{ "best_metric": null, "best_model_checkpoint": null, "epoch": 50.0, "eval_steps": 500, "global_step": 450, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.11, "learning_rate": 0.00022222222222222223, "loss": 1.7682, "step": 1 }, { "epoch": 0.22, "learning_rate": 0.00044444444444444447, "loss": 1.9883, "step": 2 }, { "epoch": 0.33, "learning_rate": 0.0006666666666666666, "loss": 1.7947, "step": 3 }, { "epoch": 0.44, "learning_rate": 0.0008888888888888889, "loss": 1.9388, "step": 4 }, { "epoch": 0.56, "learning_rate": 0.0011111111111111111, "loss": 1.8301, "step": 5 }, { "epoch": 0.67, "learning_rate": 0.0011111111111111111, "loss": 1.7121, "step": 6 }, { "epoch": 0.78, "learning_rate": 0.0011111111111111111, "loss": 1.7248, "step": 7 }, { "epoch": 0.89, "learning_rate": 0.0013333333333333333, "loss": 1.6582, "step": 8 }, { "epoch": 1.0, "learning_rate": 0.0013333333333333333, "loss": 2.0618, "step": 9 }, { "epoch": 1.11, "learning_rate": 0.0013333333333333333, "loss": 2.0288, "step": 10 }, { "epoch": 1.22, "learning_rate": 0.0013333333333333333, "loss": 6.2236, "step": 11 }, { "epoch": 1.33, "learning_rate": 0.0013333333333333333, "loss": 4.7077, "step": 12 }, { "epoch": 1.44, "learning_rate": 0.0013333333333333333, "loss": 2.8244, "step": 13 }, { "epoch": 1.56, "learning_rate": 0.0013333333333333333, "loss": 3.6065, "step": 14 }, { "epoch": 1.67, "learning_rate": 0.0015555555555555557, "loss": 2.9129, "step": 15 }, { "epoch": 1.78, "learning_rate": 0.0017777777777777779, "loss": 1.5032, "step": 16 }, { "epoch": 1.89, "learning_rate": 0.002, "loss": 1.2412, "step": 17 }, { "epoch": 2.0, "learning_rate": 0.0022222222222222222, "loss": 1.7111, "step": 18 }, { "epoch": 2.11, "learning_rate": 0.0024444444444444444, "loss": 1.6141, "step": 19 }, { "epoch": 2.22, "learning_rate": 0.0026666666666666666, "loss": 1.6745, "step": 20 }, { "epoch": 2.33, "learning_rate": 0.0028888888888888888, "loss": 2.4298, "step": 21 }, { "epoch": 2.44, "learning_rate": 0.0031111111111111114, "loss": 1.6993, "step": 22 }, { "epoch": 2.56, "learning_rate": 0.003333333333333333, "loss": 4.6999, "step": 23 }, { "epoch": 2.67, "learning_rate": 0.0035555555555555557, "loss": 3.5699, "step": 24 }, { "epoch": 2.78, "learning_rate": 0.003777777777777778, "loss": 8.7543, "step": 25 }, { "epoch": 2.89, "learning_rate": 0.004, "loss": 9.8707, "step": 26 }, { "epoch": 3.0, "learning_rate": 0.004222222222222223, "loss": 10.1086, "step": 27 }, { "epoch": 3.11, "learning_rate": 0.0044444444444444444, "loss": 9.2775, "step": 28 }, { "epoch": 3.22, "learning_rate": 0.004666666666666667, "loss": 7.8785, "step": 29 }, { "epoch": 3.33, "learning_rate": 0.004888888888888889, "loss": 7.472, "step": 30 }, { "epoch": 3.44, "learning_rate": 0.0051111111111111105, "loss": 7.2495, "step": 31 }, { "epoch": 3.56, "learning_rate": 0.005333333333333333, "loss": 13.0327, "step": 32 }, { "epoch": 3.67, "learning_rate": 0.005555555555555556, "loss": 15.2515, "step": 33 }, { "epoch": 3.78, "learning_rate": 0.0057777777777777775, "loss": 12.4308, "step": 34 }, { "epoch": 3.89, "learning_rate": 0.006, "loss": 11.2797, "step": 35 }, { "epoch": 4.0, "learning_rate": 0.006222222222222223, "loss": 9.5219, "step": 36 }, { "epoch": 4.11, "learning_rate": 0.006444444444444445, "loss": 7.6947, "step": 37 }, { "epoch": 4.22, "learning_rate": 0.006444444444444445, "loss": 11.9818, "step": 38 }, { "epoch": 4.33, "learning_rate": 0.006444444444444445, "loss": 14.0646, "step": 39 }, { "epoch": 4.44, "learning_rate": 0.006666666666666666, "loss": 12.282, "step": 40 }, { "epoch": 4.56, "learning_rate": 0.006888888888888889, "loss": 10.8636, "step": 41 }, { "epoch": 4.67, "learning_rate": 0.0071111111111111115, "loss": 8.8291, "step": 42 }, { "epoch": 4.78, "learning_rate": 0.007333333333333333, "loss": 8.4366, "step": 43 }, { "epoch": 4.89, "learning_rate": 0.007555555555555556, "loss": 8.9278, "step": 44 }, { "epoch": 5.0, "learning_rate": 0.007777777777777778, "loss": 11.5827, "step": 45 }, { "epoch": 5.11, "learning_rate": 0.008, "loss": 8.8135, "step": 46 }, { "epoch": 5.22, "learning_rate": 0.008222222222222223, "loss": 7.5243, "step": 47 }, { "epoch": 5.33, "learning_rate": 0.008444444444444445, "loss": 8.1523, "step": 48 }, { "epoch": 5.44, "learning_rate": 0.008666666666666668, "loss": 8.2508, "step": 49 }, { "epoch": 5.56, "learning_rate": 0.008888888888888889, "loss": 7.961, "step": 50 }, { "epoch": 5.67, "learning_rate": 0.009111111111111111, "loss": 7.3086, "step": 51 }, { "epoch": 5.78, "learning_rate": 0.009333333333333334, "loss": 7.3684, "step": 52 }, { "epoch": 5.89, "learning_rate": 0.009555555555555557, "loss": 7.3783, "step": 53 }, { "epoch": 6.0, "learning_rate": 0.009777777777777778, "loss": 9.3611, "step": 54 }, { "epoch": 6.11, "learning_rate": 0.01, "loss": 12.1977, "step": 55 }, { "epoch": 6.22, "learning_rate": 0.009975308641975308, "loss": 7.2867, "step": 56 }, { "epoch": 6.33, "learning_rate": 0.009950617283950617, "loss": 8.0552, "step": 57 }, { "epoch": 6.44, "learning_rate": 0.009925925925925927, "loss": 7.7156, "step": 58 }, { "epoch": 6.56, "learning_rate": 0.009901234567901235, "loss": 7.3561, "step": 59 }, { "epoch": 6.67, "learning_rate": 0.009876543209876543, "loss": 7.4133, "step": 60 }, { "epoch": 6.78, "learning_rate": 0.009851851851851851, "loss": 16.2833, "step": 61 }, { "epoch": 6.89, "learning_rate": 0.009827160493827161, "loss": 8.633, "step": 62 }, { "epoch": 7.0, "learning_rate": 0.00980246913580247, "loss": 6.6628, "step": 63 }, { "epoch": 7.11, "learning_rate": 0.009777777777777778, "loss": 6.9888, "step": 64 }, { "epoch": 7.22, "learning_rate": 0.009753086419753086, "loss": 7.3082, "step": 65 }, { "epoch": 7.33, "learning_rate": 0.009728395061728396, "loss": 7.2472, "step": 66 }, { "epoch": 7.44, "learning_rate": 0.009703703703703704, "loss": 7.0683, "step": 67 }, { "epoch": 7.56, "learning_rate": 0.009679012345679012, "loss": 7.1155, "step": 68 }, { "epoch": 7.67, "learning_rate": 0.009654320987654322, "loss": 6.8406, "step": 69 }, { "epoch": 7.78, "learning_rate": 0.009629629629629629, "loss": 7.1108, "step": 70 }, { "epoch": 7.89, "learning_rate": 0.009604938271604939, "loss": 6.7208, "step": 71 }, { "epoch": 8.0, "learning_rate": 0.009580246913580247, "loss": 6.7704, "step": 72 }, { "epoch": 8.11, "learning_rate": 0.009555555555555557, "loss": 6.8468, "step": 73 }, { "epoch": 8.22, "learning_rate": 0.009530864197530865, "loss": 6.9493, "step": 74 }, { "epoch": 8.33, "learning_rate": 0.009506172839506173, "loss": 7.0246, "step": 75 }, { "epoch": 8.44, "learning_rate": 0.009481481481481481, "loss": 6.7802, "step": 76 }, { "epoch": 8.56, "learning_rate": 0.00945679012345679, "loss": 6.8637, "step": 77 }, { "epoch": 8.67, "learning_rate": 0.0094320987654321, "loss": 6.8141, "step": 78 }, { "epoch": 8.78, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 79 }, { "epoch": 8.89, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 80 }, { "epoch": 9.0, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 81 }, { "epoch": 9.11, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 82 }, { "epoch": 9.22, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 83 }, { "epoch": 9.33, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 84 }, { "epoch": 9.44, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 85 }, { "epoch": 9.56, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 86 }, { "epoch": 9.67, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 87 }, { "epoch": 9.78, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 88 }, { "epoch": 9.89, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 89 }, { "epoch": 10.0, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 90 }, { "epoch": 10.11, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 91 }, { "epoch": 10.22, "learning_rate": 0.0094320987654321, "loss": 0.0, "step": 92 }, { "epoch": 10.33, "learning_rate": 0.009407407407407408, "loss": 7.0349, "step": 93 }, { "epoch": 10.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 94 }, { "epoch": 10.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 95 }, { "epoch": 10.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 96 }, { "epoch": 10.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 97 }, { "epoch": 10.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 98 }, { "epoch": 11.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 99 }, { "epoch": 11.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 100 }, { "epoch": 11.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 101 }, { "epoch": 11.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 102 }, { "epoch": 11.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 103 }, { "epoch": 11.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 104 }, { "epoch": 11.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 105 }, { "epoch": 11.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 106 }, { "epoch": 11.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 107 }, { "epoch": 12.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 108 }, { "epoch": 12.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 109 }, { "epoch": 12.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 110 }, { "epoch": 12.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 111 }, { "epoch": 12.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 112 }, { "epoch": 12.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 113 }, { "epoch": 12.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 114 }, { "epoch": 12.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 115 }, { "epoch": 12.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 116 }, { "epoch": 13.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 117 }, { "epoch": 13.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 118 }, { "epoch": 13.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 119 }, { "epoch": 13.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 120 }, { "epoch": 13.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 121 }, { "epoch": 13.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 122 }, { "epoch": 13.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 123 }, { "epoch": 13.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 124 }, { "epoch": 13.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 125 }, { "epoch": 14.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 126 }, { "epoch": 14.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 127 }, { "epoch": 14.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 128 }, { "epoch": 14.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 129 }, { "epoch": 14.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 130 }, { "epoch": 14.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 131 }, { "epoch": 14.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 132 }, { "epoch": 14.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 133 }, { "epoch": 14.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 134 }, { "epoch": 15.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 135 }, { "epoch": 15.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 136 }, { "epoch": 15.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 137 }, { "epoch": 15.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 138 }, { "epoch": 15.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 139 }, { "epoch": 15.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 140 }, { "epoch": 15.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 141 }, { "epoch": 15.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 142 }, { "epoch": 15.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 143 }, { "epoch": 16.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 144 }, { "epoch": 16.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 145 }, { "epoch": 16.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 146 }, { "epoch": 16.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 147 }, { "epoch": 16.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 148 }, { "epoch": 16.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 149 }, { "epoch": 16.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 150 }, { "epoch": 16.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 151 }, { "epoch": 16.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 152 }, { "epoch": 17.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 153 }, { "epoch": 17.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 154 }, { "epoch": 17.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 155 }, { "epoch": 17.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 156 }, { "epoch": 17.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 157 }, { "epoch": 17.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 158 }, { "epoch": 17.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 159 }, { "epoch": 17.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 160 }, { "epoch": 17.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 161 }, { "epoch": 18.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 162 }, { "epoch": 18.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 163 }, { "epoch": 18.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 164 }, { "epoch": 18.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 165 }, { "epoch": 18.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 166 }, { "epoch": 18.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 167 }, { "epoch": 18.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 168 }, { "epoch": 18.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 169 }, { "epoch": 18.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 170 }, { "epoch": 19.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 171 }, { "epoch": 19.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 172 }, { "epoch": 19.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 173 }, { "epoch": 19.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 174 }, { "epoch": 19.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 175 }, { "epoch": 19.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 176 }, { "epoch": 19.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 177 }, { "epoch": 19.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 178 }, { "epoch": 19.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 179 }, { "epoch": 20.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 180 }, { "epoch": 20.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 181 }, { "epoch": 20.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 182 }, { "epoch": 20.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 183 }, { "epoch": 20.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 184 }, { "epoch": 20.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 185 }, { "epoch": 20.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 186 }, { "epoch": 20.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 187 }, { "epoch": 20.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 188 }, { "epoch": 21.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 189 }, { "epoch": 21.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 190 }, { "epoch": 21.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 191 }, { "epoch": 21.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 192 }, { "epoch": 21.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 193 }, { "epoch": 21.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 194 }, { "epoch": 21.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 195 }, { "epoch": 21.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 196 }, { "epoch": 21.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 197 }, { "epoch": 22.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 198 }, { "epoch": 22.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 199 }, { "epoch": 22.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 200 }, { "epoch": 22.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 201 }, { "epoch": 22.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 202 }, { "epoch": 22.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 203 }, { "epoch": 22.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 204 }, { "epoch": 22.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 205 }, { "epoch": 22.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 206 }, { "epoch": 23.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 207 }, { "epoch": 23.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 208 }, { "epoch": 23.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 209 }, { "epoch": 23.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 210 }, { "epoch": 23.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 211 }, { "epoch": 23.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 212 }, { "epoch": 23.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 213 }, { "epoch": 23.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 214 }, { "epoch": 23.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 215 }, { "epoch": 24.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 216 }, { "epoch": 24.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 217 }, { "epoch": 24.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 218 }, { "epoch": 24.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 219 }, { "epoch": 24.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 220 }, { "epoch": 24.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 221 }, { "epoch": 24.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 222 }, { "epoch": 24.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 223 }, { "epoch": 24.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 224 }, { "epoch": 25.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 225 }, { "epoch": 25.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 226 }, { "epoch": 25.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 227 }, { "epoch": 25.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 228 }, { "epoch": 25.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 229 }, { "epoch": 25.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 230 }, { "epoch": 25.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 231 }, { "epoch": 25.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 232 }, { "epoch": 25.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 233 }, { "epoch": 26.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 234 }, { "epoch": 26.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 235 }, { "epoch": 26.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 236 }, { "epoch": 26.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 237 }, { "epoch": 26.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 238 }, { "epoch": 26.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 239 }, { "epoch": 26.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 240 }, { "epoch": 26.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 241 }, { "epoch": 26.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 242 }, { "epoch": 27.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 243 }, { "epoch": 27.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 244 }, { "epoch": 27.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 245 }, { "epoch": 27.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 246 }, { "epoch": 27.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 247 }, { "epoch": 27.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 248 }, { "epoch": 27.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 249 }, { "epoch": 27.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 250 }, { "epoch": 27.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 251 }, { "epoch": 28.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 252 }, { "epoch": 28.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 253 }, { "epoch": 28.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 254 }, { "epoch": 28.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 255 }, { "epoch": 28.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 256 }, { "epoch": 28.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 257 }, { "epoch": 28.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 258 }, { "epoch": 28.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 259 }, { "epoch": 28.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 260 }, { "epoch": 29.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 261 }, { "epoch": 29.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 262 }, { "epoch": 29.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 263 }, { "epoch": 29.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 264 }, { "epoch": 29.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 265 }, { "epoch": 29.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 266 }, { "epoch": 29.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 267 }, { "epoch": 29.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 268 }, { "epoch": 29.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 269 }, { "epoch": 30.0, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 270 }, { "epoch": 30.11, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 271 }, { "epoch": 30.22, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 272 }, { "epoch": 30.33, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 273 }, { "epoch": 30.44, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 274 }, { "epoch": 30.56, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 275 }, { "epoch": 30.67, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 276 }, { "epoch": 30.78, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 277 }, { "epoch": 30.89, "learning_rate": 0.009407407407407408, "loss": 0.0, "step": 278 }, { "epoch": 31.0, "learning_rate": 0.009382716049382716, "loss": 6.9362, "step": 279 }, { "epoch": 31.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 280 }, { "epoch": 31.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 281 }, { "epoch": 31.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 282 }, { "epoch": 31.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 283 }, { "epoch": 31.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 284 }, { "epoch": 31.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 285 }, { "epoch": 31.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 286 }, { "epoch": 31.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 287 }, { "epoch": 32.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 288 }, { "epoch": 32.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 289 }, { "epoch": 32.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 290 }, { "epoch": 32.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 291 }, { "epoch": 32.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 292 }, { "epoch": 32.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 293 }, { "epoch": 32.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 294 }, { "epoch": 32.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 295 }, { "epoch": 32.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 296 }, { "epoch": 33.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 297 }, { "epoch": 33.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 298 }, { "epoch": 33.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 299 }, { "epoch": 33.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 300 }, { "epoch": 33.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 301 }, { "epoch": 33.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 302 }, { "epoch": 33.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 303 }, { "epoch": 33.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 304 }, { "epoch": 33.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 305 }, { "epoch": 34.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 306 }, { "epoch": 34.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 307 }, { "epoch": 34.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 308 }, { "epoch": 34.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 309 }, { "epoch": 34.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 310 }, { "epoch": 34.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 311 }, { "epoch": 34.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 312 }, { "epoch": 34.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 313 }, { "epoch": 34.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 314 }, { "epoch": 35.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 315 }, { "epoch": 35.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 316 }, { "epoch": 35.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 317 }, { "epoch": 35.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 318 }, { "epoch": 35.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 319 }, { "epoch": 35.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 320 }, { "epoch": 35.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 321 }, { "epoch": 35.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 322 }, { "epoch": 35.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 323 }, { "epoch": 36.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 324 }, { "epoch": 36.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 325 }, { "epoch": 36.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 326 }, { "epoch": 36.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 327 }, { "epoch": 36.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 328 }, { "epoch": 36.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 329 }, { "epoch": 36.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 330 }, { "epoch": 36.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 331 }, { "epoch": 36.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 332 }, { "epoch": 37.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 333 }, { "epoch": 37.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 334 }, { "epoch": 37.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 335 }, { "epoch": 37.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 336 }, { "epoch": 37.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 337 }, { "epoch": 37.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 338 }, { "epoch": 37.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 339 }, { "epoch": 37.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 340 }, { "epoch": 37.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 341 }, { "epoch": 38.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 342 }, { "epoch": 38.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 343 }, { "epoch": 38.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 344 }, { "epoch": 38.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 345 }, { "epoch": 38.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 346 }, { "epoch": 38.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 347 }, { "epoch": 38.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 348 }, { "epoch": 38.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 349 }, { "epoch": 38.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 350 }, { "epoch": 39.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 351 }, { "epoch": 39.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 352 }, { "epoch": 39.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 353 }, { "epoch": 39.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 354 }, { "epoch": 39.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 355 }, { "epoch": 39.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 356 }, { "epoch": 39.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 357 }, { "epoch": 39.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 358 }, { "epoch": 39.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 359 }, { "epoch": 40.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 360 }, { "epoch": 40.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 361 }, { "epoch": 40.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 362 }, { "epoch": 40.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 363 }, { "epoch": 40.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 364 }, { "epoch": 40.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 365 }, { "epoch": 40.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 366 }, { "epoch": 40.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 367 }, { "epoch": 40.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 368 }, { "epoch": 41.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 369 }, { "epoch": 41.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 370 }, { "epoch": 41.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 371 }, { "epoch": 41.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 372 }, { "epoch": 41.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 373 }, { "epoch": 41.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 374 }, { "epoch": 41.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 375 }, { "epoch": 41.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 376 }, { "epoch": 41.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 377 }, { "epoch": 42.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 378 }, { "epoch": 42.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 379 }, { "epoch": 42.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 380 }, { "epoch": 42.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 381 }, { "epoch": 42.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 382 }, { "epoch": 42.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 383 }, { "epoch": 42.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 384 }, { "epoch": 42.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 385 }, { "epoch": 42.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 386 }, { "epoch": 43.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 387 }, { "epoch": 43.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 388 }, { "epoch": 43.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 389 }, { "epoch": 43.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 390 }, { "epoch": 43.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 391 }, { "epoch": 43.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 392 }, { "epoch": 43.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 393 }, { "epoch": 43.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 394 }, { "epoch": 43.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 395 }, { "epoch": 44.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 396 }, { "epoch": 44.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 397 }, { "epoch": 44.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 398 }, { "epoch": 44.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 399 }, { "epoch": 44.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 400 }, { "epoch": 44.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 401 }, { "epoch": 44.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 402 }, { "epoch": 44.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 403 }, { "epoch": 44.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 404 }, { "epoch": 45.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 405 }, { "epoch": 45.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 406 }, { "epoch": 45.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 407 }, { "epoch": 45.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 408 }, { "epoch": 45.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 409 }, { "epoch": 45.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 410 }, { "epoch": 45.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 411 }, { "epoch": 45.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 412 }, { "epoch": 45.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 413 }, { "epoch": 46.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 414 }, { "epoch": 46.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 415 }, { "epoch": 46.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 416 }, { "epoch": 46.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 417 }, { "epoch": 46.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 418 }, { "epoch": 46.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 419 }, { "epoch": 46.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 420 }, { "epoch": 46.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 421 }, { "epoch": 46.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 422 }, { "epoch": 47.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 423 }, { "epoch": 47.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 424 }, { "epoch": 47.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 425 }, { "epoch": 47.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 426 }, { "epoch": 47.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 427 }, { "epoch": 47.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 428 }, { "epoch": 47.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 429 }, { "epoch": 47.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 430 }, { "epoch": 47.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 431 }, { "epoch": 48.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 432 }, { "epoch": 48.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 433 }, { "epoch": 48.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 434 }, { "epoch": 48.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 435 }, { "epoch": 48.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 436 }, { "epoch": 48.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 437 }, { "epoch": 48.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 438 }, { "epoch": 48.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 439 }, { "epoch": 48.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 440 }, { "epoch": 49.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 441 }, { "epoch": 49.11, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 442 }, { "epoch": 49.22, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 443 }, { "epoch": 49.33, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 444 }, { "epoch": 49.44, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 445 }, { "epoch": 49.56, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 446 }, { "epoch": 49.67, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 447 }, { "epoch": 49.78, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 448 }, { "epoch": 49.89, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 449 }, { "epoch": 50.0, "learning_rate": 0.009382716049382716, "loss": 0.0, "step": 450 } ], "logging_steps": 1, "max_steps": 450, "num_input_tokens_seen": 0, "num_train_epochs": 50, "save_steps": 500, "total_flos": 3.7170305826816e+16, "train_batch_size": 2, "trial_name": null, "trial_params": null }