451e-08,
      "loss": 0.4414,
      "step": 327
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.2037927254497936e-08,
      "loss": 0.4718,
      "step": 328
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.9005943263146197e-08,
      "loss": 0.4576,
      "step": 329
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.619745345352003e-08,
      "loss": 0.4149,
      "step": 330
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.3612710821441777e-08,
      "loss": 0.4145,
      "step": 331
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.1251948207025576e-08,
      "loss": 0.469,
      "step": 332
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.115378273702735e-09,
      "loss": 0.469,
      "step": 333
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.203193489064009e-09,
      "loss": 0.4461,
      "step": 334
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.515566107521797e-09,
      "loss": 0.4537,
      "step": 335
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.052648154793393e-09,
      "loss": 0.465,
      "step": 336
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.8145714142055467e-09,
      "loss": 0.4483,
      "step": 337
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.801447414823687e-09,
      "loss": 0.3912,
      "step": 338
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0133674214043987e-09,
      "loss": 0.4653,
      "step": 339
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.5040242617505614e-10,
      "loss": 0.4356,
      "step": 340
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.126031424372731e-10,
      "loss": 0.4387,
      "step": 341
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 0.4266,
      "step": 342
    },
    {
      "epoch": 1.0,
      "step": 342,
      "total_flos": 2267637667594240.0,
      "train_loss": 0.49030678340217526,
      "train_runtime": 5866.5854,
      "train_samples_per_second": 3.729,
      "train_steps_per_second": 0.058
    }
  ],
  "max_steps": 342,
  "num_train_epochs": 1,
  "total_flos": 2267637667594240.0,
  "trial_name": null,
  "trial_params": null
}